U.S. patent application number 12/736430 was filed with the patent office on 2011-02-10 for location of wireless mobile terminals.
This patent application is currently assigned to Seeker Wireless Pty. Limited. Invention is credited to Stephen Frank Brown, Malcolm David, Christophe Ridgway Drane, Craig Andrew Scott.
Application Number | 20110034179 12/736430 |
Document ID | / |
Family ID | 41161465 |
Filed Date | 2011-02-10 |
United States Patent
Application |
20110034179 |
Kind Code |
A1 |
David; Malcolm ; et
al. |
February 10, 2011 |
LOCATION OF WIRELESS MOBILE TERMINALS
Abstract
Methods, systems, processor-readable media, and devices for
collecting information pertaining to the configuration of one or
more wireless networks and using this information in turn to
estimate the location of mobile wireless devices associated with
those networks are disclosed. Also disclosed are methods, systems,
processor-readable media, and devices for discovering and/or
maintaining wireless transmitter characteristics corresponding to
one or more wireless networks. Also disclosed are methods, systems,
processor-readable media, and devices for aggregating wireless
network transmitter characteristics from a plurality of wireless
network transmitters. The present disclosure can be used across
multiple networks, without requiring information from the operators
of those networks, in which the information concerning those
networks is collected and maintained in a current state efficiently
and with rapid response to changes in the network
configuration.
Inventors: |
David; Malcolm; (New South
Wales, AU) ; Drane; Christophe Ridgway; (New South
Wales, AU) ; Brown; Stephen Frank; (New South Wales,
AU) ; Scott; Craig Andrew; (New South Wales,
AU) |
Correspondence
Address: |
JONES DAY
222 EAST 41ST ST
NEW YORK
NY
10017
US
|
Assignee: |
Seeker Wireless Pty.
Limited
Gordon, New South Wales
AU
|
Family ID: |
41161465 |
Appl. No.: |
12/736430 |
Filed: |
April 7, 2009 |
PCT Filed: |
April 7, 2009 |
PCT NO: |
PCT/AU2009/000438 |
371 Date: |
October 7, 2010 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
61064977 |
Apr 7, 2008 |
|
|
|
Current U.S.
Class: |
455/456.1 |
Current CPC
Class: |
G01S 5/0236 20130101;
H04W 64/00 20130101; H04W 64/003 20130101; H04W 16/18 20130101 |
Class at
Publication: |
455/456.1 |
International
Class: |
H04W 64/00 20090101
H04W064/00 |
Claims
1. A computer-implemented method of aggregating wireless network
transmitter characteristics corresponding to a plurality of
wireless network transmitters comprising the steps of: receiving a
report from a mobile radio terminal, wherein the report includes at
least one measurement of at least one radio network parameter from
a first wireless network transmitter and GPS location information;
and estimating or updating a set of characteristics for the first
wireless network transmitter based at least in part on the reported
measurements and the GPS location information.
2. The method of claim 1 wherein the set of characteristics include
one or more location dependent parameters describing the coverage
footprint of the first wireless network transmitter.
3. The method of claim 1 wherein the method includes the step of
storing the set of characteristics for the first wireless network
transmitter.
4. The method of claim 2 wherein the one or more location dependent
characteristics comprise a mean and a covariance of a 2D Gaussian
probability density function approximating the coverage footprint
of the first wireless network transmitter.
5. A computer-implemented method of updating a database having
characteristics of a plurality of wireless network transmitters
comprising the steps of: receiving a report from a mobile radio
terminal, wherein the report includes one or more measurements of
at least one radio parameter corresponding to a first wireless
network transmitter; and one or more measurements of at least one
radio parameter corresponding to a second wireless network
transmitter; estimating a location for the radio terminal based at
least in part on the report and characteristics corresponding to
the first and second wireless transmitters; and estimating or
updating a set of characteristics for the first wireless network
transmitter based at least in part on the reported measurements and
the estimated location.
6. The method of claim 1 wherein the first wireless network
transmitter is selected from a group consisting of a GSM cell, a
Wi-Fi Access Point, a CDMA cell, a UMTS cell, an LTE cell or a
WiMax Base Station.
7. The method of claim 5 comprising the further steps of:
estimating a set of characteristics for the second wireless network
transmitter based at least in part on the estimated location of the
radio terminal and the at least one measurement of at least one
radio network parameter of the second wireless network transmitter;
and storing the set of characteristics for the second wireless
network transmitter.
8. The method of claim 5 wherein the first and the second wireless
network transmitters are selected from a group consisting of GSM,
UMTS, LTE, or Mobile WiMAX transmitters.
9. The method of claim 5 wherein the first wireless network
transmitter is a cellular base station or a Wi-Fi Access Point and
the second wireless network transmitter is a cellular base station
or a Wi-Fi Access Point.
10. The method of claim 1 further comprising the step of
authenticating a client on the mobile radio terminal.
11. The method of claim 5 wherein estimating the location of the
mobile radio terminal comprises: applying a cost function to the at
least one measurement of the at least one radio network parameter
of the first wireless network transmitter and one or more
characteristics associated with the at least one wireless network
transmitter; and minimizing a sum of costs generated by the cost
function to determine the estimated location of the mobile radio
terminal.
12. The method of claim 5 wherein estimating the location of the
mobile radio terminal comprises: applying a cost function to the at
least one measurement of the at least one radio network parameter
of the first wireless network transmitter and to the at least one
measurement of the at least one radio network parameter of the
second wireless network transmitter and one or more characteristics
associated with the first wireless network transmitter and/or one
or more characteristics associated with the second wireless network
transmitter; and minimizing a sum of costs generated by the cost
function to determine the estimated location of the mobile radio
terminal.
13. The method of claim 1 wherein the at least one measurement of
at least one radio network parameter of the first wireless network
transmitter includes an identifier.
14. The method of claim 1 wherein the report is received via a
bearer selected from the group consisting of SMS, GPRS, EDGE, HSPA,
and EV-DO.
15. A processing system that is capable of determining a location
of a radio terminal comprising: at least one wireless network
comprising a plurality of wireless network transmitters; at least
one mobile radio terminal in communication with said at least one
wireless network; at least one database containing characteristics
corresponding to transmitters of said at least one wireless
network; wherein the processing system is configured to execute
steps comprising: measuring in a mobile radio terminal one or more
radio network parameter corresponding to at least one wireless
network transmitter; receiving at a server the radio network
parameter measurements; applying a cost function to the one or more
measurements and one or more characteristics corresponding to the
at least one wireless network transmitter; and minimizing a sum of
costs generated by the cost function to calculate the location of
the mobile radio terminal.
16. The processing system of claim 15 wherein the one or more
characteristics are parameters describing the reception footprint
of the at least one wireless transmitter.
17. The processing system of claim 16 wherein the one or more
characteristics are a mean and a covariance of a 2D Gaussian
probability density function approximating the coverage footprint
of the at least one wireless transmitter.
18. The processing system of claim 16 wherein calculating the
location of the radio terminal comprises the steps of: applying a
cost function to the at least one measurement of the at least one
radio network parameter of the first wireless network transmitter;
and minimizing a sum of costs generated by the cost function to
determine the location of the mobile radio terminal.
19. A computer-implemented method of discovering and/or maintaining
wireless transmitter characteristics for wireless transmitters in a
wireless network comprising the steps of: receiving from a mobile
radio terminal, at least one measurement of at least one radio
parameter corresponding to a first wireless transmitter and at
least one measurement of at least one radio parameter corresponding
to at least one other wireless transmitter; retrieving a set of
characteristics for the at least one other transmitter from a
database; and estimating or updating in a processing system a set
of characteristics for the first wireless transmitter based at
least in part on the characteristics of the at least one other
transmitter, and the collected measurements corresponding to the
first wireless transmitter and the at least one other wireless
transmitter.
20. The computer-implemented method of claim 19 wherein the set of
characteristics for the first wireless transmitter include location
dependent characteristics describing the coverage footprint of the
first wireless transmitter.
21. The computer-implemented method of claim 20 wherein, the set of
characteristics for the first wireless transmitter include a mean
and a covariance of a 2D Gaussian probability density function
approximating the coverage footprint of the first wireless
transmitter.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] The present application is related to the following
co-pending patent applications: PCT/AU2006/000479 entitled Mobile
Location; PCT/AU2006/000348 entitled Enhanced Mobile Location;
PCT/AU2006/000347 entitled Enhanced Mobile Location Method and
System; PCT/AU2006/000478 entitled Enhanced Terrestrial Mobile
Location; PCT/AU2008/000344 entitled Enhanced Zone Determination,
filed Mar. 13, 2008 claiming priority from U.S. Provisional
Application Ser. No. 60/906,526; PCT/AU2006/001577 entitled
Detection in Mobile Service Maintenance; PCT/AU2006/001576 entitled
Mobile Service Maintenance Management; U.S. Provisional Application
Ser. No. 61/064,977 entitled Location of wireless mobile terminals,
filed Apr. 7, 2008 from which the present PCT application claims
priority; and PCT/AU2009/______ entitled Efficient collection of
wireless transmitter characteristics filed concurrent with the
present PCT. The entire content of each of these applications is
hereby incorporated by reference. Furthermore, the entire contents
of the following references are hereby incorporated by reference.
W. C. Y. Lee, Mobile Communications Engineering, McGraw-Hill, 1982,
and P. L. H. A. S. Fischer, "Evaluation of positioning measurement
systems", TlPl.5/97-110, December 1997, and IEEE VTS committee,
"Coverage prediction for mobile radio systems operating in the
800/900 MHz frequency range", IEEE Transactions on VTC, Vol. 37,
No. 1, February 1998, 3GPP TS 05.08, and C. R. Drane, Positioning
Systems, a Unified Approach, Springer Verlag, 1992, S. R. Saunders
& A. Aragon-Zavala, Antennas And Propagation For Wireless
Communication Systems: 2.sup.nd Ed, Wiley, 2007.
FIELD
[0002] Certain embodiments of the present disclosure are directed
to providing systems, methods and devices for collecting
information pertaining to the configuration of one or more wireless
networks and using this information in turn to estimate the
location of mobile wireless devices associated with those
networks.
BACKGROUND
[0003] The use of location information has been considered in
mobile services for many years. After a period of hype, so-called
location based services (LBS) suffered a downturn for a combination
of reasons. One was that the some of the technologies developed to
achieve high accuracy turned out to be too expensive to deploy or
too much of an operational burden to maintain. Another problem was
that the new technologies required new handsets which created a
problem launching the services. Another problem was that the value
chain was too complex.
[0004] Location information for a wireless device can be obtained
from at least two sources. One source is via a global positioning
system (GPS) receiver integrated into the device (or external but
coupled to the device via Bluetooth). However, GPS has a number of
limiting factors. For example, GPS is only available in a limited
number of cellular handsets, thereby limiting its availability.
Further, GPS typically performs poorly indoors due to the reliance
on line of sight for satellite fixes. Additionally, GPS has a high
latency which can frustrate some users who don't want to wait for a
GPS lock. Another source is by calculating the position relative to
the fixed transmitters of a wireless network. In the case of
cellular devices, mobile network operators (MNOs) control the base
stations and therefore have the information needed to perform the
location calculations. In the past some attempts have been made to
collect this information in order to make cellular based location
publically accessible however none of these achieved sufficient
coverage or quality to support high quality services.
[0005] As a result, cellular based location services tend to be
very heavily dependent on the particular strategies of the
individual carriers. Few MNOs make high capacity, high performance
location information for their subscribers externally accessible to
application developers and service providers. Instead, in those
cases where MNOs do actually implement some location technology,
they tend to instead use it with their own vertically integrated
suite of applications.
[0006] In the emerging Mobile Web 2.0 services ecosystem this
represents a significant impediment because fewer and fewer
applications can be viable if operated only with a single MNO or on
a small number of devices. Additionally the emerging mobile
Internet ecosystem enabled by recent advances in device UI
capabilities, wireless connections speeds and availability of
content for mobile consumption is being held back because the
majority of the independent software vendors and content providers
cannot afford to negotiate individual deals with all the carriers
to make location information available.
[0007] Application developers seeking to use location information
across multiple networks face another challenge even where
operators do make location information available. The differences
between the location technologies used by the different operators
make it difficult to abstract across the different networks to
obtain a consistent level of functionality and performance. In some
cases, third parties have attempted to implement broker type
systems which perform abstraction but in the end the need to fall
back to the lowest common denominator means that both the
functionality & performance are reduced to a very basic set of
capabilities.
[0008] Thus a need exists to make device location information
easily accessible across all wireless networks. This type of
technology naturally has to be coupled to appropriate mechanisms to
protect the information. Another challenge for the platform is the
continual evolution of the wireless networks as operators optimize
the networks and add capacity. In some markets many new cell sites
are activated each day while other cells are relocated, modified or
even decommissioned. In order to enable consistent location
performance, the location platform must be able to respond quickly
to changes in the network.
[0009] Differences between countries also mean that services which
otherwise operate on a global basis, cannot access location
information on a global basis.
[0010] Increasingly consumers use different wireless networks at
different times, sometimes operated by different carriers.
Providing a consistent level of service to users in this
environment is also difficult. Even if each of the providers were
willing to make their location information available, the
differences in the systems (for instance Wi-Fi & cellular) mean
that it may be difficult for an application developer to develop
against a common location capability set.
[0011] The issue has been solved to a limited extent by the
introduction of GPS which can operate in each handset independently
of the carrier. In practice however the limitations of GPS
necessitate assistance information (e.g., A-GPS) which typically
requires coarse cellular data to bootstrap the process. In this
case, the assistance server needs to have a database for all the
wireless carriers, with information covering all the cells in the
network. This again may be hard to achieve without negotiating with
all the MNOs. Some providers (Nokia for example) have sought to
provide a common assistance service however currently it's coverage
is limited to certain carriers only in limited regions.
[0012] A fundamental difficulty in getting coverage across all
radio networks is that the radio network configuration information
is proprietary to the mobile network operator (MNO), and embodies
proprietary technical and commercial information. MNOs are
typically unwilling to provide this information to a third party
for aggregation unless they can receive a revenue stream. Having to
source databases from different MNOs in different areas and
countries to support a wide variety of applications leads to a very
complex value chain and to date has been a commercial obstacle.
[0013] The result is that despite the significant social and
commercial benefits that could accrue from a common wireless
location service, such a service has not been available until the
recent release of services such as Google Maps for Mobile and
particularly the feature within known as My Location. Despite the
advances in such systems however there remain significant
limitations for users. For example, one such limitation is the slow
response to changes in the configuration of the radio network. For
example it has been observed that a change in the configuration of
some 3G cells in a region of London resulted in a substantial loss
of coverage for a period lasting at least 12 days. This was despite
repeated requests from handsets with a GPS activated, in an attempt
to "seed" the cell database. For users of such systems it is often
important that the system can detect and respond to changes in the
wireless network rapidly, thereby maintaining the quality of the
location service.
[0014] The limitations described in the previous paragraphs and
other limitations can be overcome using systems, methods and/or
devices which can service users across multiple networks, without
requiring information from the operators of those networks, in
which the information concerning those networks is collected and
maintained in a current state efficiently and with rapid response
to changes in the network configuration. For the foregoing reasons,
there is a need for systems and methods which can use available
information, including the measurements reported in user location
requests to refine the network model and respond to significant
changes. There is also a need for systems, methods and/or devices
that can combine information from disparate sources to update
network models. Since many of the applications or services
utilizing such location systems will operate on an international
scale, the number of transmitters to be modeled will be very large
and there is also therefore a need for systems, methods and devices
that are scalable.
SUMMARY
[0015] Certain embodiments of the present disclosure provide
methods, systems, and processor-readable media to collect and
maintain a global repository of wireless network configuration
information. Certain embodiments also use this repository to offer
a facility for applications to determine the location of a cellular
device regardless of the country in which it is situated or the
network it is registered with or the radio access technology in
use.
[0016] Certain embodiment of the present disclosure provide
systems, methods, process-readable media and/or devices which can
service users across multiple networks, without requiring
information from the operators of those networks, in which the
information concerning those networks is collected and maintained
in a current state efficiently and with rapid response to any
changes in the network configuration. In certain aspects, it may be
possible to combine the disclosed systems, methods,
process-readable media and/or devices and use in combination with
information from the operators of the networks.
[0017] Certain embodiments disclose, a computer-implemented method
of aggregating wireless network transmitter characteristics
corresponding to a plurality of wireless network transmitters
comprising the steps of: receiving a report from a mobile radio
terminal, wherein the report includes at least one measurement of
at least one radio network parameter from a first wireless network
transmitter and GPS location information; and estimating or
updating a set of characteristics for the first wireless network
transmitter based at least in part on the reported measurements and
the GPS location information. In certain embodiments, the set of
characteristics include one or more location dependent parameters
describing the coverage footprint of the first wireless network
transmitter. In certain embodiments, the method includes the step
of storing the set of characteristics for the first wireless
network transmitter. In certain embodiments, the one or more
location dependent characteristics comprise a mean and a covariance
of a 2D Gaussian probability density function approximating the
coverage footprint of the first wireless network transmitter.
[0018] Certain embodiments disclose a computer-implemented method
of updating a database having characteristics of a plurality of
wireless network transmitters comprising the steps of: receiving a
report from a mobile radio terminal, wherein the report includes
one or more measurements of at least one radio parameter
corresponding to a first wireless network transmitter; and one or
more measurements of at least one radio parameter corresponding to
a second wireless network transmitter; estimating a location for
the radio terminal based at least in part on the report and
characteristics corresponding to the first and second wireless
transmitters; and estimating or updating a set of characteristics
for the first wireless network transmitter based at least in part
on the reported measurements and the estimated location. In certain
aspects of the disclosed methods the first wireless network
transmitter is selected from a group consisting of a GSM cell, a
Wi-Fi Access Point, a CDMA cell, a UMTS cell, an LTE cell or a
WiMax Base Station. In certain embodiments, the disclosed methods
comprise the further steps of: estimating a set of characteristics
for the second wireless network transmitter based at least in part
on the estimated location of the radio terminal and the at least
one measurement of at least one radio network parameter of the
second wireless network transmitter; and storing the set of
characteristics for the second wireless network transmitter. In
certain aspects, the first and the second wireless network
transmitters are selected from a group consisting of GSM, UMTS,
LTE, or Mobile WiMAX transmitters. In certain aspects, the first
wireless network transmitter is a cellular base station or a Wi-Fi
Access Point and the second wireless network transmitter is a
cellular base station or a Wi-Fi Access Point. In certain
embodiments the disclosed methods further comprise the step of
authenticating a client on the mobile radio terminal. In certain
embodiments, estimating the location of the mobile radio terminal
comprises: applying a cost function to the at least one measurement
of the at least one radio network parameter of the first wireless
network transmitter and one or more characteristics associated with
the at least one wireless network transmitter; and minimizing a sum
of costs generated by the cost function to determine the estimated
location of the mobile radio terminal. In certain embodiments, the
estimating of the location of the mobile radio terminal further
comprises: applying a cost function to the at least one measurement
of the at least one radio network parameter of the first wireless
network transmitter and to the at least one measurement of the at
least one radio network parameter of the second wireless network
transmitter and one or more characteristics associated with the
first wireless network transmitter and/or one or more
characteristics associated with the second wireless network
transmitter; and minimizing a sum of costs generated by the cost
function to determine the estimated location of the mobile radio
terminal. In certain aspects, the at least one measurement of at
least one radio network parameter of the first wireless network
transmitter includes an identifier. In certain aspects, the report
is received via a bearer selected from the group consisting of SMS,
GPRS, EDGE, HSPA, and EV-DO.
[0019] Certain disclosed embodiments relate to a processor-readable
medium storing processor executable code for causing a processor to
perform the steps of: receiving a report from a radio terminal,
wherein the report includes at least one measurement of at least
one radio network parameter from a first wireless network
transmitter and at least one measurement of at least one radio
network parameter from a second wireless network transmitter;
estimating a location for the radio terminal based at least in part
on the report; estimating a set of characteristics for the second
wireless network transmitter based at least in part on the
estimated location of the mobile radio terminal and the at least
one measurement of at least one radio network parameter. In certain
aspects the processor readable medium storing processor executable
code causes the processor to perform the additional steps of:
storing the set of characteristics for the first wireless network
transmitter; updating a list of tasking information to reflect the
stored set of characteristics for the first wireless network
transmitter; and communicating the list of tasking information to a
plurality of radio terminals.
[0020] Certain embodiments disclose computer-implemented methods of
updating a database having characteristics of a plurality of
wireless network transmitters comprising the steps of: storing a
first set of characteristics for a first wireless network
transmitter; storing a first set of characteristics for a second
wireless network transmitter; receiving a report from a radio
terminal, wherein the report includes at least one measurement of
at least one radio parameter corresponding to a first wireless
network transmitter; and at least one measurement of at least one
radio parameter corresponding to a second wireless network
transmitter; estimating a location for the radio terminal based at
least in part on the report and characteristics corresponding to
the first and second wireless transmitters; and determining a
second set of characteristics for the first wireless network
transmitter based on the estimated location of the mobile radio
terminal, the reported measurements, and the first sets of
characteristics corresponding to the first and second wireless
network transmitters. Certain embodiments further comprise the
steps of: determining that at least one characteristic from the
first set of characteristics has changed; and updating the at least
one characteristic that has changed. In certain aspects, the at
least one characteristic that has changed is a transmitter
identifier. Certain aspects, further comprises the steps of:
identifying at least one characteristic from the first set of
characteristics that is inaccurate; and adjusting the at least one
characteristic that is inaccurate to improve accuracy of the at
least one characteristic. In certain aspects, the at least one
characteristic that is inaccurate is a location dependant
characteristic of the first wireless network transmitter.
[0021] Certain embodiments, disclosed computer-implemented methods
of determining a location of a radio terminal comprising the steps
of obtaining an initial location estimate of a radio terminal;
communicating the initial location estimate to a server; receiving
a tile including a plurality of characteristics of wireless network
transmitters near the initial location estimate of the radio
terminal; measuring one or more radio network parameters from at
least one wireless network transmitter; applying a cost function to
the one or more measurements and one or more characteristics from
the tile corresponding to the at least one wireless network
transmitter; and minimizing a sum of costs generated by the cost
function to calculate the location of the mobile radio terminal.
Certain embodiments, disclose methods wherein the one or more
characteristics are parameters describing the reception footprint
of the at least one wireless transmitter. In certain aspects, the
one or more characteristics comprise a mean and a covariance of a
2D Gaussian probability density function approximating the coverage
footprint of the at least one wireless transmitter.
[0022] Certain embodiments, disclose methods of determining a
location of a mobile radio terminal comprising the steps of:
receiving a report from a mobile radio terminal, wherein the report
includes at least one measurement of at least one radio parameter
corresponding to a first wireless network transmitter; applying a
cost function to the one or more reported measurements and one or
more characteristics corresponding to the first wireless network
transmitter; and minimizing a sum of costs generated by the cost
function to calculate the location of the mobile radio terminal. In
certain aspects, the one or more characteristics are parameters
describing the reception footprint of the at least one wireless
transmitter. In certain aspects, the one or more characteristics
are a mean and a covariance of a 2D Gaussian probability density
function approximating the coverage footprint of the at least one
wireless transmitter. In certain aspects, the report further
includes at least one measurement of at least one radio parameter
corresponding to a second wireless network transmitter and wherein
the cost function is applied to the reported measurements and one
or more characteristics corresponding to the first wireless network
transmitter and characteristics corresponding to the second
wireless network transmitter. In certain aspects, the first
wireless network transmitter and the second wireless network
transmitter are selected from a group consisting of GSM cells, UMTS
cells, CDMA cells, LTE cells, WiMAX cells and Wi-Fi Access Points.
In certain aspects, the first wireless network transmitter is
connected to a wireless network of a first type and the second
wireless transmitter is connected to a wireless network of a second
type.
[0023] Certain embodiments disclose computer-implemented methods of
aggregating wireless network transmitter characteristics
corresponding to a plurality of wireless network transmitters
comprising the steps of: receiving a report from a mobile radio
terminal, wherein the report includes at least one measurement of
at least one radio network parameter from a first wireless network
transmitter and at least one measurement of at least one radio
network parameter from a second wireless network transmitter; and
estimating a set of characteristics for the first wireless network
transmitter based at least in part on the at least one measurement
of at least one radio network parameter and characteristics
corresponding to the second wireless network transmitter. In
certain aspects, the methods comprise the step of storing the set
of characteristics for the first wireless network transmitter. In
certain aspects, the first type of wireless network is a GSM, UMTS,
CDMA, LTE, WiMAX or Wi-Fi network and the second type wireless
network is a GSM, UMTS, CDMA, LTE, WiMAX or Wi-Fi network. In
certain aspects, the first wireless network transmitter is
connected to a first type of wireless network and the second
wireless network transmitter is connected to a second type of
wireless network.
[0024] Certain embodiments disclose a processor-readable medium
storing processor executable code for causing a processor to
perform the steps of: receiving a report from a radio terminal,
wherein the report includes at least one measurement of at least
one radio network parameter from a first wireless network
transmitter and at least one measurement of at least one radio
network parameter from a second wireless network transmitter;
estimating a set of characteristics for the first wireless network
transmitter based at least in part on the at least one measurement
of at least one radio network parameter and characteristics
corresponding to the second wireless network transmitter;
estimating a set of characteristics for the second wireless network
transmitter based at least in part on the at least one measurement
of at least one radio network parameter and characteristics
corresponding to the first wireless network transmitter; and
storing the set of characteristics for the first wireless network
transmitter and the set of characteristics for the second wireless
network transmitter.
