U.S. patent number 5,508,707 [Application Number 08/314,486] was granted by the patent office on 1996-04-16 for method for determining position by obtaining directional information from spatial division multiple access (sdma)-equipped and non-sdma-equipped base stations.
This patent grant is currently assigned to U S WEST Technologies, Inc.. Invention is credited to Frederick W. LeBlanc, Alparslan M. Uysal.
United States Patent |
5,508,707 |
LeBlanc , et al. |
April 16, 1996 |
Method for determining position by obtaining directional
information from spatial division multiple access (SDMA)-equipped
and non-SDMA-equipped base stations
Abstract
A method for determining position by obtaining directional
information from spatial division multiple access (SDMA)-equipped
and non-SDMA-equipped base stations. The method is directed for use
in a wireless communication system which includes a plurality of
base stations each having a corresponding coverage area. For each
of the base stations, a plurality of RF measurements are determined
in cooperation with a receiver, including a link budget of the base
station, for a predetermined plurality of distances and directions.
Determined RF measurements for each of the base stations are
modeled as a scaled contour shape having minimum and maximum
boundaries. Base stations which neighbor the mobile unit are
determined so as to define a first bounding polygon area by their
intersecting contours. The first bounding polygon area generally
describes the relative position of the mobile unit. A second
bounding polygon area is determined in accordance with the lobes of
neighboring base stations as described in terms of azimuth angles.
The intersection of the first and second bounding polygon areas are
determined so as to define a location polygon which more precisely
describes the position of the mobile unit in terms of minimum and
maximum error estimate.
Inventors: |
LeBlanc; Frederick W. (Arvada,
CO), Uysal; Alparslan M. (Boulder, CO) |
Assignee: |
U S WEST Technologies, Inc.
(Boulder, CO)
|
Family
ID: |
23220153 |
Appl.
No.: |
08/314,486 |
Filed: |
September 28, 1994 |
Current U.S.
Class: |
342/457;
455/456.1 |
Current CPC
Class: |
G01S
5/02 (20130101); H04W 64/00 (20130101) |
Current International
Class: |
G01S
5/02 (20060101); H04Q 7/38 (20060101); G01S
003/02 (); H04M 011/00 (); H04Q 007/00 (); H04B
001/00 () |
Field of
Search: |
;342/457,357 ;379/59,60
;455/33.1,33.4,56.1 |
References Cited
[Referenced By]
U.S. Patent Documents
Primary Examiner: Issing; Gregory C.
Attorney, Agent or Firm: Brooks & Kushman
Claims
What is claimed is:
1. For use in a wireless communication system including a plurality
of Spatial Division Multiple Access (SDMA)-equipped base stations
each having a smart antenna with a predetermined number of fixed
lobes, each of the lobes oriented in a predetermined direction and
operative to communicate with a mobile unit on a corresponding
communication channel within a known coverage area, a method for
determining the position of a mobile unit, comprising:
for each of the base stations, determining in cooperation with a
receiver, a plurality of RF measurements for the base station, for
a predetermined plurality of distances and directions;
for each of the base stations, modeling its determined RF
measurements as a scaled contour shape having minimum and maximum
boundaries;
determining where the contour shapes of the neighboring base
stations intersect so as to define a first bounding polygon area
that describes the relative position of the mobile unit;
determining which lobes of the neighboring base stations are in
communication with the mobile unit;
determining the orientations of the communicating lobes of the
neighboring base stations in terms of azimuth angles so as to
define a second bounding polygon area that describes the relative
position of the mobile unit; and
determining where the first and second bounding polygon areas
intersect so as to define a location polygon that describes the
position of the mobile unit in terms of minimum and maximum error
estimate.
2. For use in a wireless communication system including a plurality
of Spatial Division Multiple Access (SDMA)-equipped base stations
each having a smart antenna with a predetermined number of floating
lobes, each of the lobes operative to track and communicate with a
mobile unit on a corresponding communication channel by changing
its orientation within a predetermined direction range, a method
for determining the position of a mobile unit, comprising:
for each of the base stations, determining in cooperation with a
receiver, a plurality of RF measurements for the base station, for
a predetermined plurality of distances and directions;
for each of the base stations, modeling its determined RF
measurements as a scaled contour shape having minimum and maximum
boundaries;
determining where the contour shapes of the neighboring base
stations intersect so as to define a first bounding polygon area
that describes the relative position of the mobile unit;
determining which lobes of the neighboring base stations are in
communication with the mobile unit;
determining the orientation ranges of the communicating lobes of
the neighboring base stations in terms of azimuth angles so as to
define a second bounding polygon area that describes the relative
position of the mobile unit; and
determining where the first and second bounding polygon areas
intersect so as to define a location polygon that describes the
position of the mobile unit in terms of minimum and maximum error
estimate.
3. For use in a wireless communication system including a plurality
of Spatial Division Multiple Access (SDMA)-equipped and
non-SDMA-equipped base stations each having an antenna with a
corresponding coverage area, a method for determining the position
of a mobile unit, comprising:
for each of the non-SDMA-equipped base stations, providing a
plurality of bandpass filters in electrical communication with the
corresponding antenna, each of the band-pass filters having a
corresponding waveguide and operative to receive mobile unit
signals from predetermined directions in predetermined frequency
slots;
for each of the base stations, determining in cooperation with a
receiver, a plurality of RF measurements for the base station, for
a predetermined plurality of distances and directions;
for each of the base stations, modeling its determined RF
measurements as a scaled contour shape having minimum and maximum
boundaries;
determining where the contour shapes of the neighboring base
stations intersect so as to define a first bounding polygon area
that describes the relative position of the mobile unit;
determining the communication frequencies of the neighboring base
stations and the corresponding directions of their received mobile
unit signals so as to define a second bounding polygon area that
describes the relative position of the mobile unit; and
determining where the first and second bounding polygon areas
intersect so as to define a location polygon that describes the
position of the mobile unit in terms of minimum and maximum error
estimate.
Description
TECHNICAL FIELD
The present invention relates generally to positioning methods.
More particularly, the invention relates to a method for
determining the position of a mobile unit in a wireless
communication system or public land mobile telecommunication
system, including a plurality of Spatial Division Multiple Access
(SDMA)-equipped and non-SDMA-equipped base stations.
BACKGROUND ART
Most metropolitan areas are now equipped with one or more forms of
wireless communication networks which provide mobile telephone and
other related services to customers across a broad frequency
spectrum. Consider, for example, what has come to be known as
"cellular" telephone services or Personal Communication Services
"PCS", i.e., radio transmissions in the frequency band between
approximately 800 MHz and 2.2 GHz.
Public wireless communications services are removing the historical
mobility constraints on the users of public communications systems.
Continuing improvements in the capabilities and services offered by
these systems will be accompanied by increasingly broad public
acceptance and adoption. Relaxing the mobility constraints on
communications systems users is thus generally considered
desirable. For service providers who rely on precise location
information, however, such as for example, Emergency "911" (E-911)
service providers, this increased user mobility is a definite
disadvantage. Uncertainty concerning the user's location translates
directly into increased response times and therefore a reduced
quality of service when the caller is uncertain of, or cannot
describe, her location.
In wired commercial communications systems, precise information
concerning the location of the user's equipment, and hence that of
the user, is available. Wireless systems, on the other hand,
typically follow a cellular model in which there are wired
connections to base stations and then wireless links from each base
station to its active users. The service area for each base station
is roughly circular in shape with a radius of 5-20 km and overlaps
that of its neighboring base stations. This corresponds to a base
station service area in the range of roughly 80-1300 km.sup.2. In
some systems, the coverage area for each physical base station is
further divided into equiangular sectors, each operating as an
independent cell. Three 120 degree sectors is representative of
this type of design, and the corresponding sector service areas
become 26-420 km.sup.2. Although the serving base station (and
sector) of a wireless E-911 caller, for example, are known, the
corresponding location information is clearly too coarse for timely
E-911 response. FIG. 1 of the drawings depicts this described
situation for conventional omnidirectional or sector base
stations.