[0025] Certain embodiments disclose processing systems for
aggregating wireless network transmitter characteristics
corresponding to a plurality of wireless network transmitters
comprising: at least one wireless network comprising a plurality of
wireless network transmitters; at least one radio terminal in
communication with said at least one wireless network; at least one
database containing characteristics corresponding to transmitters
of said at least one wireless network; and wherein the processing
system is configured to execute steps comprising: receiving a
report from the at least one radio terminal, wherein the report
includes at least one measurement of at least one radio network
parameter from a first wireless network transmitter and at least
one measurement of at least one radio network parameter from a
second wireless network transmitter; estimating a location for the
at least one radio terminal based at least in part on the report;
and estimating a set of characteristics for the first wireless
network transmitter based at least in part on the estimated
location of the radio terminal and the at least one measurement of
at least one radio network parameter.
[0026] Certain embodiments disclose processing systems that are
capable of determining a location of a radio terminal comprising:
at least one wireless network comprising a plurality of wireless
network transmitters; at least one mobile radio terminal in
communication with said at least one wireless network; at least one
database containing characteristics corresponding to transmitters
of said at least one wireless network; wherein the processing
system is configured to execute steps comprising: measuring in a
mobile radio terminal one or more radio network parameter
corresponding to at least one wireless network transmitter;
receiving at a server the radio network parameter measurements;
applying a cost function to the one or more measurements and one or
more characteristics corresponding to the at least one wireless
network transmitter; and minimizing a sum of costs generated by the
cost function to calculate the location of the mobile radio
terminal. In certain aspects, the one or more characteristics are
parameters describing the reception footprint of the at least one
wireless transmitter. In certain aspects, the one or more
characteristics are a mean and a covariance of a 2D Gaussian
probability density function approximating the coverage footprint
of the at least one wireless transmitter. In certain aspects, the
processing systems includes calculating the location of the radio
terminal comprises the steps of: applying a cost function to the at
least one measurement of the at least one radio network parameter
of the first wireless network transmitter; and minimizing a sum of
costs generated by the cost function to determine the location of
the mobile radio terminal.
[0027] Certain embodiments disclose computer-implemented methods of
discovering and/or maintaining wireless transmitter characteristics
for wireless transmitters in a wireless network comprising the
steps of: receiving from a mobile radio terminal, at least one
measurement of at least one radio parameter corresponding to a
first wireless transmitter and at least one measurement of at least
one radio parameter corresponding to at least one other wireless
transmitter; retrieving a set of characteristics for the at least
one other transmitter from a database; and estimating or updating
in a processing system a set of characteristics for the first
wireless transmitter based at least in part on the characteristics
of the at least one other transmitter, and the collected
measurements corresponding to the first wireless transmitter and
the at least one other wireless transmitter. In certain aspects,
the set of characteristics for the first wireless transmitter
include location dependent characteristics describing the coverage
footprint of the first wireless transmitter. In certain aspects,
the set of characteristics for the first wireless transmitter
include a mean and a covariance of a 2D Gaussian probability
density function approximating the coverage footprint of the first
wireless transmitter. In certain embodiments, the
computer-implemented methods of estimating and/or updating a set of
characteristics for a wireless transmitter in a wireless network
include the steps of: retrieving from a database one or more
parameters that characterize a wireless transmitter; estimating a
first coverage footprint for the wireless transmitter by radio
propagation modeling using the one or more parameters; deriving
from the first coverage footprint a first set of sample points
characterizing the footprint; deriving a second set of sample
points from at least one other source; and combining the first set
of sample points with the second set of sample points to obtain a
set of characteristics for the wireless transmitter.
[0028] Certain embodiments disclose computer-implemented methods of
collecting one or more radio measurements of wireless transmitters
in order to update a database of wireless transmitter
characteristics comprising the steps of: measuring at a mobile
radio terminal a sequence of one or more radio measurements;
reporting a selected subset of the measurements to a server,
wherein the reported subset of measurements are used to update the
database of transmitter characteristics; wherein the selection
process results in less traffic on a network and/or less server
processing for a given increment in the quality of the
database.
[0029] Certain embodiments disclose computer-implemented methods of
collecting one or more radio measurements in order to update a
database of wireless transmitter characteristics comprising:
measuring at a mobile radio terminal a sequence of one or more
radio measurements; reporting a selected subset of the measurements
to a server; wherein the reported subset of measurements are used
to update the database of transmitter characteristics; and wherein
the selection process improves the updating of the database by
reducing the amount of data transmitted, reducing the amount of
server processing, incrementally improving the coverage of the
updated database, or combinations thereof.
[0030] Certain embodiments disclose computer-implemented methods of
discovering and/or maintaining wireless transmitter characteristics
of wireless transmitters in a wireless network comprising the steps
of: receiving from a mobile radio terminal at least one measurement
of at least one radio parameter corresponding to a first wireless
transmitter and at least one measurement of at least one radio
parameter corresponding to at least one other wireless transmitter;
estimating or updating a set of characteristics for the first
wireless transmitter based on the characteristics of the at least
one other transmitter and temporal associations between the
measurements of the first wireless transmitter and the at least one
other wireless transmitter; storing in a database the set of
characteristics for the first wireless transmitter; and wherein the
first and other transmitters belong to different types of wireless
networks, and wherein the set of characteristics for the first
wireless transmitter describe the coverage footprint of the
transmitter.
[0031] Certain embodiments disclose a computer-implemented method
of discovering and/or maintaining wireless transmitter
characteristics of wireless transmitters in a wireless network
comprising the steps of: receiving from a mobile radio terminal at
least one measurement of at least one radio parameter corresponding
to at least one wireless transmitter; obtaining from a user of the
mobile radio terminal an approximate location of the mobile radio
terminal at the time the measurement was collected; estimating or
updating a set of characteristics for the at least one wireless
transmitter based on the measurements and the approximate location;
and storing in a database the set of characteristics for the at
least one wireless transmitter
[0032] Certain embodiments disclose devices for use in discovering
and/or maintaining wireless transmitter characteristics comprising:
a processing system; a plurality of wireless receivers collectively
compatible with two or more different wireless networks coupled to
the processing system; at least one positioning receiver coupled to
the processing system; a memory coupled to the processing system;
and wherein the device is capable of periodically collecting at
least one radio parameter measurement corresponding to wireless
transmitters from the two or more different wireless networks
together with a position measurement, and a timestamp; and storing
the at least one radio parameter measurement, position measurement,
and timestamp in a log file, wherein the log file can be accessed
by a network learning server to estimate a footprint of the
wireless transmitters of the two or more different wireless
networks. In certain aspects, at least two wireless networks are
cellular networks and at least one of the wireless networks is a
Wi-Fi network.
[0033] Certain embodiments disclose methods of discovering and/or
maintaining wireless transmitter characteristics wherein
contemporaneous measurements corresponding to two or more unknown
transmitters can be exploited in the discovery process, comprising
the steps of obtaining from a mobile radio terminal at least one
measurement of at least one radio parameter corresponding to a
first wireless transmitter and at least one measurement of at least
one radio parameter corresponding to a second wireless transmitter,
wherein a database does not include any information regarding
characteristics of the first or second wireless transmitter;
storing in the database a reference measurement for the first
wireless transmitter with respect to the second wireless
transmitter, without defining any location dependent
characteristics of the first wireless transmitter; receiving
location dependent characteristics of the second wireless
transmitter; estimating location dependent characteristics for the
first transmitter based on the location dependent characteristics
of the second transmitter; and storing in the database the set of
characteristics for the first wireless transmitter.
[0034] Certain embodiments disclose methods for discovering and/or
maintaining a database of wireless transmitter characteristics in
which an indication is provided about the current coverage of a
database, comprising the steps of: receiving from a mobile radio
terminal at least one measurement of at least one radio parameter
corresponding to a first wireless transmitter, wherein the database
does not include any information regarding characteristics of the
first wireless transmitter and no additional information is
available to estimate location dependent characteristics of the
first wireless transmitter; storing in the database the at least
one measurement; querying the database to determine unique
transmitter identities of transmitters for which no location
dependent characteristics have been estimated; and reporting a
count and/or the unique transmitter identities of the unknown
transmitters to operators of the system as an indication of the
current level of coverage of the database.
[0035] Certain embodiments disclose computer-implemented methods
for discovering and/or maintaining a database of wireless
transmitter characteristics in which duplicate use of transmitter
identifiers is automatically detected and managed, comprising the
steps of: receiving from one or more radio terminals at least one
measurement of at least one radio parameter corresponding to one or
more wireless transmitters; selecting from the at least one
measurement one or more reference measurements characterizing a
location dependent characteristic of the one or more wireless
transmitters, wherein each of the reference measurements includes a
transmitter identifier and an approximate date and time associated
with the collection of the reference measurement; storing in a
database the one or more reference measurements; analyzing
reference measurements having a same transmitter identifier to
determine whether a pattern of the reference measurements indicates
the presence of two or more physical devices using the same
transmitter identifier; and if multiple devices are indicated,
suppressing that transmitter identifier from the database for use
in location estimation.
[0036] Certain embodiments disclose a computer-implemented method
for discovering and/or maintaining a database of wireless
transmitter characteristics in which duplicate use of transmitter
identifiers is automatically detected and managed, comprising the
steps of: receiving from one or more mobile radio terminals at
least one measurement of at least one radio parameter corresponding
to one or more wireless transmitters; selecting from the at least
one measurement one or more reference measurements characterizing a
location dependent characteristic of the one or more wireless
transmitters, wherein each of the reference measurements includes a
transmitter identifier and an approximate date and time associated
with the collection of the reference measurement; storing in a
database the one or more reference measurements; analyzing
reference measurements having a same transmitter identifier to
determine whether a pattern of the reference measurements indicates
the presence of two or more physical devices using the same
transmitter identifier; and if multiple devices are indicated,
suppressing that transmitter identifier from the database for use
in location estimation.
[0037] Certain embodiments disclose a computer-implemented method
for discovering and/or maintaining a database of wireless
transmitter characteristics in which particular characteristics of
certain devices can be accommodated in the processing, comprising
the steps of: obtaining from an administrator one or more sets of
device options comprising a device type identifier and one or more
processing options; receiving from a device incorporating at least
one wireless receiver, at least one measurement of at least one
radio parameter corresponding to at least one wireless transmitter
associated with a device type identifier; determining whether the
device type identifier is included in any of the sets of device
options; estimating or updating a set of characteristics for the at
least one wireless transmitter based in part on the at least one
measurements, wherein the processing is modified according to any
processing options associated with the corresponding device type
identifier; and storing the transmitter characteristics in a
database for use in estimating the location of mobile wireless
devices.
[0038] Certain embodiments disclose computer-implemented methods of
discovering and/or maintaining wireless transmitter characteristics
in which measurements corresponding to a certain type of device
subsequently found to be unreliable may be modified or excluded
from the processing, comprising the steps of: obtaining from a
device incorporating at least one wireless receiver, at least one
measurement of at least one radio parameter corresponding to at
least one wireless transmitter associated with a device type
identifier; deriving a set of characteristics for the at least one
wireless transmitter based at least in part on the at least one
measurement; storing in a database one or more reference
measurements corresponding to the at least one transmitter, wherein
each reference measurement includes the device type identifier
corresponding to the device; and if a type of device is discovered
to report unreliable measurements, then modifying and/or
suppressing reference measurements including the device type
identifier corresponding to the unreliable device type in
processing to estimate transmitter characteristics using the stored
reference measurements.
[0039] Certain embodiments disclose processing systems for
discovering and/or maintaining wireless transmitter characteristics
corresponding to at least one wireless network comprising: at least
one wireless network comprising a plurality of wireless network
transmitters; at least one mobile radio terminal in communication
with said at least one wireless network; at least one database
containing characteristics corresponding to transmitters of said at
least one wireless network; and wherein the processing system is
configured to execute steps comprising: receiving a report from the
at least one radio terminal, wherein the report includes at least
one measurement of at least one radio network parameter from a
first wireless network transmitter; estimating a location for the
at least one radio terminal based at least in part on the report;
estimating a set of characteristics for the first wireless network
transmitter based at least in part on at least one measurement of
at least one radio network parameter and characteristics
corresponding to a second wireless transmitter; and storing the set
of characteristics for the first wireless network transmitter in
the at least one database. In certain aspects, the steps of
estimating the set of characteristics and storing the set of
characteristics are performed periodically at intervals of: between
10 minutes and 1 hour, 30 minutes and 4 hours, 1 hour and 12 hours,
6 hours and 24 hours, 12 hours and 48 hours, or 1 day and 7 days.
In certain aspects, the steps of estimating the set of
characteristics and storing the set of characteristics are
performed automatically when a configurable number of new
measurements have been received. In certain aspects, the steps of
estimating the set of characteristics and storing the set of
characteristics are performed automatically when measurements
corresponding to a configurable number of different transmitters
have been received. IN certain aspects, the steps of estimating the
set of characteristics and storing the set of characteristics are
performed automatically when the processing of all existing
measurements has been completed.
[0040] Certain embodiments disclose processing systems for
discovering and/or maintaining wireless transmitter characteristics
corresponding to at least one wireless network wherein separate
deployments of a same processing system can share reference
measurements by exporting some or all of the reference measurements
from a first deployment and importing some or all of the exported
measurements into a second deployment, the processing system
comprising: at least one wireless network comprising a plurality of
wireless network transmitters; at least one mobile radio terminal
in communication with said at least one wireless network; at least
one database containing characteristics corresponding to
transmitters of said at least one wireless network; and wherein the
processing system is configured to execute steps comprising:
receiving a report from the at least one radio terminal, wherein
the report includes at least one measurement of at least one radio
network parameter from a first wireless network transmitter and at
least one measurement of at least one radio network parameter from
a second wireless network transmitter; estimating a location for
the at least one radio terminal based at least in part on the
report; estimating a set of characteristics for the first wireless
network transmitter based at least in part on at least one
measurement of at least one radio network parameter and
characteristics corresponding to a second wireless transmitter;
storing the set of characteristics for the first wireless network
transmitter in the at least one database; and wherein separate
deployments of a same processing system can share reference
measurements by exporting some or all of the reference measurements
from a first deployment and importing some or all of the exported
measurements into a second deployment.
[0041] In certain embodiments, methods, systems, and
processor-readable media in accordance with the present disclosure
are characterized by the steps of receiving a report from a mobile
radio terminal, wherein the report includes at least one
measurement of at least one radio network parameter from a first
wireless network access point; estimating a location for the mobile
radio terminal based on the report; estimating a set of
characteristics for the first wireless network access point based
on the estimated location of the mobile radio terminal and/or the
at least one measurement of at least one radio network parameter;
storing the set of characteristics for the first wireless network
access point; updating a list of tasking information with a quality
of the set of characteristics for the first wireless network access
point; and communicating the list of tasking information to a
plurality of mobile radio terminals.
[0042] In certain embodiments, methods, systems, and
processor-readable media in accordance with the present disclosure
are characterized by the steps of receiving a report from a mobile
radio terminal, wherein the report includes at least one
measurement of at least one radio network parameter from a first
wireless network transmitter; estimating a location for the mobile
radio terminal based on the report; estimating a set of
characteristics for the first wireless network transmitter based on
the at least one measurement of at least one radio network
parameter. In certain aspects, the set of characteristics for the
first wireless network transmitter is stored. In certain aspects, a
list of tasking information with a quality of the set of
characteristics for the first wireless network transmitter is
updated. In certain aspects, the list of tasking information is
communicated to a plurality of mobile radio terminals.
[0043] In certain embodiments, methods, systems, processor-readable
media and/or devices in accordance with the present disclosure are
characterized by the steps of receiving a report from a mobile
radio terminal, wherein the report includes at least one
measurement of at least one radio network parameter from a first
wireless network transmitter; estimating a location for the mobile
radio terminal based on the report; estimating a set of
characteristics for the first wireless network transmitter based on
the at least one measurement of at least one radio network
parameter; storing the set of characteristics for the first
wireless network transmitter; and updating a list of tasking
information with a quality of the set of characteristics for the
first wireless network transmitter. In certain aspects the
disclosed methods, systems, processor-readable media and/or devices
may include communicating the list of tasking information to a
plurality of mobile radio terminals. In certain aspects, not
relying on the estimated location of the mobile radio terminal may
lead to several advantages in performance and/or quality control of
the network learning collection process.
[0044] In further embodiments, methods, systems, and
processor-readable media in accordance with the present disclosure
are characterized by the steps of receiving a list of tasking
information, wherein the list of tasking information includes a
quality of a set of characteristics of a first wireless network
transmitter; measuring at least one radio network parameter from
the first wireless network transmitter; if the quality of the set
of characteristics for the first wireless network transmitter is
low, generating a report having at least one measurement of at
least one radio network parameter from the first wireless network
transmitter; and communicating the report to a server.
[0045] In still further embodiments, methods, systems, and
processor-readable media in accordance with the present disclosure
are characterized by the steps of storing a first set of
characteristics for a first wireless network transmitter; receiving
a report from a mobile radio terminal, wherein the report includes
at least one measurement of at least one radio network parameter
from the first wireless network transmitter; estimating a location
for the mobile radio terminal based on the report; determining a
second set of characteristics for the first wireless network
transmitter based on the at least one measurement of at least one
radio network parameter, and the first set of characteristics;
storing the second set of characteristics; updating a list of
tasking information with a quality of the second set of
characteristics for the first wireless network transmitter; and
communicating the list of tasking information to a plurality of
mobile radio terminals.
[0046] In still further embodiments, methods, systems, and
processor-readable media in accordance with the present disclosure
are characterized by the steps of storing a first set of
characteristics for a first wireless network transmitter; receiving
a report from a mobile radio terminal, wherein the report includes
at least one measurement of at least one radio network parameter
from the first wireless network transmitter; estimating a location
for the mobile radio terminal based on the report; determining a
second set of characteristics for the first wireless network
transmitter based on the at least one measurement of at least one
radio network parameter, and the first set of characteristics;
storing the second set of characteristics; updating a list of
tasking information with a quality of the second set of
characteristics for the first wireless network transmitter; and
communicating the list of tasking information to a plurality of
mobile radio terminals.
[0047] In still further embodiments, methods, systems, and
processor-readable media in accordance with the present disclosure
are characterized by the steps of obtaining an initial location
estimate of a mobile radio terminal; communicating the initial
location estimate to a server; receiving a tile including a
plurality of characteristics of wireless network transmitters near
the initial location estimate of the mobile radio terminal;
measuring at least one radio network parameter from a first
wireless network transmitter; and calculating a location of the
mobile radio terminal based on the tile and the measurement of the
at least one radio network parameter from the first wireless
network transmitter.
[0048] In yet further embodiments, methods, systems, and
processor-readable media in accordance with the present disclosure
are characterized by the steps of receiving a report from a mobile
radio terminal, wherein the report includes at least one
measurement of at least one radio network parameter from a first
wireless network transmitter and at least one measurement of at
least one radio network parameter from a second wireless network
transmitter, wherein the first wireless network transmitter is
connected to a first type of wireless network and the second
wireless network transmitter is connected to a second type of
wireless network; estimating a location for the mobile radio
terminal based on the report; estimating a set of characteristics
for the first wireless network transmitter based on the estimated
location of the mobile radio terminal and/or the at least one
measurement of at least one radio network parameter; estimating a
set of characteristics for the second wireless network transmitter
based on the set of characteristics for the first wireless network
transmitter; and storing the set of characteristics for the first
wireless network transmitter and the set of characteristics for the
second wireless network transmitter.
[0049] In yet further embodiments, methods, systems, and
processor-readable media in accordance with the present disclosure
are characterized by the steps of receiving a report from a mobile
radio terminal, wherein the report includes at least one
measurement of at least one radio network parameter from a first
wireless network transmitter and at least one measurement of at
least one radio network parameter from a second wireless network
transmitter, wherein the first wireless network transmitter is
connected to a first type of wireless network and the second
wireless network transmitter is connected to a second type of
wireless network; optionally estimating a location for the mobile
radio terminal based on the report; updating a set of
characteristics for the first wireless network transmitter based on
the reported measurements; and an existing set of characteristics
for the second wireless network transmitter; and storing the
updated set of characteristics for the first wireless network
transmitter.
BRIEF DESCRIPTION OF THE DRAWINGS
[0050] These and other features, aspects, and advantages disclosed
herein will become better understood with regard to the following
description, appended claims, and accompanying drawings where:
[0051] FIGS. 1(a) and 1(b) illustrates an exemplary system
architecture in accordance with certain embodiments;
[0052] FIG. 2 illustrates an exemplary network model life cycle in
accordance with certain embodiments;
[0053] FIG. 3 illustrates an exemplary logic flow for determining
whether to generate a cell report using GPS location reference in
accordance with certain embodiments;
[0054] FIG. 4 illustrates exemplary logic for determining whether
to generate a report for one cell using another cell as a location
reference in accordance with certain embodiments;
[0055] FIG. 5 illustrates an exemplary logic flow for determining
whether to generate a report of a Wi-Fi AP using GPS location
reference in accordance with certain embodiments;
[0056] FIG. 6 illustrates an exemplary logic flow for determining
whether to generate a report of a Wi-Fi AP using a cell as a
location reference in accordance with certain embodiments;
[0057] FIG. 7 illustrates an exemplary logic flow for determining
whether to generate a report of a Wi-Fi AP using a cell as a
location reference in accordance with certain embodiments;
[0058] FIG. 8 illustrates an exemplary logic flow for determining
whether to generate a report of a Wi-Fi AP using another AP as a
location reference in accordance with certain embodiments;
[0059] FIG. 9 illustrates an exemplary cell model life cycle in
accordance with certain embodiments;
[0060] FIG. 10 illustrates an exemplary pre-processor in accordance
with certain embodiments;
[0061] FIG. 11 illustrates an exemplary flow chart for
preprocessing of a measurement in accordance with certain
embodiments;
[0062] FIG. 12 illustrates an exemplary 2D Gaussian fit to observed
locations for cell ID only measurements in accordance with certain
embodiments;
[0063] FIG. 13 illustrates an exemplary 2D Gaussian fit to observed
locations for cell ID and rxLev measurements in accordance with
certain embodiments;
[0064] FIG. 14 illustrates an exemplary horizontal gain pattern for
900 MHz cellular antenna;
[0065] FIG. 15 illustrates an exemplary generation of a reference
measurement when 2 cellular transmitters are observed;
[0066] FIG. 16 illustrates an exemplary generation of a reference
measurement when 2 Wi-Fi access point transmitters are
observed;
[0067] FIG. 17 illustrates an exemplary generation of a reference
measurement for a cell using the footprint of a Wi-Fi access point
transmitter;
[0068] FIG. 18 illustrates an exemplary generation of a reference
measurement for a Cellular Transmitter using GPS;
[0069] FIG. 19 illustrates an exemplary transmitter footprint for a
directional antenna;
[0070] FIG. 20 illustrates an exemplary transmitter footprint for
an omni-directional antenna;
[0071] FIG. 21 illustrates an exemplary set of reference
measurements for a directional cellular transmitter and the
resulting model of the transmitter's footprint;
[0072] FIG. 22 illustrates an exemplary logic flow for resolving a
deferred location estimate in a reference measurement
[0073] FIG. 23 illustrates an exemplary set of non-uniform
spatially sampled reference measurements leading to a bias in the
transmitter footprint model;
[0074] FIG. 24 illustrates an exemplary use of a grid overlay to
reduce the effect of non-uniform sampling on the transmitter
footprint;
[0075] FIG. 25 illustrates an exemplary logic flow for using a grid
overlay to reduce the effect of non-uniform sampling on the
transmitter footprint;
[0076] FIG. 26 illustrates an exemplary discrete propagation model
being used to predict and model the coverage of a cellular
transmitter;
[0077] FIG. 27 illustrates an exemplary use of a transmitter
footprint model to generate a set of reference measurements;
[0078] FIG. 28 illustrates an exemplary logic flow for creating
reference measurements using a third party network configuration
data;
[0079] FIG. 29 illustrates an exemplary location estimate for 3
cellular measurements and 2D Gaussian PDF transmitter models;
[0080] FIG. 30 illustrates an exemplary location estimate for 3
cellular measurements and 1 Wi-Fi access point measurement using 2D
Gaussian PDF transmitter models;
[0081] FIG. 31 illustrates an exemplary location estimate using 3
cellular measurements adjusting for movement using 2D Gaussian PDF
transmitter models;
[0082] FIG. 32 illustrates an exemplary network architecture in
accordance with certain embodiments;
[0083] FIG. 33 illustrates another exemplary network architecture
in accordance with certain embodiments;
[0084] FIG. 34 illustrates still another exemplary network
architecture in accordance with certain embodiments;
[0085] FIG. 35 illustrates an exemplary network architecture in
accordance with certain embodiments;
[0086] FIG. 36 illustrates an exemplary client application in
accordance with certain embodiments;
[0087] FIG. 37 illustrates another exemplary client application in
accordance with certain embodiments; and
[0088] FIG. 38 illustrates such an exemplary device that may be
used to collect focused measurements.
DETAILED DESCRIPTION
[0089] Certain embodiments of the present disclosure will now be
described in detail, examples of which are illustrated in the
accompanying drawings. The examples and embodiments are provided by
way of explanation only and are not to be taken as limiting to the
scope of the inventions. Furthermore, features illustrated or
described as part of one embodiment may be used with one or more
other embodiments to provide a further new combination. It will be
understood that the present inventions will cover these variations
and embodiments as well as variations and modifications that would
be understood by the person skilled in the art
[0090] In accordance with certain embodiments, the term
"aggregating" is used to describe the combined process of
collecting and then maintaining information characterizing one or
more wireless transmitters in a network. Other synonymous terms
used in the description include discovering and updating.
[0091] In accordance with certain embodiments, the terms "wireless
network access point", "transmitter" and "cell" are used to refer
to a wireless network transceiver that provides wireless network
access to one or more mobile wireless devices. For example, in a
GSM, CDMA and UMTS networks this corresponds to a BTS or base
station. In a Wi-Fi network, this corresponds to an Access Point
(AP). In a WiMAX network the term base station is also used.
[0092] In accordance with certain embodiments, the terms "carrier",
"operator", and "mobile network operator", are used to refer to an
entity that operates a wireless network. Such an operator may run
one or more networks. For example an operator may run a 2G GSM
cellular network, 3G UMTS cellular network and also a network Wi-FI
"hot spots".
[0093] In accordance with certain embodiments, the term "mobile
wireless device" is used synonymously with terms such as "mobile
radio terminal," "radio terminal", "mobile station," "mobile
phone," "user equipment," "cell phone," or "handset" and
encompasses any kind of mobile radio terminal including Personal
Digital Assistants (PDAs), laptop and other mobile computers, and
pagers. The mobile wireless device may be any type of handset or
PDA and may operate over any radio communications network such as
GSM, UMTS, or CDMA. A wireless device typically includes a display,
a network transceiver, a central processing unit (CPU), a memory
(e.g., SDRAM), a Subscriber Identity Module (SIM) card, a data
storage unit, an antenna, and one or more inputs such as a keypad
or touch screen. In certain embodiments the handset may include a
Wi-Fi transceiver. In certain embodiments the handset may include a
GPS transceiver. In certain embodiments, CDMA handsets may include
a Removable User Identity Module (R-UIM) and UMTS handsets may
include a Universal Subscriber Identity Module (USIM). In certain
embodiments, the handset may include a WiMax transceiver. In
certain embodiments, the handset may include an LTE transceiver. In
certain embodiments, the handset may include a combination of
various transceivers.
[0094] In accordance with certain embodiments, the SIM card is a
specific instance of a smart card or security/trust token for
secure wireless communication networks, i.e., in this instance for
the GSM network. Other representative examples of smart cards for
secure wireless communication networks include the Universal
Identity Module (UIM), the Removable User Identity Module (R-UIM),
and the UMTS Subscriber Identity Module (USIM). The SIM represents
the subscription contract between a specific subscriber (network
user) and the GSM network operator, i.e., providing the means for
authenticating the subscriber for network access and identifying
GSM network services to which the subscriber is entitled, i.e., the
SIM card is the subscriber's identity in the context of the GSM
network. The SIM card is portable to any GSM terminal, thereby
providing the subscriber with an unprecedented degree of personal
mobility.
[0095] The SIM card is in fact a small computer, containing a
standardized operating system (JavaCard.TM. is implemented in the
SIM card; Smart Card for Windows and Multos.TM. are other
standardized operating systems for smart cards) and system files,
RAM and flash memory (for storage of data and applications), a
microprocessor, and typically a cryptographic co-processor. The GSM
network operator controls the distribution and the stored content,
e.g., data, applications, of the SIM card. Content on the SIM card
may be provisioned by one or more of the network operator, the
handset manufacturer, the SIM card manufacturer, or the subscribers
themselves (via, for example, WAP Push, or direct USB download).
Stored on SIM cards configured for GSM networks are subscription
and security-related data, e.g., a subscriber number (International
Mobile Subscriber Identity (IMSI)) that uniquely identifies the
subscriber, a network operator-assigned subscriber-specific call
number (MSISDN), i.e., the subscriber's `phone number` in the GSM
network, the subscriber key and cryptographic algorithms for
authentication of the subscriber and encryption of subscriber
communications (specified by the GSM network operator), and
subscriber personal data, e.g., the subscriber's password or
personal identity number (PIN) for accessing the SIM card, personal
telephone directory, call charging information, a log of
recently-dialed numbers, short text messages (for use with SMS
(Short Message Service)), and a personalized subscriber services
portfolio, i.e., applications.
[0096] Also embedded in the SIM card is a SIM Application Toolkit
(STK). The STK provides the functional capability, inter alia, to
allow the subscriber to access and use embedded applications via
the user interface of the GSM terminal, and to modify the menu
structure of the GSM terminal in conjunction with the use of such
applications. The STK also allows the GSM network operator to
download new data and/or applications to the SIM card to implement
new services for the subscriber.
[0097] In accordance with certain embodiments, the term client to
refer to a software application that is deployed on a mobile
wireless device to collect and report measurements to a server. The
client functionality could be implemented in a standalone
application or integrated as a component of another application.
The deployment could be in the form of a Symbian, J2ME, BREW,
Android, SIM Toolkit or other application, even embedded in the
firmware of the device. Such clients may be pre-installed on the
device or offered for download by the user via the internet. Yet
another mechanism for the clients to be deployed includes Over The
Air (OTA) mechanisms in which either at the user's request or at
the service provider's decision the client can be transferred
wirelessly onto the subscriber's device. Such mechanisms are
described for instance in 3GPP TS 23.048 and also the Open Mobile
Alliance OMA Download OTA Specification
(OMA-Download-OTA-V1.sub.--0-20040625-A). Yet a further
implementation option for the client is as a component of either
the device firmware or operating system.
[0098] In certain embodiments, the client could be a continuously,
or substantially continuously, running application which from time
to time collects information identifying nearby wireless network
transmitters and preferably also reference position information
such as from a GPS receiver. In certain alternative embodiments,
the client functionality could be implemented as part of an
application meaning that it runs only intermittently, when launched
by the device user.
[0099] A processor-readable medium as described herein can be any
form of data storage that is accessible by a computer processor.
This can include, for example, optical storage, magnetic storage,
RAM, flash ROM, ROM, or any other suitable medium.
[0100] FIG. 1(a) provides an exemplary system architecture in
accordance with certain embodiments. A mobile wireless device 10 is
shown which may be capable of connecting to either a cellular
wireless network 12 or other wireless network 14 (depicted by a
transmitter) or both. The device may also include a GPS location
capability (either mobile based or mobile assisted). A server 16 is
shown which connects to either or both of the cellular or other
wireless network. In the latter case the connection between the
wireless access layer and the server will usually be via the
internet 18. In the exemplary illustration of FIG. 1, the server 16
represents the aspects of the present disclosure that are
implemented on the network side. Also shown is an application
server 20, which may interact with the server 16 and/or mobile
wireless device 10 either to supply information for learning about
the radio network configuration or make use of services provided in
order to provide other services to wireless device users or other
third parties. Examples of such application servers 20 include
without limitation, mapping and/or navigation platforms, content
search platforms, mobile advertising platforms, voice-over-IP
(VOIP) gateways, instant messaging platform and presence
servers.
[0101] Additionally, while FIG. 1(a) shows a single cellular
network and a single Wi-Fi Access Point, in implementation this
could be any number of cellular and fixed wireless networks. FIG.
1(b) shows the same configuration but expands the server into three
logical components, a gateway 22, network database server 24, and
location server 26. In the following description, when referring to
one of these specific components we will use these terms. Further
details on the operation of each are provided below. It should be
noted that while the illustration in FIG. 1(b) shows a single
instance of each, the logical and physical configuration of a
practical deployment could feature one or more of each of these
elements depending on capacity & scaling requirements as well
as IT architecture & security constraints.
[0102] One method for learning about the location and
characteristics of wireless network transmitters, is to collect
contemporaneous, or substantially contemporaneous, wireless network
and GPS measurements from terminals that are equipped with a GPS.
In this case the GPS measurement provides the reference location
enabling the coverage footprint of the corresponding transmitter to
be estimated and optionally the location and some characteristics
of the transmitter to be estimated.
[0103] Client software on the wireless mobile device detects those
cases where the GPS and wireless network interface are active
simultaneously and collects measurements from both, then reporting
them to the server. This reporting could be via any one of several
channels including, for example, TCP/IP connection over Wi-Fi,
TCP/IP over cellular, over wireless CS connection or via SMS, or
other suitable combination thereof.
[0104] In embodiments where there is wide distribution of client
software, this opportunistic approach to measurement collection
will provide a steady stream of measurements. However, it may be
unlikely that transmitters designed to provide primarily indoor
coverage would be reported together with GPS because of the
typically poor performance of GPS indoors.
[0105] Because simultaneous network connection for such
transmitters and GPS coverage may be infrequent, the client may
also retain the most recent wireless and or GPS measurements. In
cases where wireless and GPS measurements are observed not
contemporaneously but within an acceptably short interval of each
other, then a report may also be sent. In certain instances an
acceptably short interval may be an interval less than 5, 10, 15,
20, 30, or 60 seconds, which serves to limit the possible range of
movement of the device over the interval. In other instances an
interval may be deemed to be acceptably short if it can be
determined through other means that the mobile terminal has not
moved during the interval. One example is where a particular Wi-Fi
Access Point or other wireless transmitter typically having a
detection range of less than say 50, 100, 150, or 200 m continues
to be detected throughout the interval. If a GPS fix becomes
available without a network connection being active, the client can
store the measurements for transmission to the server at a later
time.
[0106] In certain embodiments, an enhancement is to provide a
mechanism to drive the acquisition process so that valuable
information can be reported, while less valuable information is
not. Focusing the reporting in this way can minimize the signaling
required from subscriber terminals and also reduce the
infrastructure capacity required to deal with reports while at the
same time achieving fast acquisition of the radio network
information.
[0107] Certain embodiments of the present disclosure provide a
mechanism for this by communicating information on the state of the
current database. Typically in a GSM network, each cell is assigned
a unique cell ID, which is represented using 16 bits. This means
that there are 2 46 possible different cell IDs. The system can
communicate a list to the any or all of the clients that reflects
the cell IDs that the network database server currently has
information about. This list could be expressed as a bit map of 2
16 bits, where the value of each bit represents whether the server
already has sufficient information (also referred to herein as the
quality of the information) about a cell having the corresponding
cell ID or not. In some networks cell IDs may not be assigned
uniquely. In this case the cell ID plus LAC combination is usually
unique. Accordingly, in certain embodiments the tasking encoding
may be extended to accommodate LAC and cell ID.
[0108] Using this approach, the reporting efforts of the clients
can be focused on cells that are unknown thereby enabling the
system to acquire information on unknown cells more efficiently and
rapidly than simply adopting an opportunistic approach based on
subscriber usage of the location services. The information
describing the current state of the radio network model to guide
the reporting behavior of clients is hereafter referred to as
tasking information since it effectively describes the measurement
task to be completed by the clients. Tasking information may be
maintained for cellular networks where the range of cell IDs is
limited. Separate tasking information may be maintained for each
cellular network being modeled (and also separate information for
2G/3G). In certain embodiments, tasking information is provided to
a client only for the cellular network currently being used by that
device. In alternative embodiments, tasking information may be
provided for a network that the device is capable of using but is
not currently using (e.g., a 3G phone camped on a 2G cell). The
selection of which set of tasking information to provide to a
client is determined from the network identity reported in either
network reports or location measurements from the device. In cases
where the user travels to a different country and therefore roams
onto another network, the system may update the tasking information
if that network has been configured for monitoring. Otherwise, the
existing tasking information may be left in place.
[0109] Having received the tasking information, each client
continually monitors the serving cell currently camped on by the
device. If it encounters a cell ID which is marked in the tasking
list as unknown, then it records the information (including rxLev
if available) and prepares a report for transmission to the server.
In contrast to the simpler case where measurements are reported
opportunistically, this converts each client into an active
searcher, wherever the subscriber carries the device.
[0110] In certain embodiments the tasking information changes as
the network database model is updated in response to reports from
clients. Preferably the tasking information will be versioned in
sync with the corresponding network database model. In certain
embodiments the tasking information may be pushed to any or all
active clients whenever it is updated. In an alternative
embodiment, the availability of an updated version of the tasking
information may be signaled to any or all clients, enabling clients
to pull the updated information from the server at a suitable time
(i.e. when a suitable connection is available). In this case the
availability could be signaled via, for example, SMS while the
updated information could be downloaded via SMS or an alternative
bearer such as, for example, general packet radio service (GPRS),
enhanced data rates for GSM evolution (EDGE), high-speed packet
access (HSPA), or CDMA2000 EV-DO.
[0111] The tasking information can be enhanced further to take into
account the quality of the information that could be reported. The
usefulness of a particular observation pertaining to a cell depends
partially on the observed signal power. It also may depend on the
number and type of existing measurements held by the server for
that cell. If a previously unknown cell is observed with a
relatively weak signal level (e.g., -90 dBm) and a GPS fix is
available, this conveys relatively weak information on the location
of the cell. In contrast, an observation at a signal level of -50
dBm would indicate that the device is currently situated very close
to the cell and a contemporaneous GPS fix would convey relatively
precise information on the location of the cell. As another
example, a measurement corresponding to a cell for which the server
already has 100 prior measurements would offer considerably less
value than a measurement corresponding to a cell for which only one
measurement has been received previously. Therefore a further
enhancement to the tasking information is to represent the current
state of each cell in the network database using more than one bit,
reflecting the quality of information held in the current version
of the network database. As an example consider the case where 2
bits are used.
[0112] An example of how the values could indicate the quality of
the information currently held by the network database server as
indicated in Table 1 below. When a client hears a particular cell
ID, it can consult the tasking table and determine whether to
report the cell based on the quality of the current
observation.
TABLE-US-00001 TABLE 1 use of multiple bit cell quality information
Binary value Interpretation Action 00 cell is unknown Report if
measured 01 coarse location information to Report if measured date,
few measurements with rxLev >-90 dBm 10 moderate accuracy
information Report if measured held, moderate number of with rxLev
>-70 dBm measurements 11 good quality information held in Only
report if measured the server with rxLev >-55 dBm
[0113] In this way, during the initial period of service, clients
will report cells more frequently and as the network model
converges, the reporting rate will slow automatically, except for
unknown cells. The actual threshold rxLevs corresponding to each
quality setting could be transmitted with the tasking information
to provide more dynamic control over the reporting.
[0114] As disclosed herein, either 1 or 2 bits to encode the
current state of each BTS in the network may be used in certain
embodiments. However, other bit sizes may also be used. More
generally, any acceptable number of bits may be used to convey the
quality information. The reporting thresholds could be implied,
dividing the typical signal reception level from -50 dBm to -105
dBm into fixed size steps. Alternatively the thresholds could be
explicitly encoded and transmitted to the clients. FIG. 3
illustrates exemplary logic which can be used by a client to
determine whether a report should be generated for a cell if a GPS
position fix is available. The client first determines whether the
cell ID is known to the server in step 200 based on the tasking
information. If not, the client composes a report in step 210. If
it is known, the client determines if the rxLev is greater than a
quality based threshold in step 220. If the rxLev is greater than
the quality based threshold, then the client composes the
report.
[0115] The tasking information for a GSM or UMTS network can be
encoded in 8 KB if one bit is used per cell to encode the current
cell state in the network model. For clients deployed on terminals
that are used for mobile web browsing, etc., this represents a
relatively small data download. Nevertheless it will often be
important not to disrupt the user's activities with a download like
this. Preferably the client will wait for a suitable opportunity
for performing the download in order to minimize the degree to
which the user is affected.
[0116] Optionally the tasking information can be conveyed in
smaller chunks. For instance the 8 KB could be transmitted at
different times in 1 KB blocks, accompanied by a header which
specifies the starting cell ID for that block.
[0117] In normal operation, the tasking information would be
updated from to time as the network model evolves. In this case it
may be more efficient to update in smaller blocks, perhaps only for
those blocks where there have been significant changes. Typically,
there is no lower limit on the size of the block, even an update
for a single cell can be transmitted in a single update.
[0118] Certain embodiments incorporate a further enhancement to the
measurement collection to further accelerate the acquisition
process. Because of the relatively low frequency of GPS usage by a
typical user in a familiar context, the availability of GPS based
cell location reports may be too infrequent for rapid
initialization and also to detect changes in the network
configuration.
[0119] In order to accelerate the rate of information collection,
the client can also collect and report patterns of measurements
which reflect spatial proximity between two or more transmitters.
Following the same tasking pattern as describe above, when
observing a new serving cell, the client can consult the tasking
list to determine whether to prepare a report for transmission (in
certain aspects the report can be saved for later transmission
piggybacked on a network connection established by the user).
[0120] Consider first the simple case where the current state for
each cell is represented with a single bit (known/unknown). When
the client detects a new serving cell, it checks whether a report
is warranted. This could be the case if one or the other of the
current or previous serving cells is known and the other is
unknown. In this case a report is sent to the server conveying the
information concerning the unknown cell situated nearby the known
one enables the server to compute an initial coarse model for the
unknown cell. The conditions for reporting can naturally be
extended beyond immediate adjacency. If 2 consecutive serving cell
observations corresponded to unknown cells and the 3.sup.rd cell in
the sequence was a known cell, then a report would still be
warranted. The client would report not only the two most recent
serving cells but also the 3.sup.rd most recent as it was unknown
as well.
[0121] Preferably the client also reports the time interval between
the observations of the different serving cells enabling the server
to make an allowance for possible terminal movement.
[0122] The reporting conditions could be specified more tightly by
using more than one bit to represent the quality of the cell
information in the network database server. FIG. 4 illustrates
exemplary logic used to determine whether to generate a report for
one cell using another cell reported at a similar time as a
location reference. In the following description, when referring to
information that is known to the network database server, this can
be, for example, based on the tasking information stored in the
client. The client first determines if the device is a cell ID only
device in step 400. If it is a cell ID only device, then the client
determines if the cell ID is known to the network database server
in step 410, and if so it determines if the current quality known
to the network database server for this cell is less than a first
threshold in step 420. If the cell ID is not known to the network
database server, or if the current quality known to the network
database server for this cell is less than the first threshold,
then the client determines if the cell ID quality known to the
network database server is greater than a second threshold in step
430. If so, then the client composes a report. Referring now to the
case where the device is not a cell ID only device (e.g., the
device is also capable of determining rxLev), the client determines
if the cell ID is known to the network database server in step 450.
If so, then the client determines if the rxLev is greater than a
quality based threshold in step 460. If the cell ID is not known to
the network database server or the rxLev is greater than a quality
based threshold, then the client determines if the reference cell
ID rxLev is greater than a third threshold in step 470. If the
reference cell ID rxLev is greater than the third threshold, then
the client determines if the reference cell ID quality known to the
network database is greater than a fourth threshold in step 480. If
so, then the client composes a report. In certain mobile terminals
the client may not be able to obtain useful rxLev information
either because the terminal does not support a corresponding API or
because the terminal has a defective implementation which renders
the reported rxLevs substantially useless. In this case the client
may simply issue a report if the current quality is below a
threshold and not issue a report otherwise.
[0123] A similar mechanism may be employed to report information
concerning the location of Wi-Fi Access Points with respect to GPS.
In certain embodiments, the client software monitors for the
availability of GPS and Wi-Fi measurements and records any
detection in the measurement buffer with the associated timestamp.
In some cases, depending on the capabilities of the wireless device
and the facilities supported by the platform, the client may
initiate an active scan for APs. This may be done by issuing a
request for the 802.11 hardware to transmit a probe request to all
APs in range. For Wi-Fi APs the information recorded includes both
the MAC address as well as the signal strength. Note that the
measurements may not contemporaneous. If a GPS measurement is
available at one instant and then a few seconds or minutes later
one or more Wi-Fi APs are detected, the client will retain both
sets of measurements and report them if the other reporting
criteria are met. The network database server processing takes
account of the effect of any time interval between the GPS and
Wi-Fi measurements by increasing the associated uncertainty
appropriately.
[0124] In order to avoid repeatedly transmitting reports concerning
the same AP, the client keeps a list of the MAC addresses for APs
which it has reported in a recent report list (RRL). The length of
this list depends on factors such as the available memory and the
expected density of APs in the areas where the client will be
deployed. Suitable RRL lengths may range from 10 to 10000, 50 to
4000, 80 to 200, 50 to 500, or 80 to 400 (100 may be a suitable
value in certain typical implementations). For each MAC address in
the list, one or more of the time last detected, the best quality
of any detection for that AP as and the number of times the AP has
been observed since it was added to the list may be retained. When
the RRL is full and a new AP is detected, the least important entry
may be purged and replaced with the latest AP. The order of
importance is determined based on the number of times the AP has
been detected and a combination of the time since it was last
detected and the best detection quality. This is advantageous
because there will typically be a set of APs situated in locations
that are frequently visited by the user. Retaining these APs in the
RRL is likely to have the greatest benefit in minimizing the
transmission of duplicate reports for the same AP. Compared to some
existing systems in which user devices are merely located based on
Wi-Fi measurements they report, the recently reported list
disclosed here provides sufficient protection against continual
reports of the same cells to enable the acquisition of information
about APs to also be implemented in user devices. An exemplary
processing flow for deciding whether to generate a report for a
Wi-Fi AP when a measurement is available and one or more GPS
measurements are also available is shown in FIG. 5.
[0125] In some cases it may be advantageous for a client to report
measurements for one or more Wi-Fi APs using another Wi-Fi AP which
has already been measured to a sufficient degree of accuracy as a
location reference. FIG. 8 illustrates exemplary decision logic for
this scenario
[0126] Since the coverage area of a Wi-Fi AP is typically limited
to a few tens or hundreds of meters, contemporaneous measurements
of a known Wi-Fi AP and an unknown cell can yield useful
information for the server on the location of the cell. Furthermore
in everyday usage there are likely to be more devices in a given
area with Wi-Fi & cellular enabled than GPS and cellular.
Therefore having obtained a GPS location fix for a Wi-Fi AP, it is
possible to leverage the typically smaller footprint of the Wi-Fi
AP to provide measurements for cellular base stations. A further
advantage of having location information for Wi-Fi APs is that dual
mode devices from different cellular networks are likely to share
the AP, thereby enabling location information concerning a single
AP to be leveraged to refine the estimate for cells from all
surrounding cellular networks.
[0127] The tasking framework described before can be applied here
as illustrated in Table 2 for the case where 2 bits are used to
encode the quality per cell. If a Wi-Fi AP can be detected strongly
then the client checks the serving cell. Depending on the strength
of the serving cell measurement and the state of the cell in the
tasking information, a report may be prepared for transmission.
TABLE-US-00002 TABLE 2 Binary value Interpretation Action 00 cell
is unknown Report if measured 01 coarse location information to
Report if measured date with rxLev >-90 dBm 10 moderate accuracy
information Report if measured held with rxLev >-70 dBm 11 good
quality information held Report if measured with rxLev >-55
dBm
[0128] Advantageously, certain embodiments of the present
disclosure enable the client to focus the reporting of cellular
measurements using Wi-Fi APs as a reference to only those cases
where the Wi-Fi AP is known to the network database server (thereby
yielding a reference measurement). With cellular base stations, the
use of the tasking information described above enables the client
to determine which cells have good quality information associated
with them in the database. The relatively much larger number of
Wi-Fi APs in existence compared to cells and correspondingly larger
MAC address range however makes it difficult in practical terms to
distribute the corresponding tasking information about known and
unknown MAC address (2 48 combinations). A local version of the
tasking information is available however in the Wi-Fi history list.
Since the client maintains the list of Wi-Fi APs that it has
reported to the network database server and the best quality
reported, it can determine whether one or more of the current APs
are known to the server (based on a previous report from this
client) and whether it is likely to be known with sufficient
quality to serve as a location reference for an unknown cellular
BTS.
[0129] FIG. 6 illustrates exemplary logic by which a client can
determine whether to generate a report for a cell using a Wi-Fi
measurement to provide a location reference.
[0130] The reverse reporting mechanism may also be applied in some
cases, i.e. a measurement for one or more Wi-Fi APs is reported to
the server using contemporaneous cellular measurements as the
reference source. If the network database server holds a relatively
accurate model for a cell and the terminal while camped on that
cell can also hear a Wi-Fi AP, then the client can report the pair
of measurements contributing to the network database server
learning for that AP.
[0131] In this case the client can be programmed to generate a
report only in the case that the terminal is sufficiently close to
the cellular base station to constrain the location of the Wi-Fi AP
sufficiently (i.e. the cellular rxLev exceeds a threshold). A
suitable threshold rxLev for the cell could be, for example, -60
dBm, although any other suitable value could be used such as
between -50 dBm and -80 dBm. The actual value could be defined in
the tasking information sent to clients to enable it to be adjusted
dynamically as the network model converges.
[0132] In many cases, if the terminal is connected to the Wi-Fi
rather than cellular network, the cost of transmitting a
measurement report is likely to be lower than the equivalent cost
for cellular data so the threshold might be adjusted to enable
greater reporting rates than over cellular, for example by lowering
the threshold to -90 dBm. FIG. 7 illustrates an exemplary
processing flow by which the client can determine whether to
generate a report for one or more Wi-Fi APs if contemporaneous
cellular measurements are available.