As shown in FIG. 2, prior art cellular telephone systems 10 include
a mobile telephone switching center (MSC) 12 and a plurality of
base stations such as cell site transceivers 14a-14c. The cell site
transceivers transmit radio signals to and receive radio signals
from one or more mobile units 16 that move about a cellular service
area 20. A mobile unit, as the term is used herein, refers to a
wireless voice telephone or data receiver that can be permanently
installed at a fixed location or within a vehicle or that can be
portable. Each cell site transceiver 14 is able to broadcast and
receive the radio signals within a geographic area 18 called the
cell site coverage area. Together, the areas 18 comprise the entire
cellular service area 20. Typically, a cellular service area
comprises a metropolitan area or larger region.
When a telephone call to a called mobile unit 16 originates from
either another mobile unit or a land-based telephone via a Public
Switched Telephone Network (PSTN) 22, a caller must first access
the cellular telephone system 10. This task is accomplished by
dialing the mobile unit's unique identification number (i.e., its
phone number). The MSC 12 receives the call request and instructs
the central call processor 24 to begin call processing. The central
call processor 24 transmits a signal over a dedicated line 26 (such
as a telephone line or microwave link, etc. (to each of the cell
site transceivers 14a-14c causing the cell site transceivers to
transmit a page signal that the mobile unit 16 receives. The page
signal alerts a particular mobile unit 16 that it is being called
by including as part of the page signal the paged mobile unit's
identification or phone number.
Each cell site transceiver 14 transmits the page signal on one or
more dedicated forward control channels that carry all pages, as
well as control signals, channel assignments, and other overhead
messages to each mobile unit. The forward control channel is
distinct from the voice channel but actually carries voice
communications between a mobile and another mobile unit or a
land-based telephone. Each cell site transceiver may have more than
one forward control channel upon which pages can be carried.
When a mobile unit is not engaged in a telephone call, it operates
in an idle state. In the idle state, the mobile unit will tune to
the strongest available forward control channel and monitor the
channel for a page signal or other messages directed to it. Upon
determining that a page signal is being transmitted, the mobile
unit 16 again scans all forward control channels so as to select
the cell site transceiver transmitting the strongest Signal. The
mobile unit then transmits an acknowledgement signal to the cell
site transceiver over a reverse control channel associated with the
strongest forward control channel. This acknowledgement signal
serves to indicate to the MSC 12 which of the forward control
channels (associated with the several cell site transceivers
14a-14c) to use for further call processing communications with
mobile unit 16. This further communication typically includes a
message sent to the mobile unit instructing it to tune to a
particular voice channel for completion of call processing and for
connection with the calling party.
The details of how the cell site transceivers transmit the signals
on the forward and reverse control channels are typically governed
by standard protocols such as the EIA/TIA-553 specification and the
air interface standards for Narrowband Analog Mobile Phone Service
(NAMPS) IF-88 and IS-95 air interface standards for digital
communications, all of which are well known to those of ordinary
skill in the wireless telephone communications art and therefore
will not be discussed.
While cellular networks have been found to be of great value to
mobile users whose travels span many miles, they have also been
found to be prohibitively expensive to implement for small scale
applications wherein system subscribers only desire wireless
telephone services in limited geographic areas, such as, for
example, within office buildings or in campus environments.
The Personal Communications Network (PCN) is a relatively new
concept in mobile communications developed specifically to serve
the aforementioned applications. Similar to cellular telephony
goals, a Personal Communications Network goal is to have a wireless
communication system which relates telephone numbers to persons
rather than fixed locations. Unlike cellular telephones, however,
the PCN telephones are directed to small geographic areas thus
defining "microcellular" areas designed to operate in similar
fashion to large scale cellular telephone networks. PCN
technologies are also similar to residential cordless telephones in
that they utilize base stations and wireless handsets. Unlike the
former, however, PCN technology utilizes advanced digital
communications architecture, such as, for example, PACS, formerly
called WACS, (Bellcore), DECT (European), CDMA (Omnipoint), PHS-PHP
(Japan), IS-54 (TDMA), IS-95 (CDMA), PCS-1900 (GSM), and B-CDMA
(Oki), and features which may be implemented either as private
networks or regulated services. When offered by communications
carriers as services, this PCN capability is generally referred to
as Personal Communications Services (PCS), and may be situated in a
wide variety of environments, including, for example, outdoor
urban, suburban, rural, indoor single-level and indoor multi-level
areas.
As shown in FIG. 3, prior art PCS systems 28 include one or more
control units 30 which, in accordance with the American National
Standards Institute (ANSI) T1P1 working document for stage 2
service description, as known to those skilled in the art, are
termed Radio Access System Controllers (RASCs), access managers,
etc. These control units 30 operate in similar fashion to the MTSC
12 of the cellular telephone network and, therefore, are provided
in electrical communication with the Public Switched Telephone
Network 22. A plurality of base stations or Radio Ports (RPs) 32
are also provided which transmit radio signals to and receive radio
signals from one or more subscriber wireless telephones 16, termed
mobile units or Radio Personal Terminals (RPTs) that move about a
PCS service area 34. Each Radio Port 32, like cell site
transceivers 14, is able to broadcast and receive radio signals
within a geographic area 36 called the Radio Port coverage area.
Together, the areas 36 comprise the entire PCS service area 34.
A generalized reference architecture for the PCS system of FIG. 3
is shown in further detail in FIGS. 4a-4b. The reference
architecture includes reference elements which support radio
access, wireline access, switching and control, mobility
management, and Operations, Administration, Maintenance and
Purchasing (OAM&P). As shown in the schematic, the PCS system
includes a PCS Switching Center (PSC) 38 which supports access
independent call/service control and connection control (switching)
functions and is responsible for interconnection of access and
network systems to support end-to-end services. The PCS switching
center 38 represents a collection of one or more network elements.
The system further includes a Radio Access System Controller (RASC)
40 which supports the wireless mobility management and wireless
access call control functions. It serves one or more subtending
Radio Port Controllers (RPCs) 42 and may be associated with one or
more PCS switching centers 38. As known to those skilled in the
art, Radio Port Controllers 42 provide an interface between one or
more subtending Radio Port Intermediaries (RPIs), a PCS switching
center such as 38, and RASC, air interface independent radio
frequency transmission and reception functions.
The system further includes a Radio Port Intermediary (RPI) 44
which provides an interface between one or more subtending Radio
Ports 46 and the Radio Port Controller 42, and supports air
interface dependent radio frequency transmission and reception
functions. Radio Port 46 supports the transmission of signals over
the air interface and is provided in communication with Radio
Personal Terminal (RPT) 48. This is a light-weight, pocket-size
portable radio terminal providing the capability for the user to be
either stationary or in motion while accessing and using
telecommunication services.
The system further includes variations of RPTs which are in fixed
locations, termed Radio Termination Type 1 (50) and Radio
Termination Type 2 (52), which interface Terminal Equipment Type 1
(54) and Terminal Equipment Type 2 (56) to the Radio Access
Interface.
The system of FIG. 4 further includes a Terminal Mobility
Controller 58 which provides the control logic for terminal
authentication, location management, alerting, and routing to
RPT/RTs. There is also provided a Terminal Mobility Data-store
(TMD) 60 which is operative to maintain data associated with
terminals.
Still further, the system includes a Personal Mobility Controller
(PMC) which provides the control logic for user authentication,
service request validation, location management, alerting, user
access to service profile, privacy, access registration, and call
management. PMC 62 is provided in communication with a Personal
Mobility Data-store (PMD) which maintains data associated with
users.
Finally, the system includes Operations, Administration,
Maintenance, and Provisioning, (OAM & P) systems 66 which
monitor, test, administer, and manage traffic and billing
information for personal communications services and systems. PCS
38 is also provided in communication with Auxiliary Services 68,
Interworking Functions (IWF) 70 and External Networks 72. In
accordance with the above-referenced working document for Stage 2
service description, auxiliary services 68 are defined as a variety
of services such as voice mail, paging, etc. which may not be
provided by the PCS 38. IWF 70 are further defined as mechanisms
which mask the differences in physical, link and network
technologies into consistent network and .user services. Still
further, external networks 72 are defined as other voice, digital
data, packet data, and broadband data networks.