[0133] From time to time the client application may collect
measurements pertaining to one or more wireless networks, GPS, or
any suitable combination thereof. The specific types of
measurements will vary from device to device according to the
capabilities of the device, the surrounding environment and also
user configuration settings for the device. In certain embodiments,
the client collects and stores any measurements in a time tagged
format. When a report is to be encoded for transmission to the
server the measurements at that time are analyzed and one or more
is encoded depending on the relative priorities of the measurements
as well as the capacity available in the reporting channel. For
example if a single binary SMS is to be used, then the space
available is 140 octets.
[0134] The information encoded in the report is presented using BNF
notation where + indicates accumulation, { } indicates iteration of
the item, ( ) indicates that the item is optional, and | indicates
a selection: [0135]
Report:=MCC+MNC+0{CellTableEntry}+0{ShortIdTableEntry}+0{WifiTableEntry}+-
0{MeasurementTableEntry}+0{GPSTableEntry} [0136]
CellTableEntry:=LAC+CID [0137]
ShortldTableEntry:=ARFCN|UARFCN+BSIC|PSC [0138]
WifiTableEntry:=MACAddress+Age+(rxLevel)+(SSID) [0139]
MeasurementTableEntry:=TableIndex+Age+MeasurementType+(rxLevel)+(Timingad-
vance) [0140] MeasurementType:=CellTable|ShortIDTable [0141]
GPSTableEntry:=Latitude+Longitude+Accuracy+Age
[0142] The age refers to the time between a measurement and the
most recent measurement. The age may be defined as a number of
seconds. Alternatively the age may be defined as a number of ticks
where a tick refers to a known constant, or substantially constant,
amount of time known to the relevant elements of the system that
need to process the age information. As an example the interval
could be known because all clients use a fixed sample interval.
[0143] The structure of the tables means that for a given report
any of the tables may have measurements or may be left empty
according to the information currently held in the measurement
table. The encoding format means that GPS measurements may be
included as opportunity affords. Advantageously, this minimizes
battery depletion by not powering on the GPS solely for network
acquisition but rather capitalizing on opportunities where the user
or another application has activated the GPS.
[0144] In certain embodiments the client selects information to be
reported according to the value that is likely to accrue from the
measurement in the context. As an example, if the encoding is being
done for a network acquisition report, then the objective could be
to report as much information as possible that is not known to the
server, along with information that conveys reference position
information. Consider a case where a range of cells have been
measured in recent cycles, some of which are known and some of
which are unknown to the server (determined by the client by
examining the tasking information). Assume further that a GPS fix
has been recorded recently. The encoding would start with the most
recent GPS fix and then in decreasing order of signal strength (if
available), the unknown cells, followed by the known cells, space
permitting. The reason for encoding unknown cells in reverse order
of their signal level is because the stronger the signal level, the
tighter the constraint on the location of that cell relative to a
reference position (GPS in this case). In the case where the
particular device does not support rxLev measurements, the unknown
cells would be reported in increasing order of the time interval
between their measurement and the GPS measurement time. Wi-Fi
measurements typically would be encoded (space permitting) after at
least one measurement per cellular base station has been
encoded.
[0145] In certain embodiments a device may report measurements to
support a location calculation, where the measurements correspond
to one or more cells which are currently unknown. In such a case,
the lack of any information in the database concerning those cells
would mean that no location can be calculated. From a user's
perspective this is a very negative experience and one that
warrants significant effort on the part of the location service
provider to avoid or mitigate. In some current systems when this
happens there is simply nothing for the user to do other than move
about while sending repeated requests hoping that the device may
reselect to another cell that is known to the server.
[0146] Advantageously, in certain embodiments the application
client can provide a mechanism to work around this. Consider first
the case where the cell ID is unknown to the system, but the LAC is
known, being associated with several known cells. The location
server can return a failure indication to the application server,
but can also return a location estimate which corresponds to the
centroid of the cells which are currently associated with that LAC.
The location uncertainty indicated to the user would also be
relatively large, reflecting the footprint of the controller
associated with that LAC value. The system can then present a
coarse map covering the LAC area and offer the user two options.
The first is to use the current displayed location and area to
complete the original request. In some cases, this coarse accuracy
may suffice; as an example consider a search for a business having
a relatively uncommon name. A search in an area covering a few
kilometers is likely to still resolve to the correct location. The
second alternative for the user is to use the map to try to locate
themselves by zooming in and using nearby features such as street
names etc to determine their actual location. The user can then
specify their location manually to the application via the cursor
or any other suitable UI facility. The location specified by the
user can then be translated from a screen position to a
corresponding physical location by the application and used to
update the location. The application can also provide an
uncertainty value corresponding to that position measurement based
on the current map zoom level. A suitable uncertainty value may be
the extent of the displayed map divided by a factor between 2 and
10. The system can then use the location specified by the user in
completing the original request, this can be initiated for instance
by the application refreshing the request to the server, including
the location specified by the user.
[0147] In the case that the user does provide a manual input of
their current location, aside from satisfying the users service
request, the system can also exploit the user specified position as
a reference measurement for the currently unknown cell. This
measurement can be sent to the gateway as with other network
reports. In this case the measurement type in the GPS table
indicates that the source of the measurement was human input rather
than a GPS device. The benefit of marking manual measurements as
such is that with human input there is a greater possibility of
gross error, either accidental or mischievous. If subsequent client
measurements corresponding to the cell are found to be inconsistent
with the user supplied measurement, the exemplary process
illustrated in FIG. 9 when a cell has been detected to have moved,
can be initiated more quickly to eliminate the error.
[0148] Depending on the latency between receipt of measurement
reports and network update processing, it is possible that the same
user could make further requests for location based services from
the system before the previous manual measurement has been used to
add the cell to the network database. In certain embodiments, a
further enhancement in the user device is to retain the user
specified location locally in memory as long as the device remains
camped on any unknown cell(s) as determined from the tasking
information. If further location requests are transmitted in this
case, the client will include in the GPS table a manual entry
reflecting the location previously specified by the user. In this
way in a series of requests the user will not be repeatedly
confronted with an unknown location response and forced to
re-specify the current location. Advantageously, having the client
retain the earlier manual location indication from the user enables
the subsequent service requests to be satisfied without requiring
the user's location to be persisted in the application server, a
problem from the point of view of managing server scaling and
failover.
[0149] In some cases, in addition to the cell ID being unknown, one
or more of the LAC, MNC & MCC may also be currently missing
from the network database server(s). In such cases the same
mechanism can be used to resolve the user location except that the
initial map presented is at an even coarser zoom level. The system
can use the worldwide definitions for MCC and present a map of the
corresponding country, enabling the user to zoom in to their
current location and preferably supply a manual location
indication. Overall, although the user may have to carry out some
intermediate steps this enhancement enables the location system and
associated content or application server to successfully answer
virtually all user requests which will minimize the rate of user
disappointment.
[0150] In some cases it may be possible to use existing information
to initialize the network model, without waiting for sufficient
measurements to be reported by devices. The benefit of this is that
it can unlock the circular dependency wherein users of the service
are needed to provide measurements for the network learning
process, but the service is poor or unusable until sufficient users
are active and providing measurements. Having a way to initialize
the network database can enable a useable service to be offered
initially, making the task of signing up users easier. The
following paragraphs describe several methods for initializing the
cellular network database.
[0151] In certain embodiments, initially when there is no
information available on a target cellular wireless network an
empty database is created. The MCC & MNC for the corresponding
network are initialized with the appropriate values for the network
which can be determined from public listings of this information.
Such information is available on many internet sites including the
official sites maintained by the mobile telecoms industry
associations (GSMA & CDG etc.). For models corresponding to
fixed wireless networks such as the collection of Wi-Fi APs, the
MCC & MCC are set to a system constant which effectively means
N/A since in these cases the networks are operating in unlicensed
spectrum and there is not necessarily any coordination between the
different APs. After initializing a database in this way,
subscribers will not be able to use the system for location
estimation until further information is acquired describing the
configuration of the radio network. Therefore the aim is to
accelerate the acquisition process as much as possible. The tasking
information corresponding to an empty cellular network model would
be initialized and transmitted to all available clients as early as
possible. For GSM & UMTS networks, the initial tasking
information would mark all cell ID values as unknown initially.
[0152] In some cases a MNO may be willing to make available a
one-time extract of the radio network configuration information.
This might be for instance in exchange for some consideration, but
without having to commit to the overhead of maintaining the
information and providing frequent updates.
[0153] In this case the database could be initialized using the
information provided by the operator. In some cases one or more
parameters of the models used by the network database server might
not be available from the operator's extract and would therefore be
initialized to default values. For example the covariance estimates
corresponding to location uncertainty for each cell would be set to
a value reflecting the estimated accuracy of the site information.
A suitable value might for instance be [50 2, 0; 0, 50 2]. In some
cases the information might be available in a format that is not
directly compatible with the import format supported by the network
database server. In this case, a custom script could be employed to
reformat the data for loading by the corresponding network database
server. In such cases, attributes such as environment type would
also be estimated by the network database server and used to
complete the network model initialization.
[0154] In some cases, to accelerate the initialization process a
focused measurement campaign could be initiated using either
general purpose or specialized measurement receivers together with
GPS. For instance prior to service launch one or more vehicles
could be equipped with GPS receivers and cellular terminals and
driven around target areas to collect measurements. One suitable
device could be a standard mobile phone which features cell, Wi-Fi
and/or GPS capabilities. An application running on the phone can
initialize the GPS and then collect cell, Wi-Fi and/or GPS
measurements as the user moves around the target coverage area. In
some cases, however, it may be desirable to collect measurements
for all, or substantially all, cellular networks as well as Wi-Fi.
One option may be to simply carry additional devices, one for each
network and access technology. In other words for carriers
operating both GSM and UMTS networks, two devices would typically
be required. Yet another alternative is to use a dedicated device
with the capability to monitor multiple networks simultaneously.
FIG. 38 illustrates such an exemplary device. It is equipped with
multiple cellular modems each providing a host interface through
which the controller can read the current access technology,
current serving cell information including signal strength plus
neighbor cell information and signal strength. In addition a Wi-Fi
network adapter is provided through which the controller can obtain
a list of the current Wi-Fi Access Points and associated signal
levels. The device also features a GPS from which the controller
can read location information. When activated the device
periodically scans all the cell, Wi-Fi and/or GPS devices and
records the reported information in a log file. The system provides
a facility to upload such log files, whereupon the log file is
parsed and individual reference measurements are generated and
added to the network learning collection using the GPS as a
reference source.
[0155] In other cases an airborne approach might be preferred for
collecting cellular network information. In this case a cellular
band antenna could be affixed to the fuselage of a light aircraft
and cellular terminals installed within the cockpit together with a
GPS. One suitable flight plan would be a series of lanes spaced at
one half or one third of the expected cell site separation in the
area. In most populated areas local authorities impose a ceiling
above which aircraft must fly. This ceiling is typically a few
hundred meters. From the perspective of these measurements, the
lower the better, otherwise the vertical directivity of outdoor
cellular antennas will mean that the signal level at elevation will
be greater for remote cells rather than those immediately below the
aircraft. Typically such a flight would collect information only
for macro cells. The path loss for indoor and microcells at a few
hundred meters above ground level would typically be too great for
these cells to be measured.
[0156] The measurement yield in such a campaign could be enhanced
by using special purpose receivers which attempt to decode and
report all cells in the area. In GSM for instance this would mean
that the receiver attempts to identify all the BCCH channels and
then repeatedly tries to decode the cell ID for each. This is in
contrast to the normal operation of a GSM terminal in idle mode
where cell ID reporting is typically confined to the currently
selected serving cell.
[0157] The measurements collected during such a flight could be
processed to estimate the locations of each cell by an algorithm
that seeks the parameters for a BTS model which best fit the
pattern of received measurements. One such algorithm employs a cost
function which for each measurement computes a predicted signal
level s at the location where the measurement was recorded using a
particular set of BTS parameters (x, y, height, azimuth etc.) and
compares this with the actual received level s. The cost C per
measurement is given by:
C = ( s - s ^ ) 2 .sigma. 2 ##EQU00001##
where .sigma. corresponds to the expected variation in the signal
level to fading and multi-path. For an aerial survey, shadow fading
is likely to be small compared to multi-path, therefore suitable
values of sigma are likely to be between 3 and 6 dB. The algorithm
repeats the cost calculation over a range of values of the
variables and selects the set yielding the smallest total cost.
[0158] In some countries, radio spectrum management authorities
maintain a database of registered transmitter locations. This
information could be utilized (subject to the terms of use
published by the corresponding publisher) to accelerate the
convergence and improve the accuracy of the locations more quickly.
Such databases typically provide details such as location,
frequency band, antenna type & orientation & max transmit
power. An example of such a database can be found at
http://web.acma.gov.au/pls/radcom.
[0159] In certain embodiments the database can be filtered to
extract the information pertaining to the specific network of
interest (both by operator and by frequency band) to obtain a list
of candidate locations. As measurements pertaining to specific
cells are accumulated by the network database server, the degree of
consistency between the measurements and nearby candidate locations
can be assessed. When the consistency is sufficient and the
statistical significance reaches a suitable threshold, a match can
be declared and the more accurate location assigned.
[0160] One method for performing a calculation to determine whether
a candidate location from such a database is a suitable match for a
set of measurements accumulated by the server is using the cost
function described below. To check the consistency, certain
embodiments first initialize a BTS model with the location and
antenna orientation from the database and then calculate the sum of
the residuals between the measurements and predictions using the
BTS model. If the sum of the residuals is smaller than the
threshold then the location is accepted as a good fit and the x and
y parameters for that BTS model are "snapped" to the location
obtained from the database.
[0161] In some cases, the public database may not include
information on the antenna orientation. The process above can still
be applied by first using a solver to find the model angle for
which the residual is smallest and then proceeding with the
consistency check.
[0162] The accumulated information reported by one or more devices
is combined by the system into a model which can then be used by a
location server to estimate the position of users.
[0163] While certain embodiments can be used to accumulate
information concerning a single network to support location of
wireless terminals using that network, it may also be used to
accumulate information pertaining to a large number of networks. It
is preferable therefore to partition the work in some way. One
benefit of this is that it enables the processing to be distributed
among multiple processor elements. Additionally, when the
information is used for location, if the totality of the
information is partitioned appropriately then when the location
server wishes to use the information pertaining to a few cells to
estimate the location of a specific device, the complexity of the
queries to find the required information as well as the volume of
information to be transferred can be minimized. One way to
partition the problem is by separating the models for different
networks. In the case of cellular networks, this can be done based
on the MCC & MNC which is reported by the client software along
with any measurements. Within each network it may also in some
cases be advantageous to separate the network into sub-networks
containing 2G & 3G cells respectively.
[0164] For fixed or mobile wireless networks operating in
unlicensed spectrum (for instance Wi-Fi) there is no equivalent
grouping of Access Points and therefore the system treats all such
elements individually. If it is necessary to partition the ad-hoc
collection of Wi-Fi APs for instance to provide to a Location
Server serving subscribers in a particular country, then the
partitioning can be done based on location. In other words, a sub
network containing all APs within some physical region can be
extracted. In some cases, a location server may be configured to
serve subscribers of a specific cellular network only. In such
cases, a suitable sub-network model for the unlicensed network can
be extracted by selecting all elements that fall within the
physical footprint of the cellular network.
[0165] A related aspect of the partitioning of the models is that
the storage for the reports can be segmented by network. This may
simply mean that the measurements are stored in the same database
in tables maintained by different processes or alternatively in
cases where a very large number of users are being supported across
a very large number of networks, separate databases may be used.
The benefit of the separation between networks is that it affords
considerable flexibility & scalability in operating the system
and managing computing & storage resources for it. A further
potential advantage that arises from separating the processing for
different networks is that in some regions the network database and
location servers may be able to be supplied with terrain or clutter
information which can be incorporated in the propagation modeling
process as is well known in the field. By separating the processes
it may be more convenient to provide the information when available
without interrupting the processing for those networks where no
such information is available or creating a dependency in all
network database processes on having such information.
[0166] FIG. 2 illustrates an exemplary lifecycle for a network
model according to certain embodiments. At an appropriate time the
model for the network is initialized in step 100. This could be
done when the system is initially configured and empty models are
created for all networks that are to be modeled. Alternatively this
initialization could be triggered when the first measurement
pertaining to that particular network is received by the network
database server. After initialization, the model goes through
repeated cycles of measurement collection in step 110, model update
in step 120, and tasking information update in step 130. While the
process is shown as sequential, in practice there may be some
temporal overlap between the collection and update processes. This
is because once an update has been initiated, other measurements
may continue to arrive for processing. Such measurements will be
collected and stored for the next update processing cycle. The
trigger for commencing an update can vary depending on the needs of
the applications. In certain embodiments, the update may be
triggered periodically, for example every hour, every two hours, or
once a day. Depending on the configuration of the system and
processing resources it may also e preferable to stagger the
updates across different network models so that the updating
proceeds with one network model at a time. An alternative basis for
triggering the update processing may be based on the number of
measurements that have been collected. This could for instance be
triggered once a suitable number of new measurements have been
received, for example, at least 50, 100, 200, 500, 1,000, 10,000,
or 15,000. Yet other criteria could be when new measurements have
been received which pertain to a proportion of the cells in the
network. For instance if new measurements are available for, for
example, 2%, 5%, 10%, 15%, 20% or other suitable number of the
cells in the network then the update processing could be initiated.
Yet another option for triggering the processing could be to allow
it to happen as fast as the data & processing resources permit.
In other words once a particular update cycle completes, the next
one starts, using as input all the measurements that were
accumulated while the previous cycle was being completed. The merit
of this approach is that the accuracy & coverage enhancements
enabled by recent measurements will be available for users as soon
as the available processing resources permit.
[0167] As illustrated in FIG. 2, once the update processing has
been completed, the tasking information may be updated to reflect
the state of information in the network model for use by clients to
guide their network reporting as well as measurement reporting for
location calculation. In certain embodiments, the tasking
information has an associated revision number that corresponds to
the database revision from which it originated. The network
database server may also retain previous versions of the tasking
information. Given that the database revision is likely to be
updated relatively frequently, in some cases it will be preferable
not to distribute every update to every client. Therefore it may be
preferable for each client to record the version of the tasking
information which it currently holds. At suitable times the client
may query the network database server to determine the number of
changes between the revision it holds and the latest revision of
the tasking information. The client can then initiate a download of
the updated information as required.
[0168] The quality of the model will improve with greater
availability of measurements. FIG. 9 illustrates an exemplary
lifecycle for each element of a network model, corresponding to a
BTS or other wireless access point or transmitter.
[0169] The following description will be presented using the
example of a cellular network BTS however the same lifecycle is
equally applicable to other types of wireless access points or
transmitters. When a network model is first created for the
corresponding network, there will be no cells in the model. If
information is provided to initialize the network model, the cell
may be initialized with that information. Examples of such sources
of information are included in a later section. Typically the
information provided will include the location of the cell, its
identifiers (e.g., cell ID & LAC), the frequency band and the
type of antenna. In some cases information may also be provided on
the antenna orientation & height.
[0170] In the absence of external initialization information, the
cell model may be initialized on receipt of the first measurement
pertaining to that cell. In this case the identification parameters
can be initialized before the measurements are used in the normal
update processing to set the other parameters such as x and y.
[0171] In certain exemplary embodiments the model for each cell
includes some or all of the parameters shown in Table 3 below. Some
of the parameters may be left uninitialized or may be assigned
default values in some cases depending on the type of transmitter
being modeled.
TABLE-US-00003 TABLE 3 Parameter Description Type/Units Active
FALSE if the corresponding cell has Boolean/N/A been halted
(transmission disabled). TRUE if the cell is active. antennaAngle
The direction in which the antenna Real/degrees clockwise boresight
is oriented. For non directional wrt North antennas, defaults to
zero antennaBeamwidth The 3 dB beamwidth of the antenna. Set
Real/degrees to 180 degrees for omnidirectional antennas.
antennaGain Gain along the boresight relative to an Real/dBi
isotropic radiator. antennaType Selection from one of the available
N/A antenna models in the system antennaTilt Vertical tilt of the
antenna Real/degrees ARFCN Overloaded field. N/A GSM: BCCH ARFCN
UMTS: UMTS UARFCN CDMA: Channel Number WIMAX: Frequency Wi-Fi: N/A
bcchLev Overloaded field: Integer/dBm GSM: BCCH txLev UMTS: CPICH
txLev CDMA: Pilot Channel Power WIMAX: Transmit power Wi-Fi: N/A
BSIC Overloaded field. GSM: BSIC. N/A UMTS: PSC CDMA: PN offset
WIMAX: N/A Wi-Fi: N/A Height The height of the antenna above ground
Integer/meters level ID Overloaded field. N/A GSM: Cell ID. UMTS:
Cell ID CDMA: BTS ID WIMAX: Base Station ID Wi-Fi: MAC address LAC
Overloaded field: N/A GSM: LAC UMTS: LAC Type Selection from the
following: N/A CELLULAR MACRO CELLULAR MICRO CELLULAR PICO Wi-Fi AP
WIMAX X The X location of the cell expressed in Integer/meters grid
coordinates Y The Y location of the cell expressed in
Integer/meters grid coordinates Covariance Covariance matrix (i.e.
accuracy) associated with the estimated X & Y coordinates
Latitude Latitude of the cell location Real/degrees Longitude
Longitude of the cell location Real/degrees RepeaterType Selection
from the following: N/A NOT_REPEATER REPEATER_DONOR
REPEATER_SLAVE
[0172] The update processing step 240 is carried out using all the
available measurements at the time of processing. One aspect of the
update processing may be to check whether the collection of
measurements reflect a change in the cell. For instance if a cell
is relocated, then the older measurements will reflect the original
location while the more recent ones will reflect the current
location. A test can be carried out to check this, by calculating
the residuals between predicted values using the current model and
the measured values. By maintaining a history of the residual
values corresponding to each version of the network database, a
change will be reflected in growing residuals. Use of a suitable
threshold on the residuals, for instance based on the chi-square
distribution will enable this type of scenario to be detected. In
the event that the test does not indicate any change then the
parameters of the model are refreshed based on the available
measurements. If there is a change, then further processing can be
done to identify the time at which the change is likely to have
happened and the historical measurements are trimmed, retaining
only the measurements which are consistent with the current
configuration.
[0173] Another aspect of the update processing for each cell may be
to determine the type of cell (Macro/Micro/Pico). Yet another
aspect may be to determine whether the cell antenna is
omni-directional or directional in order to use a suitable gain
pattern in the propagation modeling. Yet another parameter to be
estimated is the location of the cell. If observations of the cell
are found to exhibit an omni-directional pattern, the location of
the cell is simply the mean of the observations. The antenna angle
is set to zero since any value will do. If on the other hand the
cell is determined to exhibit a directional radiation pattern, it
may be advantageous to estimate both the antenna location as well
as its orientation.
[0174] In the initial phases, the data may not provide sufficient
information to determine robustly either the type of the antenna or
indeed the orientation. Therefore a simple, more robust model may
be used which is to estimate the location as the mean of the
reference locations available for a cell. The footprint or coverage
area of the cell can be estimated directly from the spread of these
reference locations and expressed as a 2_D Gaussian with mean X and
covariance C. FIG. 12 exemplary illustrates this using a simulated
network. A collection of observations are simulated, each
consisting of a single cell ID measurement (no rxLev) for the
highlighted sector and an accompanying GPS position fix. The
location for each measurement is marked with a dot. The calculated
mean is marked with a cross. The ellipse in the plot shows the area
represented by the covariance matrix.
[0175] FIG. 13 exemplary illustrates the same network, but this
time showing the measurements that would be reported with handsets
capable of reporting rxLev as well as serving cell ID. The rxLev
for each measurement is reflected in the size of the dot. For the
purpose of the simulation, a propagation model was used and 6 dB
log normal fading was added to the idealized rxLev at each point.
In this case the measurements can be characterized by mean and
covariance of the 3 element observation vector [x,y,rxLev]. Where
the system has accumulated a mix of cell ID only and cell ID+rxLev
measurements, the position mean is estimated across all
measurements while the mean rxLev is calculated only across the
rxLev measurements. Similarly the C11, C12, C21 & C22 are
estimated across the whole set of measurements while Ci3 & C3i
are estimated using only the rxLev measurements.
[0176] In order to maintain a consistent representation of all
cells regardless of their stage of modeling, the mean and
covariance characterizing the footprint can be converted into an
omni-directional BTS model, with the transmit power level being
adjusted to achieve a footprint of equal area to the 90.sup.th
percentile ellipse. The x & y coordinates of the BTS are set to
the mean of the ellipse.
[0177] This omni-directional model will remain in effect until
sufficient data are available to determine whether the transmission
pattern is directive or not. Note that even if a few very accurate
GPS referenced measurements are available for the cell (i.e.
associated with very strong rxLvs which constrain the cell location
very tightly), the omni-directional model will remain in effect
until sufficient data are available to determine the directivity
(or not).
[0178] If sufficient data become available for the directivity
test, a check is done. If the data indicate that the reception
patterns is sufficiently directive then the BTS model is updated
& the antenna type is set to a suitable pattern type. The
antenna orientation is then estimated from the data and used to
initialize the corresponding field in the BTS model. In this case,
the BTS location also has to be determined using a different
approach because the most likely location for the cell no longer
lies at the centre of the measurements.
[0179] In preferred embodiments, estimating the BTS location and
antenna orientation involves the minimization of a penalty
function. Such minimization techniques are well known in the field
of data analysis and statistical estimation. The model parameters
(unknowns) to be estimated are the antenna location x,y &
antenna orientation, theta. In this case, the penalty function is
evaluated over the available measurements, and for each measurement
a predicted signal level is calculated at the associated location,
using a suitable antenna pattern such as that exemplary illustrated
in FIG. 14, which is a smoothed representation of a measured
pattern for an antenna characterized by 16 dBi peak gain and 60
degrees beam width. This type of propagation modeling is well known
to those in the field, being described in texts on the subject such
as S. R. Saunders & A. Aragon-Zavala, Antennas And Propagation
for Wireless Communication Systems: 2.sup.nd Ed, Wiley, 2007. For
any given tuple of x,y and theta, a cost or penalty C is calculated
for each measurement as:
C = ( s - s ^ ) 2 .sigma. 2 ##EQU00002##
[0180] The model above requires the measurements to include rxLev.
In some cases rxLevs may not be available (serving cell ID only
measurements). In this case the penalty function C is of the
form:
C = ( s ' - s ^ ) 2 2 .sigma. 2 . ##EQU00003##
[0181] Where s' is the maximum expected rxLev for which a typical
value is -40 dBm.
[0182] A common technique is to assume that the sum of the costs
calculated in this way across the measurements is a chi-squared
random variable. In this case, one can look up the cumulative
distribution for the corresponding number of degrees of freedom and
choose a consistency threshold. In the present case a suitable
threshold may correspond to the 90.sup.th percentile or some other
percentile between 70 and 99.5. If the sum of the calculated
residuals falls below the chi-square value corresponding to this
threshold for the number of degrees of freedom, then the model may
be considered an acceptable fit to the data, otherwise the model
may not be accepted as sufficiently consistent.
[0183] The BTS model used in the system can include a radio
frequency channel parameter such as a (Universal) Absolute Radio
Frequency Channel Number ((U)ARFCN). This reflects the channel used
for the broadcast or beacon channel in the downlink. In preferred
embodiments, for each cell being modeled in the system, this
frequency parameter can be determined from the reported
measurements. This enables a suitable propagation model to be
employed in connection with this cell applied (there are
significant differences in the attenuation of signal power as a
function of range between the 450, 900, 1800, 1900 & 2100 MHZ
bands typically used by cellular networks). The ARFCN field of the
BTS model is initially set to a system constant signifying
UNKNOWN.
[0184] Some wireless mobile devices provide radio frequency
information. As an example the GetCellTowerInfo function supported
by the Windows Mobile Radio Interface Layer reports the BCCH of the
serving cell. In a deployment with a mixture of devices where some
report frequency information and some do not, certain embodiments
can exploit the measurements from those devices which do to update
the model with the correct frequency. As soon as a measurement is
received from a device which reports this information (encoded in
the short Id table), the corresponding field in the model is
updated. The benefit of maintaining this information in the network
model is that a more appropriate propagation model can be used,
i.e. the model for path loss versus range can be selected from the
set of characteristics which are typically used in the
industry.
[0185] When modeling the propagation of a cell, the propagation
modeling engine in the present system uses a suitable model for the
frequency. With a cell for which the ARFCN is unknown, a mid-range
propagation model can be used and the greater uncertainty is
reflected in a larger fading sigma value.
[0186] In many cases, cell sites are deployed in a sectored
configuration using directive antennas to enable greater frequency
or code reuse and thereby increase capacity.
[0187] For a system attempting to identify and localize all the
cells in the network this typical clustering can be leveraged to
accelerate the acquisition process and/or enhance the accuracy of
the results.
[0188] The most common configuration consists of 3 sectors spaced
approximately 120 degrees apart.
[0189] Another common pattern in cellular network configuration is
to orient the three sectors at angles (with respect to North) of
approximately 0 degrees, 120 degrees & 240 degrees.
[0190] In some cases an operator may relocate a cell. Compared to
the other types of changes which are carried out in the evolution
of a wireless network (commissioning new sites, returning etc.)
this is a relatively rare occurrence. Nevertheless, if a cell is
moved and the network database server fails to detect this and
update the database, subscribers making location requests near
these cells will start to receive spurious answers. This in turn
could damage the reputation of the service provider and decrease
the commercial potential of the service in supporting some value
added services.
[0191] Therefore as measurements corresponding to a particular cell
are received and processed, in addition to updating the current
estimate, the system will also maintain integrity information for
the cell. This can be collected in the form of residuals which are
accumulated. In the case of a relocated cell, the residuals will
tend to grow over time. By application of a suitable threshold, the
system can detect that a change has occurred and simply restart the
acquisition process for that cell, using the most recent
measurements to initialize the process (rather than eliminating
everything and having a period where there is no coverage for that
cell).
[0192] In some networks, multiple transmission points may be
deployed for a single cell. Typically one transmission point is
referred to as the donor and the one or more others as repeaters.
In some cases the repeaters may provide very limited coverage
areas, for instance indoor coverage in the shadow of the donor
macro cell. In other cases, the repeaters may themselves serve a
large physical area.
[0193] If measurements are received first for one transmission
point and then another, it may appear that the cell has moved (as
disclosed above). However if the reality is that there are multiple
transmission points then it would be advantageous for the system to
be able to resolve this and update the model to reflect the
distributed footprint associated with that cell identifier.
[0194] Certain embodiments of the present disclosure resolve this
by employing a sufficiently large accumulator threshold on the
residuals before declaring a moved cell and restarting the
acquisition process for that cell. The integrity monitoring will
also include the capability to detect multimodal concentrations of
reports that would distinguish a donor/repeater scenario from a
moved cell scenario.
[0195] In some cases, rather than a donor/repeater arrangement, the
network may be configured with multiple transmission points of
equal coverage in an arrangement sometimes referred to as a split
cell. The database server update processing can detect this because
the requirement to accumulate statistically significant sequence of
measurements before declaring a moved cell will be unlikely to be
met as measurements corresponding to each transmission point will
be received over the update period.
[0196] Another common change in cellular wireless networks which
certain embodiments of the present disclosure may be able to
accommodate is a change of one or more of the parameters used to
identify the cell. For example, this could be a change of cell ID
in GSM or a change of LAC. The latter is more common as one or more
cells are re-parented (i.e. associated with a different BSC).
Certain embodiments can identify this as a change of ID (cell ID or
LAC) based on the fact that one cell in the network will stop
receiving any measurements and a new one will start. The server
will periodically scan for cells that have not been updated for
some time and for each such cell, initiate a search for a recently
activated cell entry having radio measurements which are consistent
with those for the old cell. The time sequence for the matching
entries will typically show a clean break around the time that the
change was applied in the network. The process for confirming the
alignment of the two sets of measurements can be done using any one
of the well known tests to confirm that two sets of samples come
from the same population.
[0197] With Wi-Fi networks a similar scenario could occur if the
user replaces one Access Point with another either when upgrading
hardware or perhaps following a hardware fault. In this case the
new AP will typically serve the same area however the MAC address
detected by a Wi-Fi capable device nearby would be different.
[0198] The diversity of measurements that can be received for
processing and the use of contemporaneous measurements of different
wireless nodes to estimate the locations of each makes the task of
designing a processing framework difficult. To illustrate, assume
that at some instant TO, the network database contains a model for
2 cells as shown in Table 4 below. This version of the database is
N. Note that the X, Y values stored in the network model are in a
suitable local grid rather than latitude/longitude because grid
coordinates are more suitable for the processing to be done.
TABLE-US-00004 TABLE 4 cell ID LAC X Y C11 C12 C21 C22 1000 11 5034
11903 7500 0 0 7500 2000 11 6213 10090 10000 0 0 10000
[0199] At time T1, a report is received from a user device
containing the following information:
TABLE-US-00005 Header MCC: MNC: 505 1
TABLE-US-00006 Cell Table Num Cells 2 LAC cell ID 11 1000 11
2000
TABLE-US-00007 Short ID table Num short Ids 0 ARFCN BSIC
TABLE-US-00008 Wi-Fi Table Num APs 0 MAC SSID
TABLE-US-00009 Measurement Table Num measurements 2 Table Ind Time
Meas Type rxLev Timing Advance 1 0 CELL 1000 NOT_REPORTED 2 30 CELL
2000 NOT_REPORTED
TABLE-US-00010 GPS table Num fixes Base Lat Base Long Alt 0 0 0 0
Time Lat offset Long Offset Accuracy Type
[0200] In this case successive measurements are reported for cells
1000 & 2000 but there is no GPS or other source of independent
position information. Assuming that the current information (in
dbase version N) for cells 1000 & 2000 reflects a diversity of
sources, it would be preferable to use the information in this
measurement report to update the models for these cells in database
version N+1. There is to some extent a circular dependency.
[0201] Certain embodiments of the present disclosure can overcome
this circular dependency by extracting a separate measurement
report for cell 1000 with reference to cell 2000, using the
location parameters for cell 2000 from the previous database
(version N). Likewise cell 2000 is updated in dbase version N+1 by
creating a separate reference measurement for that cell with
reference to cell 1000, using cell ID 1000's parameters from
database version N.
[0202] In principle there is a risk of creating an unstable
feedback loop between the models for each cell in the pair. However
in practice the models for each cell will also reflect
contributions from other cells, from GPS and Wi-Fi and therefore
should not result in a self fulfilling prophecy.
[0203] The process described above may be completed by the
preprocessor for all measurements in the database. Specifically for
each cell or wireless AP reported in a measurement a new location
report is created in a temporary processing area. The update
processing per cell then consists simply of collecting all the
individual measurements corresponding to that cell and combining
them as described previously.
[0204] FIG. 10 illustrates an exemplary preprocessor. The table on
the left of the figure represents the measurements transmitted by
devices since the last database update and collected for use in the
next update. The preprocessor scans each measurement in turn and
extracts the information in it, creating smaller, individual
reports for each of the cells or transmitters featured in the
report. The table on the right represents these individual cell
measurements extracted and formatted by the preprocessor for use in
the next database update cycle.
[0205] FIG. 11 shows an exemplary flowchart of the steps carried
out by the preprocessor for each measurement report. It iterates
over all the cells featured in the report and for each carries out
the following steps: [0206] (1) Choose the next reported cell in
step 1100. [0207] (2) Obtain a location estimate (and associated
covariance) for the device at the time the measurement of that cell
was recorded in step 1110. This location estimate is preferably
calculated in a way that is designed to avoid feedback. If the
original report included a GPS fix then the location used for the
device is the GPS fix. If no GPS fix was available then the
location estimate is calculated by the location server using all
measurements, except those pertaining to the cell that is being
analyzed in this cycle to avoid a self reinforcing feedback
mechanism. From the estimated device location, a location estimate
and an associated covariance matrix for the cell under analysis are
obtained. [0208] (3) Save a measurement to the database for this
cell, containing the cell details, the estimated location of the
cell and the associated covariance in step 1120. [0209] (4)
Determine if any more cells are reported in the current measurement
report in step 1130.
[0210] For the measurement report presented above, two cell
measurement records would be created as illustrated in Table 5
below.
TABLE-US-00011 TABLE 5 cell ID LAC X Y C11 C12 C21 C22 1000 11 6213
10090 210000 0 0 210000 2000 11 5034 11903 17500 0 0 17500
[0211] For the sake of this simple example we have assumed that the
cell models for both 1000 & 2000 in version N of the dbase are
omni-directional and therefore the location estimate for the device
and in turn the measured cell correspond to the current origin of
the referring cell. The covariance estimate for each cell is
calculated as the combination of the uncertainty in the referring
cell location as well as a term reflecting the uncertainty relating
to the location of the cell with reference to the current device
location.
[0212] Although the example shown above relates to a pair of
cellular measurements, the same process could be applied for a pair
of Wi-Fi APs or a mix of cellular & Wi-Fi nodes or any other
suitable combination of measurements.
[0213] The network learning framework defines the way in which
reference measurements for transmitters are derived from
observations of radio transmitter signals, the way in which the
reference measurements are modelled and the way in which the
transmitter characteristics themselves are modelled, and the way in
which the reference measurements are used to obtain estimates for
the said transmitter characteristics. The transmitter
characteristics can subsequently be used as to support other
applications such estimating the location of a receiver.
[0214] By an observation we mean a set of wholly or substantially
contemporaneous readings from one or more wireless sensors in a
device. Such sensors include cellular radio devices and Wi-Fi
access point network adapter. For some sensors the readings in an
observation may correspond to multiple transmitters. For example,
where multiple access points are in the vicinity of a Wi-Fi access
point network adapter typically returns readings from each such
access point. By contrast a cellular receiver typically only
reports readings from the one transmitter commonly referred to as
the serving cell. An observation may also include a substantially
contemporaneous location estimate. Such a location estimate may
derive from a GPS receiver associated with the device. Where an
observation includes a location estimate it may also include an
indication of the uncertainty of the location estimate. Where an
observation includes a location estimate it may also include the
source of the location information, for example GPS.
[0215] Each reading within the observation will typically include
at a minimum the type of transmitter and an identifier for the
transmitter that was detected. A reading may also include other
attributes of the detected signal. An example of such an attribute
is a signal strength indicator such as the detected signal strength
in dBm. Another example of such an attribute is the SSID of a Wi-Fi
Access Point.
[0216] A device may accumulate a sequence of one or more
observations. This sequence may be augmented with one or more
additional attributes. These additional attributes may be applied
by the device or by a server subsequent to the sequence having been
sent to the server by the device. One such additional attribute is
the type of device that made the observation. Another such
additional attribute the ID of the device that made the
observation. Yet another such additional attribute is a timestamp.
Such a sequence may also have associated with it a location
estimate and an associated uncertainty and optionally the source of
the location estimate. Each observation within the sequence may
have associated with it an indication of the elapsed time since the
immediately preceding observation. Such an indication may be a
measure of elapsed time. Such an indication may be implicit where
the observations are made at a known and constant rate.
[0217] In certain preferred embodiments the observations that
comprise a sequence are recorded at a fixed interval, or
substantially fixed interval. When the sequence of measurements is
received by the server the timestamp is applied. The server,
knowing the observation sampling rate, is then able to extrapolate
the approximate time at which each observation was made. The server
may augment the sequence with a location estimate. Such an estimate
may be derived by the server facilitating an interaction between
the device user and a map to obtain a location estimate. This may
be of particular benefit when the sequence of observations does not
contain location estimates.
[0218] The type of transmitter indicates the nature of the signal
that has been observed. Examples of these include Wi-Fi access
point, GSM cell, UMTS cell, CDMA cell, WCDMA cell, LTE base station
and WiMax base station.
[0219] The identifier is a set of one or more attributes of a
transmitter that enables that transmitter to be differentiated from
other transmitters. One attribute that may be used as part of the
set is the type of transmitter, for example GSM cellular, UMTS
cellular, and Wi-Fi access point. Another attribute that can be
used as part of the set is an identifier that is transmitted by the
transmitter and that is unique across a given type of transmitter.
Examples of such attributes include: for GSM cellular the CGI (Cell
Global Identifier consisting of the MCC, MNC, LAC and CID)
associated with each BTS (Base transmitter station), and for Wi-Fi
the MAC address.
[0220] The location of the measurement is represented as a set of
coordinates. Typically the coordinates will either be in geodetic
format (latitude and longitude) or in grid format (easting,
northing and grid zone).
[0221] The location uncertainty (also referred to as accuracy) is a
measure of quality of the location estimate that is associated with
a measurement. When combining measurements to provide an overall
model for a given transmitter this uncertainty provides the basis
for determining the amount of emphasis applied to each observation.
Typically, the more accurate the location estimate the greater
emphasis applied to that observation in the data fusion
process.
[0222] The source of the location measurement indicates the
reference from which the location information is obtained. Examples
of such sources include direct measurement such as is provided by a
GPS device, or inference derived from an existing model for the
coverage footprint of a Wi-Fi, WiMax or cellular transmitter,
direct user input of the location, as well as location information
from third party sources such as a database. User input is process
that involves direct human intervention. This may include
interacting with a map display on a mobile device to indicate the
current position and direct input of coordinates by the user into
the mobile device or the server. Third party data sources may
include a database containing information about the location and or
footprint of one or more wireless transmitters.
[0223] The timestamp represents the date and time at which the
measurement was made. Typically an accuracy and resolution of 1 to
10 seconds for the time is sufficient. Although other ranges of
time may be used in appropriate situations.
[0224] The received strength of the observed signal from a
particular transmitter may sometimes be available. In some cases
the signal level may provide an indication of how close the
measuring device is to the transmission source, as the stronger the
signal the closer the device is to the transmitter. In particular
the signal strength may provide an indication of whether this
observation was made closer to the transmitter than other
observations.
[0225] Mobile devices may provide a means for obtaining details
about the type of device and or the specific device itself. An
example of the device type is the TAC (Type Approval Code) which
identifies the type of device in GSM and UMTS. An example of a
specific device identifier is the IMEI (International Mobile
Equipment Number) which is a unique identifier assigned to each GSM
and UMTS mobile terminal.
[0226] In certain embodiments the network learning process may use
observations derived from data captured for other systems. Provided
that the data includes information that enables the temporal
relationship between measurements to be established it should be
possible pre-processed into a format such that it can be presented
to the system as an observation containing measurements of one or
more wireless transmitters. An example of such a source of data is
zone-based location system in which messaged from the mobile device
to the server contain measurements pertaining to one or more
cells.
[0227] A sequence of observations may be deconstructed into a set
of reference measurements. One or more of the observations in the
sequence may be deconstructed. For each such observation one or
more reference measurements may be created for each transmitter
within the observation. Each reference transmitter measurement
represents an instance of a particular transmitter having been
observed relative to some other reference. A reference measurement
includes an identifier for the transmitter, a timestamp, location
coordinates for the reference measurement and the uncertainty
associated with the location, and the source of the location
coordinates. The reference measurement may also have associated
with the measured signal strength if the observation contained a
signal strength measurement for that transmitter.
[0228] The identifier in the reference measurement is the
identifier of the transmitter obtained from the observation.
[0229] The timestamp for the reference measurement is that of the
observation from which the reference measurement derives.
[0230] The location coordinates and associated uncertainty may
derive from a number of sources. One such source is location data
that is part of the observation, for example a wholly or
substantially contemporaneous GPS observation. Another such source
is location data that is part of an another observation within the
sequence. When using such an observation the uncertainty assigned
to the location estimate for the reference measurement may be
adjusted to account for movement between the time the location
estimate was made and that of the reference measurement in
question.
[0231] Another source for the location estimate for the reference
measurement is proximity to other transmitters inferred by the
detection of other transmitters. Such a transmitter may be in the
same observation. Such a transmitter may be in another observation
from the sequence. Of particular importance are the observations
adjacent in the sequence to the observation in question as the
physical distance between the points at which the observations were
made is likely to be smaller. In the certain embodiments the
location estimate used in such cases is the centroid of the
transmitter footprint model and the uncertainty is the extent of
the transmitter footprint as measured by the covariance of the
footprint model.
[0232] In certain embodiments the reference measurement accuracy is
expressed as a scalar quantity. When using the transmitter
footprint model as a location estimate, the footprint extent may
need to be converted to a scalar quantity. For example if the
footprint extent is modelled as the covariance of a 2D Gaussian PDF
then the scalar accuracy may be obtained by taking the 4th root of
the determinant of the covariance matrix.
[0233] The reference measurements derived from an observation are
obtained by separating out each of the identified transmitters and
creating one or more reference measurements for each transmitter.
In certain embodiments a set of reference measurement for a
transmitter is generated by pairing that transmitter with each of
the other transmitters from the same observation. In certain
embodiments a set of reference measurement for a transmitter is
generated by paring that transmitter with each of the transmitters
from the adjacent observations from the sequence. In certain
embodiments and where the observation includes a location
measurement, such as is provided by a GPS receiver, a reference
measurement may be created pairing each identified transmitter with
the location estimate and associated uncertainty.
[0234] In the certain embodiments the set of reference measurements
for an observation is generated by pairing each transmitter in the
observation with the location estimate, if present, with each of
the other transmitters in the observation, and with each of the
transmitters in the adjacent observations. Reference measurements
are not generated for duplicated transmitter pairings or
self-referential pairings. This is illustrated via the following
example.
[0235] An observation M references cell A and Wi-Fi APs W and X.
The next observation N references cell A and Wi-Fi APs W and Y and
includes a GPS location estimate. The next observation includes
cell B and Wi-Fi APs X and Y. The reference measurements generated
by processing observation N are as follows: A with GPS, W with GPs,
Y with GPS, A with footprint for W, A with footprint for Y, W with
footprint for A, W with footprint for Y, Y with footprint for A, Y
with footprint for W, A with footprint for X, W with footprint for
X, Y with footprint for X, A with footprint for B, W with footprint
for B and Y with footprint for B.
[0236] In certain embodiments a reference measurement is made by
pairing some or all combinations of two transmitters present in the
observations. If for a given pairing of transmitters A and B where
a reference measurement for transmitter A is referencing the
location of transmitter B and where the existing network model
contains a footprint model for transmitter B, the location estimate
is assigned to the reference measurement for A uses the transmitter
footprint model of transmitter B. Where the network model does not
include a given transmitter the reference measurement is created
with a deferred location estimate referencing the unknown
transmitter.
[0237] In certain embodiments a reference measurement corresponding
to a transmitter in an observation may be obtained by calculating
an estimate and associated uncertainty of the location of the
device when the observation was made, and using as the location
information and accuracy in the reference measurement, the
estimated location and uncertainty for the device.
[0238] In certain embodiments the sequence of observation or part
thereof is examined to establish whether the device has moved
during the period of time that applies to the observations so
examined. If the device was not to have moved a significant
distance then location estimates from one observation may be
applied to one or more other observations. As an example if one or
more Wi-Fi access points are present in two or more observations
within a sequence then the device may be deemed stationary over the
period between these observations. If the device was deemed to be
stationary then the location estimate associated with the sequence,
if present, may be applied to one or more of the observations. If
the device was deemed to be stationary then reference measurements
may be generated by referencing transmitters from non-adjacent
observations as the location for the reference measurement.
[0239] In certain embodiments reference measurements for a Wi-Fi
access point are typically created using another Wi-Fi access point
model as the location estimate is that reference transmitter was
measured substantially or wholly contemporaneously with that access
point. For example an observation O contains measurements of Wi-Fi
access points W and X and another observation P contains
measurements of Wi-Fi access point Y. For access point W a
reference measurement may be created using the transmitter model
for access point X as the location as it is part of the same
observation. Creating a reference measurement for access point W
using the transmitter model for access point Y as the location is
conditional upon the elapsed time between the observations O and P.
If the elapsed time between 0 and P is greater than 1, 2, 5, 10,
15, 20, or 30 seconds then there exists the possibility that W and
Y are not actually close enough to warrant the generation of the
reference measurement.
[0240] Locations are modelled using a probability framework. This
enables algorithms to be used to derive a model for a transmitter
using its reference measurements. A GPS location estimate and an
associated accuracy indicator such as RMS error or Circular Error
Probably (CEP) can be converted into a PDF. Similarly a coordinate
obtained from a user in the form of a map reference can be modelled
as PDF. In this case the PDF will model the accuracy of the
underlying map data and the accuracy with which the user, on
average, can specify their location from that map. An indicative
accuracy would be 1/5, 1/8, 1/10, or 1/20 of the map zoom level. In
certain embodiments the PDFs used will have a two-dimensional
Gaussian distribution.
[0241] In some observations one or more of the measured
transmitters may not be known to the system. Furthermore, there may
be no other source of location information corresponding to the
observation. For some of these observations there may not be any
known transmitters in earlier or later measurements. In such
scenarios the location information within the reference measurement
may be marked as a deferred location. In a deferred location
information is stored enabling the system to populate the location
coordinates and uncertainty at a later time when suitable
information has become available.
[0242] In the reference measurement the deferred location estimate
may be represented by a record of the proximity to an unknown
transmitter identified using its identifier. The deferred location
estimate may reference proximity to an unknown transmitter from the
same observation or the preceding observation or the succeeding
observation.
[0243] For example an observation contains transmitters A, B, and
C. The previous observation contains transmitters X and Y and the
next observation contains transmitter D. All of the transmitters
are unknown to the system at the time. For the observation
containing A, B, and C a set of reference measurements is made. For
transmitter five reference measurements are created each having a
deferred location estimate indicating proximity to transmitters B,
C, X, Y, and D respectively. Similarly five reference measurements
are created for each of transmitters B and C.
[0244] The following examples illustrate the generation of
reference measurements for specific combinations of transmitter
measurements within the same observation and across adjacent
observations. In the figures associated with these examples the
true location of the transmitter is shown by a + symbol. For
directional cellular transmitters the directionality of the
transmitter antenna is shown via a circular arc. The radius of this
arc is somewhat arbitrary. The centroid of the footprint model is
indicated by a * symbol. The extent of the footprint is shown as a
contour of constant probability which, when the reference
measurement uncertainty is expressed as a scalar, will be a circle.