FIG. 5 provides a unified functional model of the detailed system
of FIGS. 4a-4b. This functional model is derived from the PCS
reference architecture in FIG. 4 by aggregating the terminal
entities (RT and RPT) into a single functional grouping RTF, and
aggregating RP, RPI, and RPC into another single functional
grouping RCF in accordance with the Stage 2 service descriptions
for PCS. The model includes Call Control Function (CCF) 74, Service
Switching Function (SSF) 76, Service Control Function (SCF) 78,
Service Data Function (SDF) 80, Service Resource Function (SRF) 82,
Radio Control Function Access Control Function (RACF) 84, Radio
Control Function (RCF) 86, and Radio Termination Function (RTF) 88.
The functions of the terminal elements are more fully described in
the Stage 2 service description for PCS.
Wireless communication services such as the above cellular and PCS
systems, have been quickly embraced by those people whose business
requires them to travel frequently and to be in constant
communication with their clients and associates. The increased use
of wireless communication services, however, have caused headaches
for emergency operators and other position dependent service
providers who require precise location data. As known to those
skilled in the art, under current wireless technology, position
data is strictly limited to relatively large coverage areas and
sectors thereof as defined by the RF characteristics, i.e.
footprints, of the associated base station. As explained below,
these coverage areas are generally unsuitable for most commercial
and consumer applications.
In the late 1960's, federal legislation was enacted which
established the 9-1-1 telephone number as a national emergency
resource. In land-based systems, Enhanced 9-1-1 (E 9-1-1) wireline
technology provides the caller's Automatic Location Identification
(ALI) with reasonable accuracy, cost and reliability, to a Public
Safety Answering Point (PSAP) via a defacto standard. ALI is
generally accomplished by receiving the ANI, or Automatic Number
Identification, during call setup to the PSAP. A database query,
given ANI, provides ALI to the emergency call taker display
terminal as both parties establish the voice channel.
Currently, however, wireless technology does not provide ALI. As a
result, an ever-increasing percentage of emergency telephone calls
can be tracked no further than the originating base station. As
readily seen, the heart of the problem for providing E9-1-1 ALI
services for wireless communication customers lies in accurately
and reliably determining the handset location, under any
circumstance, at low cost.
Against this background, there have been previous attempts to
provide methods and systems which generally identify the positions
of wireless communication system users in cell site coverage areas
and sectors thereof. See, for example, U.S. Pat. No. 4,876,738
issued to Selby and assigned to U.S. Phillips Corporation. Selby
discloses a registration procedure in which the base station
monitors the location of the mobile unit by cell site. The effect
is to allow enlargement of the registration area if the mobile unit
consistently roams between two cells.
See also, U.S. Pat. No. 5,179,721 issued to Comroe et al and
assigned to Motorola, Inc. Comroe discloses a method for
inter-operation of a cellular communication system and trunking
communication system by transmitting an access number for each
system such that the mobile unit may be used as a cellular
telephone and a trunking communication device.
Still further, see U.S. Pat. No. 5,097,499 issued to Consentino and
assigned to AT&T Bell Laboratories. Consentino teaches a method
for preventing an overload in a reverse channel by delaying the
time of the generation of timing stamps on markers.
These methods and systems, however, have proven unsuitable for
commercial and consumer applications where users may, at any given
time, travel through very small portions of numerous cell site
coverage areas and sectors. Under current wireless technology, and
as described in the prior art referenced above, presently available
positioning methods and systems are limited to a determination of
whether the user is within one or more predetermined cell site
coverage areas or sectors. These prior art systems are incapable of
providing further detail, i.e. exactly where in the cell site
coverage area the user is located.
Prior art attempts to design higher accuracy positioning systems
which utilize commercial broadcast transmissions, for example, have
also met with limited success. See, for example, U.S. Pat. Nos.
4,054,880 (Dalabakis et al) and 3,889,264 (Fletcher) which disclose
what are known as "delta-position" systems. These prior art patents
describe systems using three spectrally spaced-apart radio signals,
each of which is an independent AM radio signal. The systems
typically have a vehicle carried mobile receiver, with a separate
tuner for each station, and a second receiver at a fixed, known
position. As disclosed, these systems count "zero crossing counts",
each of which indicates that the user has moved a certain distance
from his or her previous location. In operation, if it is desired
to determine the current position of the user, a starting position
must first be specified. A fixed position receiver detects
frequency drift of the transmitters, which is used to adjust and
coordinate zero crossing counts made by the mobile receivers.
These systems are termed "delta-position" systems because they
determine only the distance and direction traveled by a mobile user
from any particular starting point. Neither Dalabakis et al nor
Fletcher actually determines the position of the mobile user.
See also, U.S. Pat. No. 5,173,710 to Kelley et al which discloses
the use of a fixed position receiver which is adapted to determine
frequency drift along with the relative phases of various
unsynchronized FM broadcast signals originating from known fixed
locations. As disclosed by Kelley, each of the fixed transmitters
transmits a beacon signal having a phase that is unsynchronized
with the phases of the beacon signals of the other transmitters.
These signals are 19 kHz analog pilot tones generated by commercial
broadcast stereo FM stations. The fixed receiver receives the
beacon signals, determines the relative phases of the beacon
signals, and broadcasts data representing these relative phases for
receipt by the mobile receiver which is at an unknown location.
Each mobile receiver includes phase measurement circuitry that
detects the phases of the beacon signals at the mobile receiver's
current position on multiple distinct carrier frequencies such that
the current position of the mobile unit may be determined when used
in conjunction with the fixed receiver broadcast data.
See also, U.S. Pat. Nos. 5,055,851; 4,891,650; and 5,218,367, all
issued to E. Sheffer and assigned to Track Mobile, Inc. Like the
'650 patent, the '851 patent utilizes measurements of the mobile
unit's signal strength which is detected by some number of
neighboring base stations in order to calculate location. In
operation, each base station transmits a special packet of data
which includes this information for receipt by the MTSC. Another
packet of information, the actual vehicle alarm distress call (this
is not the same as a 9-1-1 call), is also sent to the MTSC. The
MTSC sends these two information packets to a Track Mobile alarm
center personal computer. The computer matches both packets using a
simple algorithm in order to find the vehicle's distance from the
base station cell center point. As disclosed, this is done
preferably with four neighboring base station cell site
measurements along with arcuation or line intersection techniques.
The results are displayed on a computer screen map. A 9-1-1 call
may then be initiated by a Track Mobile attendant, based on a
verbal request from the originating mobile user.
The '367 patent operates in much the same way although it uses a
modified handset including a modem, to send signal strength
measurements received at the mobile unit, through the cellular
network to the Track-Mobile alarm center. Only the downlink signal
strengths, received at the mobile unit, are used to estimate
location. The location is determined from the same algorithm as in
the '851 patent, but includes a refinement--antenna sector ID--if
known. As disclosed, the sector ID information reduces error by
effectively slicing the cell circle into one of three pie-shaped
sections. In the case of low power PCS installations, it is likely
that omnidirectional antennas would be used, thus precluding the
use of this sector refinement.
None of the systems referenced above, as well as general time
difference of arrival location systems such as LORAN, NAVSTAR, and
GPS, as used for example in U.S. Pat. No. 4,833,480, issued to
Palmer et al, have proven suitable for commercial applications
since, by design, they require specially adapted receivers to
receive and process the pilot tones, GPS signals, etc. at the
mobile unit. This sophisticated end equipment, of course,
significantly adds to the cost of the corresponding mobile unit. In
the case of hand portable units, this additional equipment further
results in a handset which is extremely bulky and difficult to
handle. As a result, these systems have proven unsuitable for both
large scale commercial applications, as well as ordinary consumer
use.
When applied to wireless communications of interest to the present
invention, i.e. communications in the frequency band from 800 MHz
to 2.5 GHz, these prior art systems are further considered
unsuitable for commercial applications in view of their anticipated
use of excessive frequency spectrum. More specifically, it is
anticipated that for proper operation, these systems would
necessarily require transmission of signals on separate channels
which would utilize an unacceptable amount of additional
spectrum.
Still further, the prior art systems fail to account for changes in
environmental conditions. For example, it is known to those of
skill in the art that for GPS receivers, the location calculation
will not work unless there is a clear view of at least 3-4
satellites. In dense urban areas, especially at the street level,
this condition could easily prevail. Thus, no location estimate
would be available if less than three satellite signals can be
received.