In the following example the location uncertainty is expressed as a
scalar. This scalar quantity is derived as the 4th root of the
determinant of the footprint model covariance matrix.
[0245] Typically cellular receivers only report the current serving
cell. To create a reference measurement for a cell with respect to
a different cell the reference cell will need to derive from a
different observation. An observation containing measurements of 1
cell may be used to create a reference measurement for that cell by
using measurements of another cell from another observation in a
sequence. In certain embodiments the another observation will be an
adjacent observation.
[0246] In certain embodiments an observation measuring cell B may
result in a reference measurement being created using cell A as the
reference where cell A was observed in an adjacent observation from
the sequence. Provided a model for cell B already exists in the
network model a reference measurement for cell A may be created
using the footprint model for cell B to provide the location and
uncertainty for the reference measurement.
[0247] FIG. 15 illustrates the generation of a reference
measurement for cell 21701 based on an observation containing cell
21701 and an adjacent observation containing cell 20503. The figure
shows the reference measurement for 21701 is created using the
footprint model for cell 20503. In this example the reference
measurement for cell 21701 may be made irrespective of whether
there is an existing footprint model for cell 21701 in the current
network model.
[0248] In certain embodiments an observation measuring access
points B and A may result in reference measurements to be made for
A with respect to B or B with respect to A or both. Provided a
model for access point B already exists in the network model a
reference measurement for access point A may be created using the
footprint model for access point B to provide the location and
uncertainty for the reference measurement. Similarly a reference
measurement for access point B may be created using the footprint
model for access point A provided a model for access point A
exists. Where a footprint model does not exist a reference
measurement may still be created by using a deferred location
estimate.
[0249] In certain embodiments where an observation measuring access
point A is succeeded by an observation measuring access point B a
reference measurement for A can be create using B as the location
reference. Provided a model for access point B already exists in
the network a reference measurement for access point A may be
created using the footprint model for access point B to provide the
location and uncertainty for the reference measurement. If no such
model exists for B then the location estimate can be set as a
deferred location estimate with respect to B.
[0250] FIG. 16 illustrates the generation of a reference
measurement based on the observation containing 2 Wi-Fi Access
Points having MAC addresses 87723982376 and 62644269161, The figure
shows the reference measurement for 87723982376 is created using
the footprint model for access point 62644269161. In this example
the reference measurement for access point 87723982376 can be made
irrespective of whether there is an existing footprint model for
access point 87723982376 in the current network model. A reference
measurement may also be made for access point 62644269161. If there
is a footprint model for cell 87723982376 then the reference
measurement location and uncertainty will use this footprint model.
Otherwise the reference measurement would need to include a
deferred location reference to access point 87723982376.
[0251] In certain embodiments an observation measuring access point
A and cell B may result in reference measurements to be made for A
with respect to B or B with respect to A or both. Provided a model
for access point B already exists in the network model a reference
measurement for access point A may be created using the footprint
model for access point B to provide the location and uncertainty
for the reference measurement. Similarly a reference measurement
for cell B may be created using the footprint model for access
point A provided a model for access point A exists. Where a
footprint model does not exist a reference measurement may still be
created by using a deferred location estimate.
[0252] In certain embodiments where an observation measuring access
point A is adjacent to an observation measuring cell B a reference
measurement for A can be create using B as the location reference.
Provided a model for cell B already exists in the network a
reference measurement for access point A may be created using the
footprint model for cell B to provide the location and uncertainty
for the reference measurement. If no such model exists for B then
the location estimate can be set as a deferred location estimate
with respect to B.
[0253] In certain embodiments where an observation measuring cell C
is adjacent to an observation measuring access point D a reference
measurement for C can be create using D as the location reference.
Provided a model for access point D already exists in the network a
reference measurement for cell C may be created using the footprint
model for access point D to provide the location and uncertainty
for the reference measurement. If no such model exists for D then
the location estimate can be set as a deferred location estimate
with respect to D.
[0254] FIG. 17 illustrates the generation of a reference
measurement based on the observation containing Wi-Fi access point
87723982376 and cell 21701. The figure shows the reference
measurement for 21701 is created using the footprint model for
access point 87723982376. In this example the reference measurement
for cell 21701 be made irrespective of whether there is an existing
footprint model for cell 21701 in the current network model. A
reference measurement may also be made for access point
87723982376. If there is a footprint model for cell 21701 then the
reference measurement location and uncertainty will use this
footprint model. Otherwise the reference measurement would need to
include a deferred location reference to cell 21701.
[0255] When an observation contains a location estimate, or a lack
of movement enables a location estimate from another observation to
be applied to the observation, a reference measurement may be made
for each measured transmitter setting the location and uncertainty
as the GPS location estimate and associated uncertainty. FIG. 18
illustrates an example of such a reference measurement using a
cellular transmitter. The observation in which cell 21701 was
measured has an associated GPS measurement. A reference measurement
for cell 21701 is created using the GPS measurement and associated
uncertainty as the location estimate.
[0256] An observation measuring transmitter A is paired with an
observation of a cell B to create a reference measurement for
transmitter A. The observation measuring a transmitter B may derive
from the same observation as A or a proceeding or a succeeding
observation. Transmitter B may be from the same network or a
different network to that of transmitter A. The current network
model does not contain a reference to transmitter B. In this
instance to create a reference measurement for transmitter A
deferred location estimate measuring transmitter B is used. The
reference to transmitter B includes the type of transmitter as well
as the identifier.
[0257] The system maintains a model for each transmitter which
includes a representation of the area over which the transmitter
may be detected by a mobile or client device. This area is referred
to as the footprint of the transmitter. In cellular networks a
mobile terminal typically reports the current serving cell which is
the cell corresponding to the best, typically the strongest,
received signal. As such the footprint model for that transmitter
is the area over which that transmitter is likely to be selected by
a client device as the best signal. For Wi-Fi networks the client
devices may report a set of one or more access points that are
currently able to be detected. As such the footprint of a Wi-Fi
access point is the area over which the signal can be detected by a
client device.
[0258] Transmitter footprints are modelled as a probability density
function (PDF). The parameters of this PDF are obtained by
processing all of the reference measurements associated with the
transmitter. This provides a particular advantage for generating
location estimates based on measurements from different types of
transmitters as all transmitters may be modelled in a unified
manner. The use of a PDF provides the mathematical basis for
combining measurements of different transmitters to produce an
estimate of the location as well as an indication of the quality of
that estimate. At the same time the system can learn about
previously unknown transmitters or improve the quality of
information about known transmitters. Similar principles can be
applied to other types of transmitter such as CDMA and WiMAX.
[0259] The model for a given transmitter may be obtained by
processing all of the reference measurements for that transmitter
to derive the parameters of the PDF which models the footprint of
that transmitter.
[0260] Both the transmitter reference measurements and transmitter
footprints are modelled via probability density functions. One
useful PDF is a two-dimensional Gaussian distribution. Such a
distribution confers a number of advantages.
[0261] When two or more Gaussian measurements are combined the
result also has a Gaussian distribution. A number of measurements
can be combined and the output has the same form, or substantially
the same form. Thus there may be no need for empirical rules to
deal with certain combinations of measurements or limit to the
number of measurements that can be processed at a given time.
[0262] Processing the measurements is a linear operation. This
means that the computations need not be complex and may scale
linearly with the number of measurements being processed.
[0263] Accordingly, certain embodiments result in an implementation
that is computationally efficient and robust.
[0264] One form of the PDF for modelling each reference measurement
of a transmitter is a two-dimensional Gaussian distribution. The
mean of the distribution is specified as the location estimate and
is denoted Xm(i) where the index i indicates that it is the i-th
reference measurement of that transmitter:
X m ( i ) = [ x ( i ) y ( i ) ] ##EQU00004##
[0265] The uncertainty of the measurement is represented by the
covariance denoted Cm:
C m ( i ) = [ .sigma. x 2 ( i ) .sigma. xy ( i ) .sigma. yx ( i )
.sigma. yx 2 ( i ) ] ##EQU00005##
[0266] The matrix Cm is symmetric and thus
.sigma.xy(i)=.sigma.yx(i) Each such measurement represents the fact
that a mobile device detected the identified transmitter while
situated at an estimated location Xm(i) having associated
positional uncertainty Cm.(i)
[0267] The 2D Gaussian PDF can be represented graphically via a
contour of constant probability. Such a contour is an ellipse
centred on the position measurement.
[0268] Measurements from independent sources such as GPS typically
report the measurement uncertainty as a scalar accuracy such as the
2DRMS or CEP. Such accuracies are readily converted to an
equivalent covariance matrix form. The first step is to convert the
scalar to a quantity equivalent to 1 standard deviation denoted
a(i). The covariance matrix is then constructed with the diagonal
elements equal to .sigma.(i)2 and the off-diagonal elements set to
0.
C m ( i ) = [ .sigma. ( i ) 2 0 0 .sigma. ( i ) 2 ]
##EQU00006##
[0269] In this case a constant probability contour will be a circle
centred on Xm.(i)
[0270] The preferred form of the PDF for modelling the footprint of
a transmitters is a two-dimensional Gaussian distribution. The
centroid of the footprint is the mean of the distribution and is
denoted denoted Xt:
X t = [ x t y t ] ##EQU00007##
[0271] The extent of the footprint is represented by the covariance
denoted Ct:
C t = [ .sigma. xt 2 .sigma. xyt .sigma. yxt .sigma. yt 2 ]
##EQU00008##
[0272] The matrix Ct is symmetric and thus .sigma.xyt=.sigma.yxt.
As with the measurements the footprint can be graphically
represented by a contour of constant probability.
[0273] FIG. 19 illustrates the footprints for sectorised cellular
transmitters. The cell IDs of these transmitters are 21701, 21702,
and 21703. The circular arcs indicate the directionality of the
antennas. The transmission power each antenna is focused in the
direction of the antenna boresight with the result that
observations of that transmitter will be concentrated in front of
the antenna. As a consequence the transmitter footprint are
typically centred some distance to the front of the antenna
(indicated by an asterisk) and the extent of the footprint will
cover the area in front of the antenna (indicated by an ellipse).
The ellipse is a contour of probability=0.95 for the 2D gaussian
PDF mode for the corresponding antenna.
[0274] FIG. 20 illustrates the footprint for a transmitter with an
omni-directional antenna. The network is the same as that used in
FIG. 19 except for the transmission site which was changed to be a
single omni-directional antenna. The transmission power of the
antenna is distributed equally in azimuth with the result that
observations of the transmitter are expected to be distributed
equally around the transmitter. As a consequence the transmitter
footprint will be centred closer to the transmitter than for a
directional antenna and the extent will be more evenly distributed
around the transmitter. This is evident in FIG. 20 with the
centroid being much closer to the transmitter than for the
sectorised antennas and the 0.95 probability contour encloses the
transmitter whereas in the sectored examples the 0.95 probability
contour does not.
[0275] The footprint model for a transmitter is obtained by
processing all of the measurements corresponding to that
transmitter. The mean of the PDF is the weighted mean of the
measurements. This means the more accurate a measurement is the
more influence it will have on the result. Conversely a measurement
that is not accurate is not given much influence. Provided the
measurement uncertainty is reliably determined, this has the
advantage that measurements that contain larger errors do not
distort the result as significantly as would happen if all
measurements were treated equally.
[0276] The estimated location of a measurement of transmitter is
denoted Xm(i) with the corresponding measurement covariance denoted
C m(i). For measurements modelled with 2D Gaussian PDFs, the mean
of the PDF, the centroid of the transmitter footprint, is defined
by
X t = ( i C m ( i ) - 1 ) - 1 i ( C m ( i ) - 1 X m ( i ) ) ' Mean
of Transmitter footprint for 2 D Gaussian PDF Equation 1
##EQU00009##
[0277] Where the -1 superscript denotes the matrix inverse and the
superscript indicates the matrix transpose and multiplication
operations are matrix multiplications.
[0278] The extent of the transmitter footprint is given by the
weighted sample covariance of the measurements. For measurements
modelled with 2D Gaussian PDFs the sample covariance is calculated
as:
C t = ( i C m ( i ) - 1 ) - 1 i ( ( X m ( i ) - X t ) ' C m ( i ) -
1 ( X m ( i ) - X t ) ' ) Covariance of Transmitter footprint for 2
D Gaussian PDF Equation 2 ##EQU00010##
[0279] For the case where the measurement accuracies are all
expressed in terms of scalar quantities then the mean and
covariance of a 2D Gaussian PDF modelling the transmitter footprint
becomes
X t = ( i 1 .sigma. ( i ) 2 ) - 1 [ i x m ( i ) .sigma. ( i ) 2 i y
m ( i ) .sigma. ( i ) 2 ] Mean of Transmitter footprint for 2 D
Gaussian PDF where the location accuracy is a scalar Equation 3 C t
= ( i 1 .sigma. ( i ) 2 ) - 1 [ i ( x m ( i ) - x t ) 2 .sigma. ( i
) 2 0 0 i ( y m ( i ) - y t ) 2 .sigma. ( i ) 2 ] Covariance of
Transmitter footprint for 2 D Gaussian PDF where the location
accuracy is a scalar Equation 4 ##EQU00011##
[0280] Where
(xt,yt) are the elements of Xt and are the mean of the location
estimates from the reference measurements, and (xm(i),ym(i)) is the
estimated location for the ith measurement, and .sigma.(i) is the
associated scalar accuracy converted to a standard deviation.
[0281] For example, with a GPS receiver reporting the accuracy as a
2DRMS value, the accuracy needs to be divided by 2 to obtain the
equivalent value as a Gaussian standard deviation.
[0282] FIG. 21 illustrates reference measurements for a directional
cellular transmitter for which the uncertainty associated with the
location coordinates was specified as a scalar quantity. The
accuracy varied between 25 m and 100 m. The centroid of the
footprint model, the weighted mean of the reference measurements,
is plotted as an asterisk. The extent of the footprint mode, the
weighted covariance of the reference measurements, is plotted as a
0.95 probability contour.
[0283] Deferred location estimates may be resolved during the
update network model (120) process within the network model cycle.
When the network model is being updated the system will have access
to the existing model which resulted from the previous update as
shown by the loop in the network model lifecycle (FIG. 2). The
lifecycle of a transmitter (FIG. 9) is repeated in each iteration
of the network update (120).
[0284] FIG. 22 illustrates an exemplary process for resolving a
deferred location estimate during a network update cycle (240) by
referring to the existing network model as the unknown transmitter
may have since had a model constructed. If the transmitter
referenced in the deferred location estimate is modelled in the
network then the coverage model for that transmitter is applied to
that measurement. In the preferred embodiment the deferred location
estimate is replaced by the mean of the transmitter footprint PDF
and the deferred location estimate uncertainty is replaced by the
covariance of the transmitter footprint PDF. If the transmitter
referenced in the deferred measurement does not have a model in the
existing network model then the associated reference measurement is
suppressed from the current network update cycle.
[0285] The observations collected by the system corresponding to a
particular transmitter may be non-uniformly distributed in space.
If this sampling distribution is not sufficiently accounted for in
some manner this may lead to a bias in the transmitter footprint
model due to more reference measurements originating from one area
within the transmitter's coverage area than from other areas. An
example of this issue is a cell that covers a public transport
interchange. Such an interchange may generate many more transmitter
observations compared to that from other areas within the coverage
of the cell due to the greater number of users visiting the
interchange compared to other regions covered by the cell.
[0286] In certain embodiments the bias potentially introduced by
spatially non-uniform sampling of the transmitter observations is
reduced by limiting the number of reference measurements that are
processed for a given area. In certain embodiments the area over
which reference measurements have been observed for a given
transmitter is overlaid with a rectangular grid of n.times.m
rectangular regions of equal area. Typically n and m will be
between 5 and 500. The side of the rectangular regions will be
restricted to a minimum value so as to avoid the granularity of the
grid becoming too small which defeats the purpose of the grid. A
typical minimum value for the side of the grid rectangle is 50 m.
However, other values may be used such as those between 25 m and
200 m, 25 m and 100 m, 50 m and 100 m, or 50 m and 150 m. The
reference measurements are grouped according to the grid region
within which the estimated location for that measurement lies. Each
such group of measurements is then processed to produce a reference
measurement to represent that region of the coverage footprint in
the modelling of the transmitter footprint. If a grid region
contains no reference measurements then that grid region is not
represented in the modelling of the footprint transmitter. The
result is that a set of reference measurements for a transmitter
are reduced to a set of at most n.times.m reference
measurements.
[0287] The exemplary means for processing the footprint
measurements is to create for each square in the grid containing
one or more reference measurements a representative reference
measurement with the location estimate set as the centre of the
square and the associated uncertainty being set to that of the
reference measurement within that square with the smallest
uncertainty. Where the uncertainty is represented as a covariance
matrix the smallest uncertainty is comparable to the covariance
yielding the smallest determinant.
[0288] FIG. 23 illustrates a set of 155 reference measurements for
a cellular transmitter. The measurements are biased with more of
the measurements to the North-East. The result of normalising the
data by using a 26.times.14 grid is shown in FIG. 24. Where one or
more original reference measurements falls within a grid a
reference measurement for that grid was created with the reference
measurement location set as the centre of the grid and the accuracy
set as the minimum of the accuracies of the reference measurements
within that grid region. This process resulted in the 150 original
reference measurements being normalised over the grid to 102
reference measurements. The effect of the normalisation process can
be seen by comparing the transmitter model footprints. With out the
normalising process the cell footprint mode centroid is at (328933,
6263063). With the normalising process the cell footprint mode
centroid is at (328897, 6263053). The effect of the bias in the
reference measurements causes the footprint model to be biased
towards the area where the measurements are denser. For the
normalised set of reference measurements the footprint is more
symmetric about the boresight of the sectorised antenna of the
cellular transmitter.
[0289] In some cases an external entity, for instance a network
operator, may be willing to make available an extract of the
network configuration corresponding to part of or the whole of a
network. An example of such data is an MNO providing the details of
one or more cells in a cellular network. This might be, for
instance, in exchange for some consideration but without having to
commit to the overhead of maintaining the information and providing
frequent updates. Another example of is the operator of a series of
Wi-Fi access points, for example those operated by some fast food
outlets, providing an extract of the configuration details
concerning of one or more Wi-Fi access points.
[0290] For a cellular network the supplied data may include for
each transmitter one or more of the parameters listed in Table 3.
The data supplied will typically comprise at a minimum the
identifiers for each cell (for example Cell ID and LAC in GSM &
UMTS) along with the location of the corresponding cell antenna. In
some cases additional information such as antenna height, azimuthal
orientation and antenna characteristics may also be supplied. Here
we describe how this information can be assimilated into the
network learning framework of certain embodiments of the present
disclosure. FIG. 28 illustrates an exemplary process.
[0291] For a Wi-Fi network the data supplied will typically
comprise at a minimum the identifiers for each cell (for example
MAC address) along with the location of the corresponding cell
antenna. In some cases additional information such as the SSID may
also be supplied. In some cases additional information such as the
transmit power may also be supplied.
[0292] In some cases, the information may be supplied in a format
that is not directly compatible with the import format supported by
the network database server. In this case, a custom script may be
employed to reformat the data for loading by the corresponding
network database server. Additionally in some cases one or more
parameters of the models used by the network database server might
not be available from the operator's extract and would therefore be
initialized to default values. For example missing antenna heights
for macro cells could be initialized to a default value between 10
and 20 metres, antenna gain values for sectorised cells can be
defaulted to a value between 12 and 16 dB, broadcast channel
transmission powers for macro cells can be defaulted to values
between +33 and +43 dBm. The specific default values used can be
adjusted according to local conditions. For example casual
observation of macro cell heights in Australia and UK reveals that
in Australia cell towers tend to be perceptibly taller than in the
UK. Accordingly a suitable default for height in Australia could be
20 m whereas a value of 10 m could be used in the UK.
[0293] Once the supplied data have been reformatted to the internal
NWDB format used by the NWDB server and any missing parameters have
been initialized using default values, a propagation model may be
used to estimate the serving footprint of each supplied cell. Such
simulation techniques are known by those in the industry and are
typically applied to estimate so-called best server polygons,
namely the region within which the cell of interest is likely to be
selected by a mobile terminal as the serving cell. Using a discrete
approach to the simulation, yields a collection of discrete points
at which the cell of interest is likely to serve. This set of
points can then be used to derive a footprint model for the
transmitter.
[0294] In certain embodiments the transmitter footprint is modelled
as a 2D Gaussian PDF. The parameters of this model are obtained
from the propagation modelling by taking the mean and covariance of
the points at which the transmitter is deemed hearable. FIG. 26
illustrates the propagation modelling and footprint modelling for a
GSM cell (Cell ID 21701). The network is simulated over a grid of
points. At each of the points marked with an asterisk the signal
levels for each cell were simulated. Where 21701 is the strongest
cell the asterisk is enclosed by a circle. The 2D Gaussian PDF
model for cell 21701 is obtained by calculating the mean and
covariance of the locations at which cell 21701 was deemed the
strongest cell. The resulting mean is as a plus symbol. The
resulting covariance is shown as a contour of constant
probability=0.95.
[0295] Having obtained this model for a transmitter, the aim is to
integrate this information with other sources of information about
that transmitter. This is done by generating from the footprint
model, a set of reference measurements which accurately reflect the
estimated footprint. In other words, a set of reference
measurements which if processed using the learning algorithms will
return a footprint corresponding to the one estimated using
propagation modelling. This can be done, for example, by selecting
5 points, one situated at the centroid of the footprint and 4
others situated at the extremes of the semi-major and semi-minor
axes of the constant probability contour ellipse corresponding to
the covariance.
[0296] Two other parameters may be useful when creating the 5
reference measurements. One is the timestamp. Reference
measurements derived from live terminals are tagged with a
timestamp corresponding to the time when the measurements were
observed. The timestamps for the 5 simulated reference measurements
for an externally supplied cell, are set to the date and time at
which the information was last known to be valid. Typically this
will be the time when the measurements were extracted from the
operator's planning database. In some cases however if the
information is obtained from a historical archive, the timestamp
for the reference measurements should be set to an estimated date
corresponding to when the information was valid. Setting the date
in this way enables these measurements to be mixed seamlessly with
live reference measurements and for any changes in the network
since the externally supplied information, reflected in more recent
live reference measurements to take effect. The second important
parameter is the accuracy associated with each reference
measurement. This should be set to a value which reflects both the
quality of the source data as well as the estimated accuracy of the
best server propagation modelling. Typically cell site location
information recorded by radio planning departments is obtained
using GPS receivers and therefore the accuracy is 50 m or better.
The spatial accuracy of the best server polygon prediction will
vary with the propagation model used, including whether digital
elevation information as well as local clutter models were
employed. As a default, an accuracy value of 100 m may be assigned
to each of the simulated reference measurements.
[0297] FIG. 27 continues the example from FIG. 26. The footprint
model for cell 21701 is used to generate five reference
measurements of the cell. The location of the reference
measurements is shown as an asterisk. The large ellipse is the 0.95
probability contour for the transmitter model PDF. The smaller
ellipses surrounding each of the reference measurements are the
uncertainty that is associated with the reference measurement. The
uncertainty was set to a standard deviation of 50 m. The
uncertainty is illustrated in the figure as a 0.95 probability
contour which appear as ellipses around each of the reference
measurement locations.
[0298] An exemplary flowchart for the process of generating
reference measurements for a transmitter using the third party
transmitter characteristics is shown in FIG. 28.
[0299] In certain embodiments the coverage footprint model of one
or more transmitters may be based directly on the type of and
classification of the transmitter. An example of this is Wi-Fi
access point transmitters. Such transmitters typically use
omni-directional antennas and have a relatively limited range,
typically 50 m or less. Furthermore the footprint may be dependent
upon the environment, walls, floors, etc. For such a transmitter
the footprint can be specified without using propagation modelling.
In certain embodiments this model is a 2D Gaussian PDF centred on
the transmitter coordinates. The covariance of the PDF would
reflect the uncertainty in the antenna coordinates and the range of
the access point. A typical value for the covariance matrix is for
the diagonal elements to be 502 with the off-diagonal elements set
to 0. Another example is cellular transmitters classified as a pico
cell for which a typical diagonal element of the covariance would
have a value of 1002. Another example is cellular transmitters
classified as a micro cell for which a typical diagonal element of
the covariance would have a value of 2002.
[0300] The system in accordance with certain embodiments provides a
mechanism through which externally obtained radio network
configuration information can be incorporated into the overall
network modelling and maintenance process. In certain aspects this
incorporation can be seamless or substantially seamless. The
information provided can be partial or can cover the entire
network. This capability yields further benefits for deployments
done in cooperation with a Mobile Network Operator (MNO). Unlike
conventional installations where the location system requires an up
to date network database, the present system in accordance with
certain embodiments can accept updates from the MNO whenever it is
convenient for the radio network engineering department and in the
meantime, the system continues to update its master database using
information sourced from user activity. The capability of the
system to use data covering part of the network may be of
particular value where a network has been deployed as a series of
turn-key projects and the vendor for one of the projects provides
the network configuration data for that project whereas a vendor
for a different project does not provide the data or provides it at
a later time.