In many office buildings, the metal content of the windows is also
sufficient to preclude effective satellite reception. To this end,
if all wireless antennas were isotropic and were located in flat
and open terrain, estimating the location of a handset/mobile unit
using the prior art Trackmobile signal strength technology might be
sufficient. Unfortunately, the known advantage of the PCS world,
and to a reasonable extent, cellular, does not have a flat and open
terrain. None of the prior art patents work in areas where there
are obstructions to the radio signal's path like buildings, trees,
hills, and automobiles.
Seasons are also known to have a dramatic affect on propagation
where radio waves are significantly attenuated by tree leaves in
the summer, but less so in the winter. Thus, actual RF field data
gathered in one season may not be accurate in another season. As
readily seen, precisely predicting location based on RF propagation
loss has generally been an intractable problem, due to the
complexity of factors, as well as the data collection difficulties
in constructing the necessary databases needed to supply the actual
field data. Thus, the principles relied upon by the
above-referenced prior arts patents, free space loss, rarely
exists, as obstructions and interference increases daily, even in
the most optimal RF environments.
Consequently, a need has developed to provide a positioning method
which may be practically and economically implemented for use in
wireless communication systems and, in particular, in the microwave
band from 800 MHz to 2.5 GHz.
Still further, a need has developed to provide such a method which
may be used by service providers to provide precise location
information for use in emergency situations such as locating an
Enhanced 9-1-1 (E9-11) 1) caller, enforcing restraining orders and
house arrests, confirming the intended location of a user at
predetermined times and the like. It is further desirable that such
a method be compatible with existing wireless telephone technology
and should not degrade the operation of an existing system.
Finally, such a system should neither require the allocation of
more radio frequencies than are currently allocated to wireless
telephone systems, nor require a substantial portion of existing
wireless frequencies.
DISCLOSURE OF THE INVENTION
It is a general object of the present invention to overcome the
limitations of the prior art by providing a positioning method for
accurately determining the location of a mobile unit.
More particularly, it is an object of the present invention to
provide a method for determining the position of a mobile unit by
obtaining directional information from Spatial Division Multiple
Access (SDMA)-equipped and non-SDMA-equipped base stations.
In carrying out these and other objects, features and advantages of
the present invention, a method is provided for determining the
position of a mobile unit such as, for example, a wireless
telephone, Personal Digital Assistant (PDA), or similar interactive
electronic device. The method is also applicable to spread-spectrum
residential cordless telephones which operate in the 900 MHz
frequency band.
According to the invention, the method is provided for use in a
wireless communication system, sometimes also referred to as a
public land mobile telephone system, which includes a plurality of
base stations each having a corresponding coverage area. For each
of the base stations, a plurality of RF measurements are determined
in cooperation with the receiver, including the link budget of the
base station, for a predetermined plurality of distances and
directions. The determined RF measurements for each of the base
stations are modeled as a scaled contour shape having minimum and
maximum boundaries and which is capable of being projected on a
mapping system such as an orthophotograph. The base stations which
neighbor the mobile unit are thereafter determined so as to define
a first bounding polygon area by their intersecting contours. The
first bounding polygon area generally describes the relative
position of the mobile unit.
In one preferred embodiment, each of the base stations is
SDMA-equipped and includes what is known to those skilled in the
art as a smart antenna. The Smart antenna has a predetermined
number of fixed lobes each of which is oriented in a predetermined
direction and is operative to communicate with a mobile unit on a
corresponding communication channel within a known coverage area.
In keeping with the invention, the lobes of the neighboring base
stations which are in communication with the mobile unit must be
determined along with their orientations in terms of azimuth angles
so as to define a second bounding polygon area that describes the
relative position of the mobile unit. By determining where the
first and second bounding polygon areas intersect, a location
polygon may be defined which more precisely describes the position
of the mobile unit in terms of minimum and maximum error
estimate.
In further keeping with the invention, an alternative embodiment is
similarly provided wherein each of the SDMA-equipped base stations
includes a smart antenna with a predetermined number of floating
lobes. In this embodiment, each of the lobes is operative to track
and communicate with a mobile unit on a corresponding communication
channel by changing its orientation within a predetermined
direction range. Like the above-described method, for each of the
base stations, a plurality of RF measurements are determined in
cooperation with the receiver, including the link budget of the
base station, for a predetermined plurality of distances and
directions. The determined RF measurements for each of the base
stations are modeled as scaled contour shapes each having minimum
and maximum boundaries and which is capable of being projected on a
mapping system such as an orthophotograph. The base stations which
neighbor the mobile unit are thereafter determined and their
corresponding contours are analyzed to further determine where they
intersect. These intersections will define a first bounding polygon
area that describes the relative position of the mobile unit.
By determining which lobes of the neighboring base stations are in
communication with the mobile unit, their corresponding orientation
ranges may be described in terms of azimuth angles so as to define
a second bounding polygon area that describes the relative position
of the mobile unit. Like the above embodiment, it may thereafter be
determined where the first and second bounding polygon areas
intersect so as to define a location polygon which more precisely
describes the position of the mobile unit in terms of minimum and
maximum error estimate.
In further keeping with the invention, there is provided yet
another alternative embodiment which is similarly provided for use
in a wireless communication system including a plurality of base
stations each having a non-SDMA antenna with a corresponding
coverage area. In this embodiment, each of the base stations is
further provided a plurality of bandpass filters in electrical
communication with a corresponding antenna. Each of the bandpass
filters includes a corresponding waveguide and is operative to
receive mobile unit signals from predetermined directions in
predetermined frequency slots.
Again, for each of the base stations, a plurality of RF
measurements are determined in cooperation with a receiver,
including the link budget of the base station for a predetermined
plurality of distances and directions. The determined RF
measurements for each of the base stations are thereafter modeled
as scaled contour shapes having minimum and maximum boundaries and
which are capable of being projected on a mapping system such as an
orthophotograph. The base stations which neighbor the mobile unit
are thereafter determined and their corresponding contours are
analyzed to further determine where they intersect. These
intersections define a first bounding polygon area that describes
the relative position of the mobile unit.
By determining the communication frequencies of the neighboring
base stations in the corresponding directions of the received
mobile unit signals, a second bounding polygon area may be defined
which describes the relative position of the mobile unit. Again, it
may thereafter be determined where the first and second bounding
polygon areas intersect so as to define a location polygon which
more precisely describes the position of the mobile unit in terms
of minimum and maximum error estimate.
In each of the above embodiments, once the location polygon area
has been defined, the latitude and longitude of the center of the
polygon area may also be determined whereupon the street addresses
contained therein may be learned through reference to one or more
databases.
In keeping with the invention, the modeling of the determined RF
measurements as scaled contour shapes requires the initial
segmenting of the coverage areas of each of the base stations into
a plurality of arc segments. For each of the arc segments, a
plurality of single or multiple regressions are thereafter
performed so as to convert actual data into a corresponding
plurality of mathematical curve-fit equations, each representing a
relationship between a predetermined measurable variable, i.e.
Relative Signal Strength Indication uplink (RSSI.sub.up), Relative
Signal Strength Indication downlink (RSSI.sub.down), Word Error
Rate uplink (WER.sub.up), Word Error Rate downlink (WER.sub.down),
Quality Indication uplink (QI.sub.up), Quality Indication downlink
(QI.sub.down), Time Differential uplink (TD.sub.up), Time
Differential downlink (TD.sub.down), instantaneous power of each
transmitter, start-up power of each transmitter, etc. and distance
from the base station. Note that in certain cases of TDMA and CDMA
techniques, the initial and instantaneous transmitter power levels
must also be known.
For each of the arc segments, the degree of fit of the
corresponding mathematical equations may thereafter be determined
by comparing each of the mathematical equations with actual data.
The equations may further be optimized by determining which has the
best correlation and least standard error for a predetermined
portion of each arc segment. Finally, the optimized mathematical
equations may be combined for each arc segment so as to form the
scaled contour shape corresponding to each base station.
In further keeping with the invention, a Genetic Algorithm (GA) may
also be used to optimize the parameters of each of the single or
multiple regressions so as to further improve the degree of fit for
greater correlation and minimum standard error.
Still further, in cases where there is generally poor correlation
between all of the mathematical equations of an arc segment and the
actual data, the corresponding base station may be instructed along
with the receiver (i.e., the mobile unit) to each temporarily
change their transmission frequencies by 10-40 MHz. Thereafter,
additional RF measurements may be obtained for the base station at
the changed frequency, including its link budget, for the same
predetermined plurality of distances and directions, thus yielding
an increased number of variables for consideration and
analysis.
BRIEF DESCRIPTION OF THE DRAWINGS
The foregoing aspects and many of the attendant advantages of this
invention will become more readily appreciated as the same becomes
better understood by reference to the following best modes for
carrying out the invention, when taken in conjunction with the
accompanying drawings.
FIG. 1 is a schematic diagram illustrating conventional base
station 911 position accuracy;
FIG. 2 is a generalized schematic diagram of a prior art cellular
telephone system;
FIG. 3 is a generalized schematic diagram of a prior art Personal
Communications System (PCS);
FIG. 4 is a detailed schematic diagram of the system of FIG. 2;
FIG. 5 is a unified functional model of the system of FIGS. 2 and
3;
FIG. 6 is a schematic diagram of a generalized positioning system
which may be used in accordance with the method of the present
invention;
FIG. 7 is a schematic diagram of a fixed lobe antenna;
FIG. 8 is a schematic diagram of a floating lobe antenna;
FIG. 9 is a schematic diagram of a bandpass filter equipped
antenna;
FIG. 10 is a schematic diagram of a bandpass filter equipped
sectorized antenna;
FIG. 11 is a representative curve fit graph obtained from the
generic curve fit database used in accordance with the present
invention;
FIG. 12 is a schematic of typical data obtained when utilizing the
Bollinger band database in accordance with the present
invention;
FIG. 13 is a schematic diagram of representative bounding polygons
obtained by using a run-time database in accordance with the
present invention;
FIG. 14 is a schematic diagram of representative arc segments drawn
around a Radio Port;
FIG. 15 is a schematic diagram of a first sample curve fit plot
before and after a manual search;
FIG. 16 is a schematic diagram of a second sample curve fit data
plot before and after a manual search;
FIG. 17 is a schematic diagram of a third sample curve fit data
plot before and after a manual search;
FIG. 18 is a schematic diagram of a fourth sample curve fit data
plot before and after a manual search;
FIG. 19 is a schematic diagram of a best fit confidence interval
with maximum and minimum bands;
FIG. 20 is a schematic diagram of a representative Bollinger
band;
FIG. 21 is a schematic diagram of a location band;
FIGS. 22-23 provide representative schematics of RSSI.sub.downlink
and WER.sub.uplink ;
FIG. 24 is a schematic diagram of a first location polygon obtained
in accordance with the present invention;
FIG. 25 is a schematic diagram of a second bounding polygon area
formed by the determined azimuth angles of the communicating lobes
of neighboring base stations;
FIGS. 26-27 are schematic diagrams of the intersection of first and
second bounding polygon areas; and
FIG. 28 is a block diagram of the method steps of the present
invention.
BEST MODES FOR CARRYING OUT THE INVENTION
As referenced above, the present invention is adapted for use in a
wireless communication system, sometimes also called a public land
mobile telecommunications system. It should be understood, however,
that the method of the present invention may be suitable for use
with most interactive electronic devices such as the Personal
Digital Assistants (PDAs) and the like. For example, in the case
where a PDA is available, through a series of mutually agreed upon
message formats between the lap and the PDA, location information
can be sent to the PDA device instead of forwarding information to,
or in addition to, the PSTN. The following disclosure is thus
intended as one of example and not limitation.
With reference to FIG. 6 of the drawings, there is provided a
schematic diagram of a generalized positioning system which may be
used in accordance with the method of the present invention. The
positioning system is designated by reference numeral 90 and
includes at least one base station 92 such as a Radio Port which is
operative to receive calls from one or more mobile units 94 such as
Radio Personal Terminals (RPTs) over air interface channels. The
system further includes a control unit 96 which is provided in
electrical communication with the Public Switched Telephone Network
(PSTN) 98. The functions of control unit 96 may be implemented in a
Mobile Telephone Switching Center (MTSC) when used in a cellular
telephone network or they may be implemented in a Radio Port
Controller, RASC, etc. when used in a PCS system or the like. A
location databank 100 is also provided which is operative to store
real-time RF measurements for the base stations 92, including their
link budgets.
Each of the base stations 92 may be Spatial Division Multiple
Access (SDMA)-equipped and may include a fixed-lobe or
floating-lobe "smart antenna" as the terms are known and used by
those skilled in the art. FIG. 7 illustrates a smart antenna 102
having a predetermined number of fixed lobes 104 which are oriented
in a predetermined direction and operative to communicate with a
mobile unit 94 on a corresponding communication channel within a
known coverage area. Similarly, FIG. 8 illustrates a smart antenna
106 having a predetermined number of floating lobes 108, each
oriented in a predetermined direction and operative to communicate
with a mobile unit 94 on a corresponding communication channel by
changing its orientation within a predetermined direction
range.
The fixed lobe smart antenna of FIG. 7 is shown with eight lobes
which provide direction readings of 0.degree., 45.degree.,
90.degree., 135.degree., etc. As readily seen, the accuracy would
be plus or minus 22.5.degree.. In contrast, the floating lobe smart
antenna of FIG. 8 is shown with three lobes. As known to those
skilled in the art, theoretically, the maximum number of lobes
possible for a floating lobe smart antenna is equal to the number
of antenna elements. Thus, for example, the floating lobe antenna
shown in FIG. 8 corresponds to a maximum of three elements. In
operation, floating lobes track users as they move. Thus, a lobe
can move from position 1 to 1B as the user moves. At the time of an
emergency call, the direction (azimuth) of the user would be
obtained from the smart antenna. For example, if the user was in
position 1, the azimuth angle was 0 and in position 1B, about
40.degree.-45.degree.. An estimate of the accuracy of this
measurement would then be sent to a location databank. The accuracy
would, of course, be limited by the width of the lobe which is
known to the smart antenna.
In an alternative embodiment shown in FIG. 9, each of the base
stations 92 may be non-SDMA-equipped. In the case of bandpass
filter use with nonsmart antennas, each filter is designed to
receive signals in different frequency slots. For illustrative
purposes, it is therefore assumed that there are eight frequency
slots available in a (Time Division Multiple Access (TDMA) system
for a handset to communicate to the antenna. In regular operation,
a handset/mobile unit sends a pilot signal to a base station
informing the base station that it desires to transmit and the base
station assigns a frequency slot (one of eight) to the handset.
Assume further that bandpass filters are built that filter 1 110
would only pass those signals that correspond to frequency 1. In
order to have separation between filters passing frequencies, every
other frequency slot is therefore assigned to a filter. Thus,
filter 2 112 passes only frequency 3. Similarly, filter 3 114
passes only frequency 5. Still further, filter 4 116 passes only
frequency 7. The waveguides that feed into the filters are designed
in such a way that they pick signals from a specific direction (in
this case, for example, North, West, South, East or 0.degree.,
90.degree., 180.degree., 270.degree.). Now in the case of an
emergency call, for example, the base station tells the mobile unit
to go to frequency 1 and transmit a brief (coded or known) pulse,
then to frequency 3 and the same pulse, then to frequency 5, then
to frequency 7.
Depending on the location of the user, one or more of these filters
can receive signals. If a first user is dead North, for example, of
the antenna, most likely only filter 1 will receive a signal while
the mobile unit is jumping from frequency to frequency. This
information would therefore be sent to the location databank (in
degrees or maybe the ID of the filter, just like the "smart
antenna" case). An accuracy estimate would also be sent. In this
case, the accuracy would depend on how many filters are used and
what kind of processing is done in the filters.
Further enhancements to the estimate can be obtained with
processing at the filters. For example, if the signal strength for
each received pulse is calculated, then that could be used to
refine the estimate. Imagine a user 2 which is North-Northwest.
Both filter 1 and filter 4 would receive signals while the mobile
unit is jumping from frequency to frequency. However, the received
signal strength of filter 1 would be larger than filter 4 which
would suggest that the user is to the North of Northwest. If signal
strength is not measured, the user could only guess Northwest since
both filters 1 and 4 received signals.