[0301] In some systems models are constructed iteratively where
each measurement is processed and used to update the model to
create a new version of the model. The measurement is then
discarded. Using certain disclosed embodiments, the present
application of modelling transmitter footprints may be done by
applying each measurement of a transmitter in a recursive mean
covariance calculation. One benefit of this type of recursive
approach is that it minimises the data storage requirements as well
as the computation time to update the model following the input of
one or more new measurements. One disadvantage of such an approach
however is that it is more vulnerable to errors in individual
measurements. Furthermore in cases where the underlying physical
process changes, observations made of the changed system are still
added to the existing model with the result that the updated model
may be neither an accurate reflection of the old system nor the new
system. Consider a system where the location of a Wi-Fi access
point is being estimated and that this access point changes
location due to the owner of the access point moving to a new
address. An iterative system would add observations made of the
access point at its new location to the estimate made from
observations of the access point when at the original location. The
result is that the location estimate will neither be an indication
of the original location or the new location.
[0302] In certain embodiments, significant measurements
corresponding to a transmitter may be retained. When a model for a
transmitter is updated it is in effect re-computed using the
retained measurements. This has the advantage that, instead of
irretrievably folding errors into the model, it provides an
opportunity to detect changes to one or more parameters associated
with the transmitter and then limit the processing of measurements
to the subset that reflect the time after which the change was
made. Examples of such changes are a Wi-Fi access point being
relocated, a COW (cell on wheels) being relocated, cell site
reconfiguration such as antenna re-alignments, distributed antennas
and being converted to distinct cells.
[0303] In a similar manner the system, in accordance with certain
embodiment, can also detect transmitter identity changes. In GSM
and UMTS cellular networks the LACs may change as part of system
performance management. Because the history of measurements is
maintained the abrupt end of measurements for one CID and LAC and
the commencement of measurements for another CID and LAC can be
detected enabling the system to rapidly adapt to the change. The
system can detect the change, determine that although the
transmitter identity has changed it is the same underlying
transmitter and thus join the two sets of measurements enabling the
footprint of the model of the "new" transmitter to be accurately
determined without having to start from scratch. In many networks
the CID is not changed thus simplifying the measurement alignment
process.
[0304] By storing the device type and/or the specific device
identifier in the network learning collection, quality control
mechanisms are enabled to deal with defective handsets and
particular problem devices to preserve the integrity of the
collection. For example in certain devices when the device enters
standby mode, the cell measurements cease being updated however
there is no indication to the client application that this is the
case. As a result network learning measurements based on adjacent
measurements might incorrectly derive a reference measurement for
one cell based on an adjacent reported cell when in fact the two
cells are separated by a large distance, which the user traveled
while the device was in standby mode. The presence of the TAC in
each network learning reference measurement means that if such
errors are discovered after one or more such handsets have
contributed measurements to the collection, a script can be run to
isolate and purge all measurements originating from a device of
that type. Similarly the inclusion of the identifier in each
reference measurement enables a series of erroneous measurements
originating from a particular client to be isolated and purged from
the collection.
[0305] This section describes certain embodiments for estimating
the location of a subscriber using the wireless network information
accumulated in the database and measurements reported by the
device.
[0306] The client software on a user's device collects and
maintains a history of wireless network measurements as well as GPS
measurements when available. When a location request is invoked
(either locally by the user of an application or remotely in the
form of a message received from the network, the client may refresh
the measurement information by requesting the latest wireless and
or GPS measurements from the terminal and then encode and transmit
the available measurements to the server for processing. Methods
for processing this information to obtain a position estimate and
associated uncertainty estimate are described in the following
sections. In certain aspects the specific processing carried out
will depend to a large extent on the type of information reported
by the client. This in turn depends on the environment in which the
device is currently located the capabilities of the devices and any
user configuration settings. In the present context, because the
radio network models are likely to be in a constant stage of change
as available information is used to refine the models, the
processing carried out to estimate location may also vary according
to the current state of the network models corresponding to the
elements featured in the measurement reports.
[0307] If the measurements reported by the client include GPS based
coordinates, in the majority of cases, these coordinates will be
given priority over any other information reported by the client.
In some cases however the GPS information may be aged. In such
cases, the wireless measurement history may be examined to see
whether it indicates that the device may have been moving in the
time that has elapsed since the last GPS position fix. If the
wireless measurements do not indicate any movement, then the
earlier GPS position will be returned. The uncertainty indication
returned in this case would also be tight--reflecting the intrinsic
accuracy of the GPS position fix.
[0308] On the other hand if the wireless measurements indicate a
change in received information in the intervening interval then a
hybrid position estimate & associated uncertainty estimate
would be returned based on the combination of the GPS fix and a
position estimate from the more recent wireless measurements. The
combination could be calculated as a covariance weighted mean of
the two location estimates where the covariance associated with the
GPS location has been increased from the intrinsic GPS position fix
accuracy to reflect the uncertainty due to time (and likely
movement of the device in that period).
[0309] In some cases, there may be a limit on the number of
measurements that can be reported--for instance if they have to be
sent in a single SMS. In this case, the availability of the tasking
information enables the client to identify the most valuable
information to report. It can then convey the most useful
information, enabling a more accurate result. Typically, the
strongest rxLevs constrain position more tightly. Additionally,
cells which have better quality indicated in the tasking
information will typically yield less positional uncertainty in the
results.
[0310] The location calculation processing can be implemented as a
framework which can utilize one or more measurements (of possibly
different types) to estimate location. In certain embodiments each
measurement is applied to a cost function which is appropriate to
that type of measurement to calculate a cost at a given putative
location. An optimization process is applied using the cost
function to evaluate a set of possible locations and choose the one
which is most consistent with the available information. This
framework makes it possible to combine information of different
types and weight them in appropriate proportion to their
significance. The following paragraphs describe the cost functions
for the different types of measurements.
[0311] For a cellular serving cell measurement (no rxLev reported),
the form of the cost function used depends on the state of the
model for the corresponding cell. Naturally if there is currently
no corresponding cell model in the network, then that particular
measurement is overlooked in the position calculation.
[0312] If the model is still relatively coarse, and being modelled
as a collection of points represented by mean and covariance, then
the cost for a given point X is calculated simply as the
Mahanalobis distance
cost=(X-M)'C.sup.-1(X-M)
where M and C are the mean and covariance of the accumulated
observations for the corresponding cell.
[0313] Alternatively, if the cell is being modelled as a BTS, the
cost is calculated as
cost=-log(p(S|X))
[0314] Where p(S|X) represents the probability that S will be the
strongest cell at location X. [0315] Typically in a location
server, the cost function for such this type of measurement
reflects the anticipated variation in signal levels and therefore
in the serving cell selection process, but assumes that other
parameters such as the base station location are known perfectly.
In the present context however where the parameters of the wireless
network configuration are themselves (separately) being estimated,
the uncertainty associated with these model parameters also has to
be taken into account in the location estimation. This ensures for
instance that is a pair of measurements are reported, where one
pertains to a cell whose location is known very precisely and the
other pertains to a newly added cell whose location is still
uncertain, the location estimation process will weigh the former
measurement more heavily.
[0316] If the model is still relatively coarse, and being modelled
as a collection of points represented by mean and covariance, then
the cost for a given point is calculated simply as the Mahanalobis
distance
cost=(X-M)'C.sup.-1(X-M)
where M and C are the mean and covariance of the accumulated
observations for the corresponding cell. Compared to the cell ID
only case above, the M-distance is computed using a 3 element
vector consisting of [x, y, rxLev] parameters of the cell that have
been measured. S may consist of a cell identify for which the cost
represent the probability that at X S will be the serving cell. S
may consist of a cell identify and a received signal strength for
which the cost will represent the probability that at X S will be
the serving cell and will have the observed signal strength.
[0317] If a more detailed BTS model is available for the
corresponding cell, the form of the cost is
cost = ( r ^ - r ) 2 .sigma. 2 . ##EQU00012##
[0318] Where sigma in this case reflects not only the expected log
normal fading in this region but also the effect of the location
uncertainty. This is estimated by checking the variation in
predicted levels over the location and antenna angle uncertainty.
r-hat is the predicted rxLev and r is the measured rxLev.
[0319] A common challenge in using this type of cost function for
cellular is that the propagation model used to predict the rxLev at
a putative position uses a general model which is more suited for
outdoor propagation. If the device is situated indoors, there is a
significant likelihood of greater attenuation due to indoor
penetration. Typically the effect of this is to make the predicted
levels stronger than the corresponding measurements and therefore
introduce a bias into the estimation process. In the present
context there may be some cases where in addition to cellular
wireless measurements, one or more measurements which correspond to
fixed wireless networks are reported. For fixed wireless network
types such as Wi-Fi, deployments are primarily indoors. Therefore
if such measurements are co-reported with cellular measurements,
the location estimation process assumes that the device is indoors
and the propagation model is adjusted by adding a loss term
representing typical indoor penetration loss for the corresponding
cell. Note that the specific value applied would be adjusted
according to the frequency band of the cell. For 900 MHz a value of
around 9 dB can be used as an average across different building
materials, different floors and different environments. The
additional variation in the received signal can be modelled by
adding a log normal term with a standard deviation of 4 dB to the
standard deviation representing fast & slow fading in the
channel. While some differences have been measured in building
penetration loss between 900 MHz, 1800 MHz and 2100 MHz, the above
values are a suitable approximation across all of these.
[0320] For short range fixed wireless measurements (for instance a
Wi-Fi AP), typically embodiments will use a simple mean &
covariance representation for the footprint of the AP rather than a
detailed radio transmitter model.
[0321] One option when hearing a Wi-Fi AP would be simply to return
the location of the AP due to the relatively short range of the AP
compared to cellular wireless links. This could be instead of
incorporating these measurements into a combination with cellular
measurements etc.
[0322] In practice there are a few benefits to combining the Wi-Fi
measurements with any cellular measurements in a complete solution.
One reason is that while the Wi-Fi link range may be short, the
positional uncertainty associated with a Wi-Fi may still be
relatively large at some point of time. This could be the case for
instance if no GPS based location has been measured for the AP-only
measurements relative to cellular transmitters. Therefore the form
of the cost function for the Wi-Fi measurement reflects this
uncertainty and enables this measurement to be included and
weighted appropriately, reflecting the uncertainty associated with
the location of the AP. Another benefit relates to the potential
mobility of Wi-Fi APs. It is possible that an AP which was
previously measured in one location has been moved and has now been
reported by a device from a different location however the radio
network database has yet to be updated. In such a case, the
combined solution will reveal the inconsistency between the other
measurements and the Wi-Fi measurement in the form of a large
residual. In this case, the Wi-Fi measurement can be excluded and a
location recalculated.
[0323] The form of the cost function for a Wi-Fi measurement is
similar to that described above for a cell ID only measurement. At
a given putative position, the cost
cost=(X-M)'C.sup.-1(X-M)
where M and C are the mean and covariance of the footprint model
for the corresponding Access Point.
[0324] As noted earlier, for cases where a mix of different types
of measurements are available, certain embodiments employ a
framework where individual measurements contribute a cost and the
sum of all costs is minimized to choose the optimal answer.
[0325] In cases where a diversity of measurement types is
available, additional benefits may be obtained in some cases. The
measurements reported by a device will typically span some interval
in time, perhaps as much as a few minutes. In this case the
possibility has to be considered that the device may have moved
some distance over this period. A location calculation which
combines all measurements into a single location calculation will
yield significant errors if in fact the measurements span some
significant physical distance. On the other hand if it can be
determined that all the measurements correspond to the same
physical location, then the greater number of measurements
(assuming independent random errors) will on average yield a more
accurate result.
[0326] The availability of short range fixed wireless measurements
in such cases can enable such a determination. If the multiple
measurements corresponding to the same Wi-Fi AP are reported across
the entire time interval then the likelihood is that the device has
remained in virtually the same location (within the same building
at least). This benefit can accrue even if the Wi-Fi AP is
currently not present in the radio network database. Therefore in
such cases, all the measurements can be treated as corresponding to
the same location and a combined solution calculated.
[0327] In the event that it cannot be concluded that the device
remained in the same location for the duration of the measurements,
one solution is to use only the most recent set of measurements,
perhaps allowing a window of 30 seconds.
[0328] Having selected the measurements to be used, an optimization
technique is used to determine the location at which the combined
costs reach a minimum. Such techniques are well known in the art.
The data processed in this context are likely to yield irregular
manifolds over x & y, meaning that techniques to avoid false
minima may be required to ensure the globally optimum solution is
found.
[0329] In certain embodiments the transmitter models may be used to
estimate the location of a device based on a set of one or
measurements of wireless signals from one or more wireless
transmitters made by that device. At a minimum each measurement
will typically contain an identifier for the signal measured. The
measurement may also contain the received signal level. The
measurement may also include time information. The time information
may be an absolute timestamp. The time information may be an
indication of the elapsed time between the measurement and another
measurement. For example the such another measurement would be the
most recent measurement. The indication may be elapsed time in
seconds. The indication may be elapsed time measured in a more
abstract measure such as clock ticks where a clock tick is an
amount of time known to those elements of the system that are
measuring or processing time related information.
[0330] The measurements may be in the same form as the observations
used for network learning. The measurements may be in a different
form from which the data for location estimation can be
extracted.
[0331] In certain embodiments an exemplary transmitter model uses a
2D Gaussian PDF. The use of such a model enables the location
estimates and associated covariance to be calculated using a
closed-form equation which is more efficient than the iterative
numerical techniques required to minimise a cost function. The
estimated location for a set of wireless signal measurements is
denoted X-hat with the corresponding measurement covariance denoted
C-hat. Xt(i) and Ct(i) denote the transmitter footprint model
centroid and covariance of the transmitter measured in the ith
measurement. The estimated location is calculated by
X ^ = ( i C t ( i ) - 1 ) - 1 i ( C t ( i ) - 1 X t ( i ) ) '
##EQU00013##
[0332] The uncertainty of the location estimate is calculated
as:
C ^ = ( i C t ( i ) - 1 ) - 1 ##EQU00014##
[0333] In certain instances, a report may comprise one or more
wireless transmitter observations as well as location information
from a GPS. In certain embodiments the location estimate used for
the reference measurement is derived by combining the GPS position
information with the constraints obtained from the footprints
associated with the reported wireless transmitters. To align the
GPS information with the 2D Gaussian PDF corresponding to the
transmitters, the GPS measurement can be modelled as a Gaussian PDF
by setting the mean of the distribution to the GPS location and
deriving the covariance from the associated uncertainty. This then
enables the estimation equations above to be applied to combine the
GPS location with the transmitter footprints to derive an overall
location estimate. In certain alternative embodiments the GPS
location used for the location estimate may be the associated
location estimate from the observation when it is present and an
estimate derived from the transmitter footprints only when there is
no location estimate in the observation. This has the advantage of
minimising the computation where the uncertainty of the observed
location is much smaller than the transmitter footprint. This
advantage is of particular interest when a GPS location is present
in the observation along with cellular measurements. In certain
instances the presence of GPS position information plus one or more
transmitter observations will see the resulting location estimate
dominated by the GPS information due to the relatively smaller
uncertainty associated with the GPS location compared to the
footprint associated with the reported transmitters. In some
instances however, particularly in dense urban areas, the GPS
uncertainty may be relatively much larger, of the order of 200
metres or more while the footprint of one or more reported
transmitters may be considerably smaller, say a few tens of metres,
in which case the transmitter observations contribute substantially
to the accuracy of the resulting location estimate.
[0334] In certain embodiments the measurements are made at
intervals over a period of time. Such an example is a sequence of
cellular measurements as cellular receivers typically only return
the identity of one cell for any given measurement. The accuracy of
the location estimate derived from such a sequence is, on average,
better than the estimate resulting from a single measurement. There
is a problem, however, that the device making the measurement may
have moved whilst the sequence was being gathered. Processing such
a sequence of measurements may result in the accuracy of the
location estimate of the device's current location being reduced
due to the effect of the movement.
[0335] In certain embodiments the location estimation process
compensates for possible movement of the device during the
measurement process by weighting the measurements in the sequence
according to their age. The older a measurement is relative to the
most recent measurement, the less weight it receives in the
location estimation process. Where the transmitter model is a PDF
the weighting may take the form of decreasing the cost assigned to
a measurement by applying a multiplier to the cost. The older the
measurement the smaller the multiplier. An exemplary form of the
cost multiplier is
1 1 + .DELTA. T n ##EQU00015##
where k and n are constants and AT is the age of the measurement
relative to an epoch or an event. An example epoch is the current
time. An example event is the time of the most recent measurement.
Typical values for k where AT is measured in seconds are between 0
and 1. Typical values for n are between 1 and 5. Where AT is
measured in ticks the factor would need to be scaled based on the
number of seconds per tick.
[0336] In certain embodiments where the transmitter footprint model
is a 2D Gaussian PDF an exemplary means for determining the weights
is to apply a multiplier to the transmitter footprint covariance.
The older the measurement the greater the multiplier. An exemplary
means for estimating the location and uncertainty where the
measurements are to be weighted according to age.
X ^ = ( i [ ( 1 + k .DELTA. T ( i ) 2 ) C t ( i ) ] - 1 ) - 1 i ( [
( 1 + k .DELTA. T ( i ) 2 ) C t ( i ) ] - 1 X t ( i ) ) '
##EQU00016##
[0337] The uncertainty of the location estimate is calculated
as:
C ^ = ( i [ ( 1 + k .DELTA. T ( i ) 2 ) C t ( i ) ] - 1 ) - 1
##EQU00017##
[0338] Where k is as defined above and .DELTA.T(i) is the age of
the ith measurement. Typical values for k are as described
above.
[0339] In certain embodiments the system makes a determination as
to whether the measurement device was moving or not during the
period over which the measurements were made. Means for making this
determination, for example a Wi-Fi Access Point common to two or
more observations, are described elsewhere in this disclosure. In
certain embodiments where this determination is made it may be used
to control the application the measurement weighting. Where the
device has been determined not to have moved significantly then all
the measurements may be treated equally. This is equivalent to
setting the cost multiplier or covariance multiplier to unity which
in certain embodiments means setting k to 0. Doing so has the
advantage of increasing, on average, the accuracy of the location
estimate.
[0340] FIG. 14 illustrates a location estimate derived from a set
of measurements referencing Cell IDs 28652, 60383, and 20701 where
the transmitter model is a 2G Gaussian PDF. The footprint centroid
for each of these cells is shown as an asterisk. The covariance of
these footprints is the ellipse encircling the centroid. The
location estimate is shown by a triangle and the associated
uncertainty is the covariance matrix indicated by the encircling
ellipse. The effect of the footprint covariance effectively
weighting each observation is evident in that the location estimate
is closest to 20701 which has the smallest footprint. The benefit
to accuracy of using multiple measurement is evident as the
covariance of the location estimate is smaller than all three
footprints. Had the location estimate been based on a single
measurement then the covariance of the location estimate would be
the same as that of the footprint of the measured cell.
[0341] FIG. 15 illustrates the same scenario as FIG. 14 but with a
measurement of Wi-Fi access point 124013255425 in addition to the
three cells. The Wi-Fi access point is to the east of cell 17621
with the cell footprint illustrated by the asterisk and encircling
ellipse showing the footprint covariance. The estimated location
combining the three cells and one Wi-Fi measurement is indicated by
the triangle and the associated uncertainty by the encircling
ellipse. Compared to the estimate from FIG. 14 the uncertainty is
much smaller due to the influence of the Wi-Fi access point which
has a much smaller footprint covariance.
[0342] FIG. 31 illustrates the same scenario as FIG. 29 but this
time the location estimate is adjusting for possible movement
during the period over which the measurement were made. The
measurement for cell 21701 is the most recent. The measurement for
cell 28652 is the least recent. The measurements were taken 30
seconds apart. The value of n was 1.5. The value of k was 0.0061
with the result that the term k=1. Compared to the example where
there is no compensation for movement and as expected, the
resulting location estimate is closer to the most recent cell,
21701, by 344 m. Similarly the uncertainty associated with the
estimate is larger by approximately 16%.
[0343] For location requests originating from other than the
immediate user of the device (e.g., a network originated location
request), it is typical for the location client to indicate the
desired Quality Of Service (QOS). In such cases, QOS can be
expressed both in terms of the desired location precision as well
as cost to compute and latency. In certain embodiments where the
client software can assess the capabilities of the device as well
as the current conditions, it is possible to map the requested QOS
to a particular location method or even a combination of
methods.
[0344] Consider a scenario where a user requests a location,
expressing only that high precision is required (i.e. no
constraints on latency). The client software would attempt to
obtain a fix from the GPS, if the device includes a GPS capability.
In the absence of a GPS capability, the client would encode the
maximum number of measurements possible for the available
communication channel and forward them to the location server for a
wireless network orientated solution calculation. In another
scenario, if the user indicates that low latency is paramount, the
client would check whether a recent GPS fix was available in the
measurement buffer. If such a fix was available, the client
software would forward it to the location server. If no recent GPS
solution was available, the time required to start the GPS and
obtain a fix would most likely be too great to satisfy the
requested QOS so instead the client software would refresh the
radio measurements and then transmit the measurements to the
location server to facilitate a wireless network orientated
location estimate. In yet other cases, it may be advantageous to
provide an estimate with minimum latency, and then as soon as
practicable after that, a high precision estimate. In such cases
the client would immediately send to the location server the
wireless measurements (and any stored GPS fixes) and then activate
the GPS and await a position fix. Once a GPS fix was achieved, this
would be forwarded as a secondary response to the location request.
This type of initial & deferred response can significantly
enhance user experience for instance in loading a map and marking
the user's location on it. With the coarse immediate fix, a map of
the immediate surrounds can be provided. The relative uncertainty
in the position can be reflected appropriately. Subsequently, when
a more precise solution becomes available, the display can be
updated to reflect this.
[0345] For end user applications on a mobile device, the UI in the
application may also provide configuration options enabling the
user to select between GPS based and wireless based or even a
combination.
[0346] One advantage of certain embodiments of the present
disclosure may be that it can enable location based services to be
offered on devices which have only unlicensed wireless network
adapters. An example is the Nokia N800. Because in certain
embodiments the network database server can accumulate location
information for 802.11 networks, obtaining location reference
information based on any combination of GPS, cellular or other
802.11 nodes means that a client integrated in the limited
connectivity device can still make a request from a location
server, supplying information on the nearby 802.11 APs and have its
location estimated.
[0347] In a network where there are a proportion of cells that
remain unknown to the location server, there is a proportional
chance that a subscriber sending a location request with the
current serving cell will fail to be served because the location
server has no information about that cell. In some cases even
reporting the two most recent or three most recent cells may not
avoid this if all of these recent cells are unknown. Research
indicates that this is one of the greatest sources of consumer
dissatisfaction with a positioning service (i.e. failure to obtain
a location fix).
[0348] A method to dramatically reduce such cases is for the mobile
client to maintain a history of serving cell and as each new cell
is entered in the accumulator to check from the tasking information
whether that cell is known to the location server. In the normal
course of purging old measurements from the accumulator, the client
never purges a cell if it is the most recent cell that is known to
the server. The most recent known cell is maintained along with the
elapsed time since the last measurement corresponding to that cell.
In the event of a location request, where several recent cells are
unknown, the presence of the most recent known cell in the report
ensures that the location server can deliver a location estimate,
with the position uncertainty adjusted to reflect the elapsed time
since that measurement.
[0349] In some cases where an unknown cell is reported, the system
can check to see whether that cell might be an element of a
sectorised site.
[0350] This check can be done by checking for a known site which
has only two known sectors having cell IDs that are within +/-1 of
the unknown cell ID.
[0351] If a candidate site is found, a validation step can be
carried out if other measurements of known cells were also
reported. Using only the known cells, a location estimate can be
computed. This can be checked against the hypothesized location and
orientation of the unknown cell. If a match is indicated then the
unknown cell can be assigned that location and the additional
information incorporated in a revised location estimate.
[0352] In some cases, if an unknown cell is reported in a location
request and there is no additional information from which to derive
an estimate, the system can check whether the LAC is known. If so,
the system can return a location representing the centre of all
cells having that LAC and indicate the much larger positional
uncertainty. In many cases, this will at least enable the user to
obtain a map encompassing the location and other techniques can be
used to refine the position, for instance a secondary search based
on street name.
[0353] In such cases, the UT for the system may enable the user
after determining his location more accurately to highlight this
location to the system for instance by clicking on the map in order
to provide the system with a reading for this cell to initialize
the acquisition process for that cell.
[0354] In cases where one or more cell identities featured in a
location request correspond to installations where there are
multiple transmission points (donor/repeater(s)), then it is
preferable for the location server to attempt to resolve the
measurements to a particular transmission element. If other
measurements are reported then the system attempts to select the
element which is most consistent with the other reported
measurements.
[0355] In the event that there are no other measurements available,
the system can default to the element having the largest coverage
area (being the most likely assuming uniform spatial distribution
of subscribers). The positional uncertainty reported in this case
can be increased to reflect the ambiguity in the location.
[0356] Certain embodiments employ a variety of techniques to model
the propagation of wireless signals. In many cases, such models can
be enhanced with information on the terrain and clutter. If such
information is available, it can be incorporated in embodiments of
the present disclosure to obtain better performance. For example,
terrain can be useful for cellular modelling and clutter can be
useful when modelling building density for determining shadow
fading sigma and path loss decay.
[0357] In certain embodiments, methods, systems, and/or
processor-readable media may be used to enable and/or to improve
location based services offered by a mobile network operator. This
may be useful if the radio network configuration database
maintained by the operator contains an unacceptable number of
omissions or other errors, which in turn degrade services that
depend on the radio network configuration information. Examples
include the various location based services (LBS) which are may be
offered by operators including local search, friend and family
locator. What is unacceptable in terms of omissions or others
errors and/or degradation of service will vary from network
operator to network operator depending on a number of factors,
including the specific key performance indicators (KPIs) for the
service being offered. For example, a local search service may be
able to tolerate a network database in which 1%, 3%, 5%, 7% or 10%
of the cells in the network are missing. By contrast a home zone
service for which a 95%, 97%, 99% or 99.5% in zone reliability is
promised, typically cannot. In some cases the network operator may
be unable due to practical or commercial reasons to address these
issues in their database. Certain embodiments of the present
disclosure may be employed to provide a more complete and/or more
accurate network database for provisioning on the zone server. By
more complete in the present context we mean that more of the cells
actually present in the wireless network have corresponding entries
in the network database. Similarly, by more accurate in the present
context we mean that the average deviation between the true values
of certain parameters in the database and their real values in the
deployed network is smaller.
[0358] In cases where the system is already modelling the network
of interest using certain disclosed embodiments, it may simply be a
case of exporting the database and providing it to the zone server.
In other cases a system according to certain embodiments of the
present disclosure may be deployed and provided with measurements
collected in a focused survey measurement campaign. The resulting
database may then be provided to the zone server. In yet other
cases, where a substantial proportion of the operator provided
database is known to be sufficiently accurate, it may be desirable
to use this data but augment it with the network model information
held by a system according to certain disclosed embodiments. In
this case, the operator provided database can be used to export a
collection of reference measurements as described in the present
application. These measurements may then be combined with reference
measurements held in the network learning server and used to obtain
an updated network model on the basis of both sets of reference
measurements. This updated model may then be exported and provided
to the zone server. In this way a variety of issues such as missing
transmitters, transmitters with incorrect locations either due to
data entry errors or recent relocation or even transmitters having
incorrect azimuth information associated with their antenna may be
ameliorated, substantially ameliorated, and/or sufficiently
ameliorated to give acceptable results.
[0359] In certain cases, instead of exporting a complete database,
a comparison may be carried out between the network model held by
the network learning server and a database provided by the
operator. The differences may be provided to the operator to enable
their database to be corrected.
[0360] The following paragraphs deal with the architectural aspects
of certain embodiments of the present disclosure. Certain
embodiments accumulate information on the configuration of one or
more wireless networks using measurements from one or more wireless
devices. The wireless networks may be fixed or mobile and may
comprise anywhere from one to many thousands or hundreds of
thousands of transmitters. In turn, the wireless devices may number
anywhere from one to many millions of devices. Certain embodiments
also provide support for location determination as the devices roam
between different networks.
[0361] FIG. 32 illustrates an exemplary configuration of a system
to support location based services in connection with a single
mobile wireless network. Note that in this case, it is the
acquisition of the wireless network configuration information that
is the focus of the discussion, therefore some aspects of the
actual location services aspects including for instance the
interconnection to location clients or other application servers
are not shown.
[0362] In this exemplary embodiment, communication between the user
device and the server is via one or more of SMS, packet (UDP/IP or
TCP/IP) (over GPRS, EDGE, HSPA, or EV-DO) or even CS data. Indeed
different bearers may be used for different exchanges. The main
exchanges and their flow are described in the following paragraphs.
Note that although the connection is shown as direct from the
wireless network to the gateway, this need not be the case. In one
scenario, a direct connection could be possible, for instance if
the gateway is provisioned with an account on the SMSC or directly
connected to the GGSN. On the other hand the connection could be
via the public internet. For SMS connectivity, this could be
through an SMSC aggregator. For packet data exchanges it could be
via the internet, in other words the client relies on a suitable
APN to be configured on the device that permits access to the
internet and the gateway.
[0363] Tasking information from the network database server is sent
to the client application on the wireless mobile device. This can
be initiated either by a push from the network side if the tasking
information has been updated or a pull from the client side if the
information needs to be refreshed, based for instance on a time
elapsed since last refresh or an indication from the network side
that updated information is available.
[0364] A network report is sent from the client to the network
database server based on triggering criteria at the client
including optionally the availability of a connection to the
network. The report could be sent via SMS or packet, directed to
the gateway. The gateway forwards the report to the network
database server for processing.
[0365] A location measurement request may be sent from the location
server to the client if triggered by a location request from an
authorized location client. In this case the location request is
shown as being transmitted via the gateway however in some cases
the location server may have a direct connection to a suitable
channel, without needed to send via the location gateway.
[0366] A location measurement response is sent from the client to
the location server either in response to a request from the server
or possibly in response to an event within the client. This could
include a user request via the GUI, or an existing trigger such as
a periodic reporting regime. In this case the response is directed
to the gateway. While the ultimate destination for the response is
the location server, to support a location calculation, the gateway
forwards one copy to the location server as would be done in a
conventional location services network configuration, and another
copy to the network database server for use in the network model
maintenance. A further role for the gateway in this context is a
security role. It is likely that a desirable aspect of achieving
user acceptance of this kind of service and the use of radio
measurements from their devices is a strong assurance that
information reflecting their location and movement patterns are not
accessed by unauthorized parties. The network acquisition process
described in this application does not require any user
identification information. Therefore in forwarding measurement
information to the network database server, the gateway discards
any user identification information.
[0367] In exemplary FIGS. 15, 16, and 17 the gateway, network
database server and location server are shown as separate entities.
It should be understood that this is a logical decomposition only.
Optionally the physical configuration could correspond to this
however in some cases two or more of these components could be run
on a single host. Equally a distributed implementation could be
employed with multiple hosts sharing each function for enhanced
resilience and scaling as is well known in computer networking. In
some cases the gateway and network database server could be
deployed in conjunction with an existing location server product.
Such an existing location server in carrier sponsored deployments
would typically have a facility to receive a database describing
the configuration of the radio network, such database being
provided by the radio network management function within the
carrier organization. Certain embodiments of the present disclosure
enable such a product to be deployed in a carrier independent
fashion with the network database server being substituted in the
role previously carried out by the carrier radio network planning
department. In some cases a reformatting step may be required
between the network database server and the location server to
accommodate the format required by the existing location server.
The interfaces between the different entities will depend on the
physical distribution and may include any inter process
communication mechanism if deployed on a single host or any
suitable network communication such as via FTP or HTTP or other
application protocol over a LAN. Yet another possibility is via a
database through shared table space.
[0368] FIG. 33 shows an exemplary extension from the single mobile
wireless network shown in the previous section, in which a dual
mode device is operated by the user. In this case the user can
connect the device either through the cellular network and/or a
fixed wireless network. The fixed wireless network in this case
could be, for example, a public Wi-Fi AP or alternatively a private
Wi-Fi AP integrated with the user's fixed broadband router at
home.
[0369] The network database server in this case can be configured
to maintain information on the fixed wireless network access points
as well as the cellular network base stations, depending on the
requirements of the applications that will use the location
determining capabilities of the location server. In some cases it
may only be desirable to determine location while the device is
connecting through the mobile wireless network. For the purposes of
this explanation we will assume that both the cellular and the
fixed networks are to be monitored.
[0370] In this case the interactions between the servers and the
client application are essentially the same as described
previously. The only differences are that the measurements
collected by the client and reported to the server may from time to
time include measurements pertaining to the fixed wireless network
(SSID, MAC address & RSSI) and the reports may be delivered to
the gateway via either the mobile wireless network or via a fixed
wireless network connection and the internet. For network
originated messages to the server, the gateway will typically use
the cellular network since the destination address (MSISDN) remains
fixed regardless of where the device is located.
[0371] FIG. 33 shows the network database server maintaining models
for both the mobile wireless network as well as the fixed wireless
network. Note that although the fixed wireless access points
accessed by the device may include access points that are either
privately owned or operated by one or more access providers, they
are modeled in the network database server as part of a single
Wi-Fi network. The fact that they are operated by different
providers cannot be determined from the radio information reported
by the device. In any case, the affiliation of the access point is
of no consequence in determining the location of the access point
and in turn using that information to determine the location of
devices accessing it.
[0372] FIG. 34 shows an exemplary architecture for supporting
location services across multiple networks without the respective
carriers providing network databases. As before the illustration
shows direct connection between the gateway and each wireless
network however this could also be a connection via the
internet.
[0373] A representative wireless mobile device is shown attached to
each wireless network. The client running on each device makes
measurements of its respective network with or without reference to
GPS. These measurements are conveyed over wireless bearer at
suitable times to the gateway. The gateway forwards messages
containing measurements to the network database servers. Note that
while separate servers are indicated, this is a logical separation.
Capacity permitting, multiple networks could be modeled on a single
host. In theory all networks could be modeled in a single composite
collection where each cell is distinguished by its cell global
identity (MCC+MNC+LAC+cell ID). In practice several benefits accrue
from maintaining each network separately.
[0374] The update process between the copy of the network model
maintained by the network database server and the location
server(s) can be performed independently for each network,
triggered at suitable times or based on certain events such as a
threshold number of changes or new cells being detected.
[0375] In this case a single location server is indicated using
multiple network models. In practice, there could be a dedicated
location server coupled to each individual network model.
[0376] The illustration shows a single gateway. In practice,
multiple gateways could be configured. This could be necessitated
by capacity requirements as the number of network models being
maintained grew large. Optionally a separate gateway could be
provisioned for each country. Further alternative segmentations are
possible.
[0377] The configuration in this figure includes only mobile
wireless (cellular) networks. As described in the earlier section
however it is possible to also accommodate one or more fixed
wireless network segments.
[0378] An exemplary alternative architecture illustrated in FIG. 35
could be used in which one or more operators choose to participate
in the service and provide their radio network configuration
information. In this case, for clarity, only cellular networks are
shown however the configuration could be extended to support fixed
wireless networks as well as shown previously.
[0379] A significant limitation in existing Location Based Services
(LBS) as standardized by 3GPP or OMA is in the roaming scenario. As
a general matter, the consumption of location based services by an
individual is greater when traveling than when in familiar
surroundings. Currently however because of the dependence on radio
network configuration information etc. Location Services are often
network specific. This means that when subscribers travel and their
cell phones roam on other networks, location information is no
longer available unless the roamed network supports the services.
While the standards do provide extensive support for roaming
location services, commercial issues including market fragmentation
mean that there are very few cases where roaming cellular
subscribers can utilize location services.
[0380] Certain embodiments of the present disclosure address this
because the location calculation can be entirely independent of the
underlying network (other than a requirement to accumulate
measurements pertaining to that network in order to develop a model
of its configuration). Because the identity of the network may be
conveyed in the measurements to the location server, users will be
able to roam completely seamlessly across networks receiving
consistent service.
[0381] FIG. 36 and FIG. 37 illustrate two exemplary architectures
for the client software. These figures focus on the network
acquisition aspect of the application. The remaining application
functionality could involve mobile mapping or instant messaging or
any other functionality. Indeed the network acquisition related
functionality described here could be implemented in the form of a
library, suitable for integration by third party application
developers into their applications thereby accessing the benefits
of widely accessible location.
[0382] The network acquisition aspect monitors the wireless
network(s), detects any measurements that could be useful in
maintaining the network model(s), and transfers these to the
server. In FIG. 36, the application features a single component,
(in some implementations corresponding to a single operating system
process). This process makes use of various system interfaces to
obtain information and services. In the figure the Symbian
CTelephony interface is used as an example of the interface for
accessing radio network measurements. By invoking the applicable
methods the application can obtain the following information for
the current serving cell: MCC, MNC, LAC, cell ID & RSSI. The
application also can use other application programming interfaces
(APIs) to retrieve GPS location information (if supported on the
specific device). Tasking information may be provided from the
network database server to the application. Such information is
stored for use in identifying patterns of measurements to report to
the server.
[0383] In certain embodiments, this application runs only when
launched by the user until the user exits the application, in which
case the measurement processing occurs only during this period. In
alternative embodiments, the user may be able to select whether
they want the application to execute automatically. Once the user
has launched the application, measurement collection may commence
regardless of whether the user is currently executing any actions
that require radio network measurements. In alternative
embodiments, the user may be able to select how and/or when the
measurement collection commences. The measurement collection may be
done periodically, triggered by a timer. The interval could be any
suitable value between a few seconds or several tens of seconds.
The lower limit on the interval is the minimum interval at which
one could expect the terminal to reselect serving cell. If a
pattern of measurements is observed which matches some criteria in
the tasking information, a measurement report is created and stored
in the network reports. A separate component of the application is
responsible for monitoring for the availability of a suitable
communication link to send network reports to the server.
[0384] FIG. 37 shows an alternative architecture in which there are
two separate components, typically implemented as two separate
processes. One is a background process which is launched when the
application is first installed and activated. The other is the main
component of the application that provides the UI. Both components
would typically be part of a single install file. The interface
between the two components can be via any suitable Inter-Process
Communication (IPC) mechanism supported by the particular device
platform, including sockets. Several data stores are shown in the
figure. It is uncommon for different processes to be able to access
the same shared data. Therefore in the present context the data
store could be maintained by one or other process and necessary
information passed to the other process on request via IPC. To
illustrate, the tasking information would typically be received by
the UI component which handles user initiated connections with the
network. The information would then be transferred to the
background component which makes more frequent use of it, and save
by that process in private storage.
[0385] The background process is responsible for periodically
collecting radio measurements, updating the measurement filter and
also analyzing the measurements for any patterns that match
criteria in the tasking information. In the event that a pattern is
found, a network report is prepared and saved for subsequent
reporting to the network database server. The background process
also reads GPS measurements from the GPS API (on supported devices)
whenever available. In some platforms, the GPS API supports a
register/notify pattern, in this case enabling the background
process to register to be alerted when a GPS fix is obtained or
when the GPS position changes substantially from a previous
fix.
[0386] The main component is responsible for the main application
functionality. Typically this will include some form of location
based functionality. When the user activates this functionality
(for instance requesting a map of the current location from an
application server), the main process triggers a location request
to the server along with the rest of the information pertaining to
the user's request. To send a location request the main process
retrieves a measurement report from the measurement filter (i.e.,
via IPC with the background process) and then conveys this to the
server via the current preferred connection (for instance via HTTP
as a MIME attachment or via raw TCP/IP over Wi-Fi, GPRS, EDGE,
HSPA, EV-DO or even SMS). The remainder of the processing on the
application server to pass the measurement information to a
location server, plus satisfy the user request is not covered here
as the focus is on the architectural aspects of the client
application, in particular as it relates to wireless network
database acquisition.
[0387] In an alternative situation, the background process might
also directly access the communications services supported by the
device in order to communicate with the server. In general for
mobile devices, where data exchanges incur a charge as well as
significantly draw on the battery, applications designers prefer to
leave it to the user to initiate connections.
[0388] The following paragraphs describe the platform facilities
that a Window Embedded CE client could use to implement the
processing described above. To obtain the cellular information, the
client can use the GetCellTowerInfo function supported by the Radio
Interface Layer dll. Among the information available by using this
function are the MCC, MNC, LAC & cell ID and rxLev.
[0389] GPS location information (on devices which include GPS
hardware) the client can read the GPS using the Location Framework
by calling LocationGetReport( ). The client would specify the
LOCATION_LATLONG_GUID report type and pass an appropriately large
value for the maximumAge. This could for instance correspond to the
interval at which the client measurement filter is being refreshed.
By requesting a report without first calling
LocationRegisterForReport( ) the client application can avoid
powering on the GPS and depleting the battery, but can also take
advantage of fixes that are available when the user or another
application has activated the GPS.
[0390] For reading Wi-Fi information, once again the Location
Framework can be used. The client application can invoke a
LocationGetReport( ) request, specifying the
LOCATION.sub.--802.sub.--11_GUID report type and passing a
relatively large maximum age as above for the lat/long request. As
with the GPS request, the application can take advantage of Wi-Fi
information if it is available without having to power on the
802.11 network hardware and consume power in a scan. Other details
about initializing libraries etc would be easily determined by one
familiar with Windows CE programming, using the reference
documentation.
[0391] The client implementation on other mobile device platforms
uses equivalent facilities supported by those platforms as would be
understood by one familiar with mobile applications
development.
[0392] The continual improvement in storage, processing and display
capabilities of mobile connected devices means that certain
embodiments of the present disclosure can be extended to provide
on-device location determination capabilities. In some cases,
devices already incorporate GPS hardware providing local
positioning facilities. For non-GPS equipped devices however, it is
usually necessary to connect to a location server in the network to
obtain a position calculation. The same is also the case if the
device is currently indoors or in other locations where the GPS
cannot achieve a position fix.
[0393] For many applications this requirement to complete an
exchange with a network based server to obtain a single fix is
cumbersome. Consider for instance a mapping application which if
GPS is available updates automatically as the user moves. If the
same application is used on a device without a GPS, the user has to
continually initiate an exchange with the network to obtain a
position update.
[0394] In such cases it would be preferable if a position fix could
be computed immediately on the device, where required this could be
refreshed rapidly. In certain embodiments information may be stored
locally on the device to compute location estimates locally using
wireless signal measurements.
[0395] This can be achieved by transferring "tiles" of information
representing the configuration of the wireless network(s) being
used by the device into storage on the device for local usage.
Techniques for managing collections of tiles of spatially indexed
information are well known in the field being used for instance by
mobile applications which display maps obtained in small segments
from a repository in the network.
[0396] In the present case, after obtaining an initial location
estimate the system can provide one or more tiles of information
describing the configuration of the cellular network currently used
by the device along with any other fixed network elements in the
area which might be used, for instance Wi-Fi APs etc. This
information can then be used locally with the location estimation
algorithms described earlier. In densely populated areas, the
density of Wi-Fi APs is likely to be considerably greater than the
cellular networks. It is preferable therefore to manage the
cellular and Wi-Fi network information as independent "layers",
updating the Wi-Fi layer on a finer scale.
[0397] In some cases the system may provide a facility for users to
download the radio network configuration for an entire area and
store it in device memory to enable offline operation without
continually needing to access the network while moving. This
mirrors the facility which is often available with mapping
applications to store the information covering a region for faster,
disconnected operation and for applications using the former to
support the latter will enhance the user experience.
[0398] In previous sections, techniques were described for
estimating the location of a wireless device based on one or more
measurements pertaining to a wireless network transmitter. For
applications where the location is to be calculated on the device
to enable greater responsiveness or avoid the cost or latency of
exchanges with a network based server, sequences of individual
location estimates can be applied to a Kalman filter to smooth the
random errors in the individual measurements and provide a more
consistent indication of the user's position.
[0399] This Kalman filter can be aided by constraints derived from
the measurements. For instance if a Wi-Fi AP is being measured but
is not present in the database and therefore cannot be exploited in
the location estimate, if it is continually being received by the
device then it provides a very strong indication that the device is
either stationary or moving within a very limited range. Applying
this velocity constraint to the Kalman filter can increase the
suppression of the random variations that could otherwise result
from the temporal variations typically observed in cellular radio
measurements and the dependent location estimates and also increase
the effective averaging interval or time constant.
[0400] In a system where a large number of measurements from user
devices are being accumulated, it may be advantageous to implement
quality control capabilities to protect against either unintended
or malicious degradation of the system by the transmission of false
information. One example in the present context would be a hacker
implementing a fake client which transmitted a stream of reports
combining cellular or fixed wireless access point identifiers with
incorrect GPS coordinates. Without suitable quality control
measures, this stream of reports would corrupt the database,
degrading or even disabling service for legitimate users.
[0401] Measures for protecting against this type of attack include
authentication between the client and the server and use of secure
transport to preserve the contents of the messages from
modification in transit.
[0402] Another type of attack would be a flood of either network
reports or location requests. Certain embodiments of the present
disclosure provide a collection of methods for protecting against
such denial of service (DOS) attacks. Certain embodiments also
include a fallback where in the case of a DOS attack the system can
discard requests that don't originate from cellular networks (this
may be effective because it is unlikely that users seeking to
maliciously introduce erroneous data into the system would use
cellular data with the associated charges to implement such an
attack).
[0403] As the penetration of clients grows, the server may
accumulate a substantial volume of data on the coverage and
performance of the wireless networks used by the client devices.
This can be used to detect network issues such as black spots where
there is frequent reselection between cells or between RATs.
Locations where dual-mode cellular terminals frequently reselect
between GSM & UMTS are a common source of irritation for
cellular subscribers. Certain embodiments of the present disclosure
also can provide the ability to directly measure the footprint of
cells. These measurements can be provided to radio network
engineers to compare with the footprints predicted by their
planning tools in the form of best server polygons etc, in turn
enabling the tools to be updated to more accurately reflect the
real coverage of the cells. Cell overshoot is a common performance
problem that can be detected in this way. A further capability of
certain embodiments is the ability to measure the effect of network
adjustments over time. Data collected before and after an
adjustment can be compared to determine whether the change had the
desired effect. Using the example of a cell having excessive
overshoot in a particular direction, data collected after an
increase in the tilt was applied can be compared to the data
recorded beforehand to assess the degree to which the adjustment
was successful in addressing the problem.
[0404] In addition to the performance and footprint of the network,
certain embodiments can also collect information on the spatial
patterns of usage. If the server is integrated with application
servers supporting VOIP, IM, browsing etc., then those servers can
accumulate statistics on the location distribution of traffic by
obtaining location information for archiving along with traffic
information. This information can be made available to the access
provider for use in their network planning and optimization. The
data can also be segmented by service type enabling more accurate
planning, taking into account the different requirements of these
different services, for instance the lower latency requirements of
VOIP compared to browsing.
[0405] Another scenario where such information can be useful is for
operators establishing a new network, for instance after the award
of a new radio frequency spectrum license. A common challenge in
such cases is to determine the capacity required and its spatial
distribution. Assuming that the operator of a server in accordance
with certain embodiments of the present disclosure is not
affiliated with the current cellular access providers, spatial,
traffic density data for existing networks could be provided to the
new entrant to guide their network planning activities.
[0406] This type of spatial traffic distribution data could also be
used by different access providers to determine where users are
favoring alternative access providers over their own network. For
instance a cellular carrier could obtain data showing the relative
proportions of VOIP calls being placed on their network to the
proportion being placed on competing access provider networks, for
instance Wi-Fi hotspots. This type of data can enable an access
provider to characterize their competition and develop new access
product offerings tailored to the usage patterns of
subscribers.
[0407] Certain embodiments described herein may have one or more of
the following advantages.
[0408] One advantage of certain embodiments is that wireless
network configuration data can be aggregated independently of the
MNOs.
[0409] Another advantage of certain embodiments is being able to
use data from non-GPS enabled wireless devices to further refine
the configuration data.
[0410] Another advantage of certain embodiments is that the
cellular network configuration data can be accumulated more rapidly
because of the focused collection enabled by the tasking data. This
is typically performed more quickly than a purely opportunistic
approach such as is used in other systems.
[0411] Another advantage of certain embodiments is that the Wi-Fi
network configuration data can be accumulated more rapidly because
of the focused collection enabled by the tasking data. This is
typically performed more quickly than a purely opportunistic
approach such as is used in other systems. Moreover, this can be
quicker, more accurate, and less expensive than relying on drive
tests to collect the information that any user device can provide
input.
[0412] Another advantage of certain embodiments is greater
coverage, for instance with indoor pico cells that would be
unlikely to be located by a GPS enabled handset because the GPS
coverage indoors would prevent a simultaneous measurement of the
cell and GPS fix.
[0413] Still a further advantage of certain embodiments is being
able to determine location for dual mode wireless devices whether
they are operating on a cellular network or have connected to a
network operating in the unlicensed spectrum such as a Wi-Fi
network.
[0414] Another advantage of certain embodiments is that the
wireless network configuration data can be updated quickly
responsive to changes in the wireless network.
[0415] Still a further advantage of certain embodiments is that a
mobile radio terminal can determine its own location without
relying on a location server.
[0416] Yet another advantage is that distributing tasking
information to a plurality of mobile radio terminals allows certain
embodiments to rapidly and selectively obtain information regarding
wireless network transmitters that have not been accurately
described in a wireless network database.
[0417] The disclosure has been described with reference to
particular embodiments. However, it will be readily apparent to
those skilled in the art that it is possible to embody the
disclosure in specific forms other than those of the embodiments
described above. The embodiments are merely illustrative and should
not be considered restrictive. The scope of the disclosure is given
by the appended claims, rather than the preceding description, and
all variations and equivalents which fall within the range of the
claims are intended to be embraced therein.
[0418] The reader's attention is directed to all papers and
documents which are filed concurrently with this specification and
which are open to public inspection with this specification, and
the contents of all such papers and documents are incorporated
herein by reference. All the features disclosed in this
specification (including any accompanying claims, abstract, and
drawings) may be replaced by alternative features serving the same,
equivalent or similar purpose, unless expressly stated otherwise.
Thus, unless expressly stated otherwise, each feature disclosed is
one example of a generic series of equivalent or similar
features.
[0419] It will be understood that the term "comprise" and any of
its derivatives (eg. comprises, comprising) as used in this
specification is to be taken to be inclusive of features to which
it refers, and is not meant to exclude the presence of any
additional features unless otherwise stated or implied.
[0420] The reference to any prior art in this specification is not,
and should not be taken as, an acknowledgement of any form of
suggestion that such prior art forms part of the common general
knowledge.
* * * * *
References