As can readily be seen, this type of scheme is useful and practical
for antennas placed at intersections. Along the streets, the
signals would be funneled to the filters. In the case of anomalies
(a user that is dead North) transmits a pulse at frequency 3 (the
frequency for filter 2) and through a long path, this is picked up
by filter 2. In this case, the user could only hope to receive a
pulse at frequency 1 (the frequency for filter 1) that is stronger
than the one at frequency 3 since the frequency 1 pulse followed
more straight path. Thus, measuring signal strength of the pulse
received at the bandpass filter would help in such cases.
In a further alternative embodiment, the antennas may be
non-SDMA-equipped and sectorized. With reference to FIG. 10 of the
drawings, it can be seen that in such a case, direction information
can only be obtained which is plus or minus 60.degree. accurate.
Therefore, in order to make this more accurate, selective filters
can be placed on the sectors so as to increase the accuracy of
direction information. As seen, if two filters are placed in a
sector, plus or minus 30.degree. of accuracy would be obtained.
These selective filters, such as waveguides, can be rectangular,
circular, cylindrical, or any other suitable shape. By using the
geometry, waveguides are designed to block certain frequencies to
be passed (selective high pass filters). In operation, the
waveguides will feed into an RF processing circuit where the signal
will be analyzed for its contents.
The importance of this direction information is that it is not
possible in all cases to triangulate and obtain direction
information. For example, triangulation in 2D requires at least
signals from three Radio Ports. Similarly, triangulation may not be
reliable in some areas due to environmental conditions,
obstructions, etc.
In the system of FIG. 6 which is operative to be used in accordance
with the teachings of the present invention, a location databank
100 is also provided which is operative to store real-time
measurements for base stations 92, including their link budgets. As
explained in further detail herein, these RF measurements may
include, for example, Relative Signal Strength Indication uplink
(RSSI.sub.up), Relative Signal Strength Indication downlink
(RSSI.sub.down), Word Error Rate uplink (WER.sub.up), Word Error
Rate downlink (WER.sub.down), Quality Indication uplink
(QI.sub.up), Quality Indication downlink (QI.sub.down), Time
Differential uplink (TD.sub.up), Time Differential downlink
(TD.sub.down), instantaneous power of each transmitter, start-up
power of each transmitter, etc. and distance from the base
station.
Finally, the positioning system 90 includes a Location Adjunct
Processor (LAP) 118 which may be an Intelligent Peripheral (IP) or
other suitable device which is in electrical communication with the
location databank 100 and control unit 96. The LAP 118 is operative
to access the location databank 100 and determine and forward the
location of the mobile unit 94 to the control unit 96.
As shown, positioning system 90 is directed for use with the Public
Switched Telephone Network (PSTN) 98 which is provided in
electrical communication with control unit 96. Control unit 96 is
therefore operative to receive calls forwarded by base stations 92
temporarily suspend call processing, and generate call information
request signals. The LAP 118 receives the call information request
signals, accesses databank 100 and determines and forwards the
location of the mobile unit 94 to the control unit 96. The call is
thereafter forwarded to the PSTN 98 along with the determined
mobile unit location.
Applicants recognize that various alternative embodiments of
positioning system 90 may be used in accordance with the teachings
of the present invention as shown and described, for example, in
co-pending U.S. patent application Ser. No. 08,314,477 filed Mar.
28, 1994 which has at all times relevant hereto been commonly owned
with the present application.
At the threshold, it should be understood that each of the systems
referenced above and used in accordance with the teachings of the
present invention requires detailed location processing. This
processing utilizes scaled contour shapes. The shapes are modeled
based upon determined RF measurements for each base station 92. The
location processing of the present invention thus focuses on the
ability to predict and model RF contours using the actual RF
measurements, then performing data reduction techniques such as
curve-fitting techniques, Bollinger bands, and genetic algorithms,
in order to locate a mobile unit and disseminate its location.
An example of a suitable software analysis tool is a program by
Axcelis, Inc. termed "Evolver 2.0". This is an Axcelis spreadsheet
program that can perform a genetic algorithm optimization of the
parameters generated in the above curve fitting techniques.
Location Processing
More specifically, the location processing steps include the
initial modeling of determined RF measurements for each of the base
stations as a scaled contour shape having minimum and maximum
boundaries which is capable of being projected on a mapping system
such as an orthophotograph which may be digitally recorded.
Thereafter, it must be determined which of the base stations can be
"heard" by the mobile unit, i.e., which base stations are neighbors
of the mobile unit. Once this information is known, it may further
be determined where the corresponding contours of the neighbor base
stations intersect so as to define a first bounding polygon area
that describes the position of the mobile unit in terms of a
minimum and maximum error estimate.
As readily seen, a key component of the present invention is the
initial ability to diagram and model the RF propagation loss from a
given Base Station/Radio Port, for various RF measurement arc
segments, which will define entire contours. As those skilled in
the art will recognize, in theory, if the "free space" power loss
is known for all useful distances in all directions from a base
station, then individual circular power loss contour shapes may be
drawn around the base station. Assuming two or preferably three
base stations are neighbors of the mobile unit, then RF
measurements may be used to determine location via intersecting
contours. The particular shape of the contour intersections is the
bounding polygon that describes the location, in terms of the
maximum error estimate.
Unfortunately, the principle of free space loss rarely exists when
attempting to predict base station coverage areas since the
surrounding buildings, trees, traffic signs and other geographical
"clutter" blocks transmitted signals. To account for these
variables involved in propagation prediction, the present invention
therefore utilizes a number of segmented models and analysis
techniques for data reduction purposes. The resulting output
becomes the location databank which consists of a collection of
component databases, many of which may be designed on a per base
station basis. The component databases may include a base station
database, a prediction database, a measured RF database, a generic
curve fit database, a Bollinger band database, equipment-specific
corrections database, and a run-time database as described in
further detail below.
Base Station Database
In keeping with the invention, the base station database provides a
detailed list of the attributes of every installed and proposed
base station. Applicants contemplate that this database would
contain the following data elements:
1. Name or identification of base station.
2. Base station vendor name, model number, serial number.
3. Latitude (LAT), Longitude (LONG), or at least accurate street
location detail for conversion to/from LAT and LONG, and Altitude
(ALT) of physical placement of base station.
4. Base station transmitter default power, instantaneous power for
each active transmission channel, and power range.
5. Antenna gain contours (if omni-directional, otherwise sector
make-up, and gains within each sector).
6. Whether or not a distributed antenna scheme is used, and if so,
placement (LAT, LONG, ALT) of all remote antennas.
7. Nearby surrounding obstructions (e.g., the mounting surface of
the RP: is it on a metal wall, in an elevator, or hanging in free
space).
8. Base station transmitter operating frequency band (licensed,
unlicensed), and allowed frequencies.
9. Whether or not a duplicated transmitter is used, and if so,
include the identifying characteristics of each transmitter.
10. The PSAP associated with each base station.
11. Type of air interface: protocol and signaling (e.g., PACS,
CDMA, GSM, DECT, CDMA, PHS-PHP, IS-54, IS-95, PCS-1900, B-CDMA,
etc.) This information should be derived from the base station
vendor name, model number, and serial number. Any dual or
multi-mode capabilities must also be known and characterized.
12. Base station antenna gain contour. This information could be
derivable from knowledge about the antenna's characteristics and
surrounding obstructions.
13. The control unit associated with the base station, neighboring
communication network topology and the associated central office.
This information may be derived from knowledge of the control unit
and its connected central office at the time the wireless
communication system is originally engineered. Nonetheless, the
network topology may change, due to a variety of reasons. For
example, future base stations may use a signaling protocol
arrangement with their control unit such that the base station can
be easily moved around without prior notification to a centralized
work manager system. A control unit may automatically discover the
addition/deletion or in/out change of a particular base station. To
the extent this automatic capability exists, a forwarding event
report message must be sent to a system associated with the
location service. In cases where the control unit is associated
with a PBX, foreign exchange circuit, or similar remoting facility,
the identification and end-to-end topology circuit arrangements
will be needed.
14. Frequency Assignment Characterization (FAC). This should be
derivable from the RP vendor, make/model information. If the FAC is
automatic, then a potential incompatibility may exist during the
performance of the location function. Knowing these details, and/or
having the ability to control the occurrences of frequency
assignment, can resolve incompatibilities.
15. Current operational RP status. This information should be
derivable from the wireless communication network OAM and P systems
that should routinely receive current information about the
in-service state of the base stations. This information is needed,
for example, because a planned, but not in-service base station, or
a faulty base station, could disturb the location algorithm, if
this information is otherwise not known.
16. Traffic load characteristics of the base station and its
superior network. This may be derivable from the network planning
activity, base station model characteristics, and dynamic
monitoring by OAM and P systems, or each base station. For example,
if a base station needed to perform an emergency location function,
it cannot be invoked because it is at 100% of capacity, with no
possibility to shed "non-emergency" load, then other techniques may
be applied.
Prediction Database
This is a planning database primarily populated by, and used to
support/interact with base station site planners and installation
engineers. In accordance with the invention, it is used primarily
to predict coverage. The location function accesses this database
in order to require a rudimentary understanding of intended
coverage area of newly planned cell sites and their operational
status. Using the various RF propagation models and special plane
curves, propagation coverage will be predicted for all base
stations by examining the placement of the base station, local
street widths, and the surrounding clutter. This provides a quick,
inexpensive estimate of each base station's coverage.
Measured RF Database
In keeping with the invention, the measured RF database consists of
actual measurements taken from the area surrounding the base
station. These measurements could be taken by technicians during
base station site installation or any other collection technique.
Both uplink (handset to base station) and downlink (base station to
handset) measurements will be made for data such as Received Signal
Strength Indicator (RSSI), Word Error Rate (WER), Quality Indicator
(QI), and Time Differential. Each of these variables are known to
those skilled in the art and will therefore not be discussed in
further detail. These measurements will be recorded along with the
exact location at which the measurements were taken. All
measurements are made within an arc segment region as discussed in
further detail below.
Generic Curve Fit Database
This database is contemplated for use in accordance with the
invention when no equipment-specific data is required/available.
The generic curve fit database is created in the following
manner:
1. Using the measurements database, load the data for each
measurement type (i.e. RSSI.sub.down), per an arc segment region,
and per a base station, into a curve fitting program. One such
program known to applicants is Table Curve 2D distributed by Jandel
Scientific Software. Using any random or pseudo-random method,
"holdback" 15% of the data points from the curve-fitting exercise,
to be used as verification points later. This process will produce
an equation for each measurement type, per region.
2. Inspect the resulting graphs for each measurement. Measurements
that produce smooth, well-fit curves will be noted.
3. Simultaneously inspect all graphs for a given region. If one
measurement produces a much smoother graph than the others,
determining location in that region will require only one
parameter. Alternatively, there may be areas within the region that
correlate well with some measurements and poorly with others. As
shown in FIG. 11, for example, it can be seen that the correlation
in area A is fairly good for WER and poor for RSSI. Similarly, the
correlation in area B is good for RSSI and poor for WER. These
graphs suggest that determining location will require multiple
parameters. In the example of FIG. 11, WER would be used in areas A
and D, RSSI would be used in area B, and another measurement would
be used in area C.
4. Test the equations by using the data points that were excluded
from step 1. If the results are satisfactory, go on to the next
step. If the error-bounds are too large using the existing
equations, it may be necessary to use genetic algorithms to enhance
the predictive technique for the region. Genetic algorithms could
be used here to simultaneously combine the six (or more) equations
in every conceivable manner to produce the best fit.
5. Store the equations for each region in the location database for
use during a location request, along with the error estimate.
By analyzing the surrounding characteristics for each model region
(i.e. street width, distance from base station to nearest building,
etc.) along with a corresponding location equation, it may be
possible to reuse this information in a predictive manner for
future base station installations. Applicants contemplate that this
could reduce costly manual RF measurement testing.
Bollinger Bands
As known to those skilled in the art, the basic idea behind
Bollinger Bands is to read data points and create a moving average
and a moving standard deviation. The bands are determined by
calculating the average of a certain number of data points plus and
minus two times the standard deviation of the data. A "sliding
window" is used for the volatility of the data. The optimal window
size will vary with the condition of the data.
As shown in FIG. 12, Bollinger Bands provide: (1) the ability to
handle discontinuities and vast multi-model, noisy search spaces;
and (2) they optimize error wherever possible, i.e., wherever field
measurements have a low volatility, then Bollinger Bands will
generally have a low bandwidth, which results in a more accurate
bounding polygon.
In accordance with the present invention and as explained in
further detail below, RF measurements will be analyzed using the
Bollinger band technique in the following manner:
1. Load the data for each measurement type (i.e. RSSI downlink),
per arc segment region, into a program to calculate the sliding
window average and standard deviation.
2. For each distinct measurement value (e.g. -70 Db, -71 Db, -72
dB, etc.), store the measurement value and the corresponding
average distance (in feet) in both the upper and lower band (in
feet), based on the sliding window. Equipment-Specific Corrections
Database
This database is contemplated for use with the present invention if
vendor-specific, and/or model-specific equipment characteristics
are available and are used in the areas of interest, which deviate
from the generic curve fit database assumptions. For example, in
GSM, different vendors use slightly different mapping or transfer
functions, in relating true Word Error Rate, with the vendor's
quantized indicator. It is anticipated that public, open standards
will be defined, that mitigate the need for the Equipment-Specific
Corrections Database. Data for this database would normally be
provided from lab tests performed by mobile unit manufacturers,
which are then used for correction purposes with respect to the
generic curve fit database, and its assumed internal baseline
standard.
Run-Time Database
This database is contemplated by Applicants to be stored directly
in the format of the GIS software being used (e.g. map info or
ARC/info). It is derived from the data reduction processes, for
example, the curve-fitting in Bollinger Band databases. Each arc
segment per base station contains a number of entries. The first
entry defines the independent variables used to calculate location
within this arc segment. There is also one entry for each distinct
measurement value of the independent variables selected (e.g. RSSI
down=-70 dB, -71 dB, -72 dB, etc.) These entries are actually
graphical objects (bounding polygons) that are selectable by the
GIS software.
For example, with reference to FIG. 14 and the table below, assume
the curve fitting in Bollinger Band analysis for base station 1 has
determined that RSSI.sub.up is the best location predictor for arc
segments 1, 2 and 3, while WER.sub.down is the best predictor for
arc segments 4 and 5. The run-time database would contain the
following entries:
______________________________________ RUN-TIME DATABASE Arc
Segment Predictor Variable ______________________________________ 1
RSSI.sub.up 2 RSSI.sub.up 3 RSSI.sub.up 4 WER.sub.down 5
WER.sub.down ______________________________________
In addition, the database would contain many bounding polygons per
arc segment. FIG. 13 illustrates this concept for the five arc
segments mentioned. In this figure, the bounding polygons for
RSSI.sub.up values of -70 dB, -71 dB and -72 dB are displayed for
arc segments 1-3. Additionally, the bounding polygons for WER down
values of 1.1% and 1.2% are displayed for arc segments 4 and 5.
While only 2-3 bounding polygons per arc segment are displayed in
the figure, there would actually be many polygons to cover the
entire range for variable being used.
The run-time database is displayed with one predictor variable per
arc segment as shown above. The Position Location System (PLS)
process will actually use more than one predictor variable per arc
when a single variable does not reliably predict distance. The
run-time database for each arc segment will be constructed by using
the results of the curve fit and Bollinger band databases, and will
actually consist of two tables. The first table will be used to
construct a set of fuzzy logic rules, while the second table will
provide a predicted distance value, along with a minimum and
maximum boundary.
For example, if arc segment 1 of radio port 5 is predicted well by
RSSI.sub.down for values of -40 dB to -70 dB, and WER.sub.down for
values of 1% to 3%, the following entries would appear in the
run-time database rule table:
TABLE 1 ______________________________________ Run-Time Database
Rule Table Radio Arc Port Segment Variable Min Range Max Range
______________________________________ 5 1 RSSI.sub.down -40 -70 5
1 WER.sub.down 1.0 3.0 ______________________________________
The second table for arc segment one would contain entries such as
these:
TABLE 2 ______________________________________ Run-Time Database
Values Table Radio Arc Mean Min Max Port Segment Variable Value
Dist Dist Dist ______________________________________ 5 1
RSSI.sub.down -40 100 0 200 5 1 RSSI.sub.down -41 120 20 220 5 1
RSSI.sub.down . . . . . . . . . . . . 5 1 RSSI.sub.down -70 500 400
600 5 1 WER.sub.down 1.0 400 350 450 5 1 WER.sub.down 1.1 440 390
490 5 1 WER.sub.down . . . . . . . . . . . . 5 1 WER.sub.down 3.0
800 700 900 ______________________________________
During a location request, the LAP would access the run-time
database rules table and construct the following code to determine
the caller's predicted distance from radio port 5 for arc segment
1:
______________________________________ Pseudo-code:
______________________________________ rule.sub.-- 1 = FALSE
rule.sub.-- 2 = FALSE /* look for active rules */ if -70 <=
RSSI.sub.down <= -40 then rule.sub.-- 1 = TRUE if 1.0 <=
WER.sub.down <= 3.0 then rule.sub.-- 2 = TRUE if rule.sub.-- 1
is TRUE and rule.sub.-- 2 is TRUE /* both rules apply, so we have
to perform a weighted average using the difference between
predicted max and min */ weight.sub.-- 1 = (RSSI.sub.down
max-RSSI.sub.down min) / (RSSI.sub.down max-RSSI.sub.down min+
WER.sub.down max-WER.sub.down mean) weight.sub.-- 2 = (WER.sub.down
max-WER.sub.down min) / (RSSI.sub.down max-RSSI.sub.down min+
WER.sub.down max-WER.sub.down mean) /* reverse the weights because
the one with the smaller difference is better and should be
weighted more heavily */ mean = weight.sub.-- 1*WER.sub.down mean +
weight.sub.-- 2*RSSI.sub.down mean min = weight.sub.--
1*WER.sub.down min + weight.sub.-- 2*RSSI.sub.down min max =
weight.sub.-- 1*WER.sub.down max + weight.sub.-- 2*RSSI.sub.down
max else if rule.sub.-- 1 is TRUE use RSSI.sub.down mean, min and
max else use WER.sub.down mean, min and max
______________________________________
The detailed steps of preparing the run-time database and thus the
PCS location databand may be illustrated with reference to FIG. 14
of the drawings. FIG. 14 is a schematic diagram of a Radio Port
that has arc-segments 120 of 6 degrees. The arc-segments create
discrete sections of the area around the Radio Port. With these
sections clearly defined, the RF behavior of the Radio Port can be
characterized in each section independently. After the locations
have been partitioned into arc-segments, a spreadsheet file can be
produced for each arc-segment.
The preparation steps include the initial gathering of field data.
The desired parameters (RSSI.sub.up, RSSI.sub.down, WER.sub.up,
WER.sub.down, QI.sub.up, QI.sub.down, etc.) will be collected at
locations surrounding the Radio Ports. In a preferred embodiment,
these locations will be approximately 10 meters apart from one
another. All measurements will be placed with location tags in a
suitable spreadsheet file such as, for example, Microsoft
Excel.
The locations will thereafter be partitioned into arc segments 120
as indicated above. In keeping with the invention, the locations
need to be partitioned into arc segments 120 in order to accurately
model the parameters around corresponding Radio Ports. After the
data has been collected and partitioned into arc segments, a
suitable curve fitting program such as TableCurve 2 D will be used
to curve-fit the data (distance versus each parameter) for each
individual arc-segment. The software generates a list of functions
that could possibly characterize the data and sorts the functions
(best to worse) by means of lowest Fit Standard Error
(FitStdErr).
Sometimes, the best fit (lowest FitStdErr) that the curve-fitting
software packages produces is not the best fit for the RF data.
There are many different examples of the software package fitting a
curve to the data that is not physical (not possible in the RF
environment). Some examples of non-physical fits are fits that
swing into negative distances, fits that have high sinusoidal
content, and fits that have many slope reversals or large swings in
areas where few or no actual data points reside.
FIG. 15 illustrates two TableCurve 2D curve-fit on the same data.
The plot on the left shows the curve-fit that the software package
chose as the best fit (it is the fit with the lowest FitStdErr).
One skilled in the art would recognize that the plot on the left is
highly unlikely to be representative of the data because of the
large swings where few data points lie. With the data from FIG. 15,
a manual search for the most logical fit is needed. One skilled in
the art would therefore search the fits until she found a fit that
is more logical (like the fit on the right in FIG. 15).
FIG. 16 provides another example of a TableCurve 2D fit that is not
logical. The fit on the left has one swing to a very large distance
(off of the top of the plot) in an area where there are no data
points. The plot on the right is much more likely to describe the
data accurately in the area where there are no data points, even
though it has a higher FitStdErr than the plot on the left.
FIG. 17 illustrates yet another fit (left) that has a large
negative distance swing (again, where no data points lie) and a
sharp, large positive distance swing. In keeping with the
invention, negative distances are not valid because they do not
represent the RF environment properly. The sharp, large distance
swing is not reliable because of the low number of data points in
the area. The plot on the right has a much higher probability of
being accurate.
The lowest FitStdErr fit in FIG. 18 displays a more subtle problem.
The points along the distance axis (vertical) are not well
represented, yet they make up the majority of the data point
population. The plot on the right better represents those data and
also eliminates questionable swings that are in the left plot.
Although manually searching for the most logical fit may result in
a larger FitStdErr, the fit will also be more representative of the
actual RF environment. The number of invalid fits by TableCurve 2D,
for example, can be minimized by collecting a high number (50-60)
of evenly spaced data points within each arc-segment.
After the curve fitting program produces a valid fit, 95%
confidence intervals (or bands) can be created. These bands
(minimum and maximum) are produced by adding and subtracting twice
the FitStdErr to the average fit. Any negative distances will be
eliminated from the band. FIG. 19 shows a best fit with maximum and
minimum confidence bands. It should be noted that through simple
numeric integration, the area of the interval can be computed. The
area of the band will describe how volatile the data is throughout
a complete arc-segment.
After the confidence intervals have been determined, Bollinger
bands can be created for the data in each arc-segment 120. As
indicated above, Bollinger bands are similar to the confidence
intervals in that they represent a range in which data points are
likely to reside. However, Bollinger bands widen according to the
volatility of the data in a certain area of a particular
arc-segment. Basically, the Bollinger interval is wide in areas
where the deviation of the data points is large, and is narrow in
areas where the deviation of the data points is small. FIG. 20
shows how Bollinger bands widen in areas of data volatility.
As discussed above, Bollinger bands use a "sliding window"
technique to compute a moving average across a data set. The
sliding window size for location purposes will be 20% of the data
population for each arc-segment. As with confidence intervals, the
area of the Bollinger bands can be computed through simple numeric
integration. The advantage of the Bollinger band over the
confidence interval is that the area of the Bollinger band in a
discrete section of an arc-segment can describe the volatility of
the data in that section. The area of the confidence interval can
only describe the volatility of the data throughout a complete
arc-segment.
A problem with Bollinger bands is that they have a phase lag that
is introduced in calculating the moving average. Because of this
phase lag, the Bollinger band widens slightly beyond the volatile
data. The amount of phase lag is dependent on the size of the
sliding window.
To "clip" the phase lag, the Bollinger band and confidence
intervals can be intersected. The intersection of these two bands
becomes the location or distance band 122, as shown in FIG. 21. The
location band 122 is what will be used to generate (for the
location databank) minimum and maximum distances for any valid
values of any of the parameters. The area of the location band 122
can be computed with simple numeric integration and is an
indication of the data volatility.
At this stage, location bands have been produced for all parameters
in each arc-segment. Now, a method of determining which parameters
to use is needed. Fuzzy logic will be used to determine which
parameters will be used when estimating a distance. Fuzzy logic, as
known to those skilled in the art, consists of fuzzy patches or
rules which try to explain the behavior of fuzzy systems. Fuzzy
patches or rules are simply if-then-else statements that describe a
discrete section of the system's output. The goal is to have a
group of fuzzy patches that accurately describe the system's
complete output. In this location system, fuzzy rules will be
created to use the parameters with the least volatility to estimate
a distance.
FIGS. 22 and 23 provide examples of two different parameters from
the same arc-segment. An example of a fuzzy rule would be as
follows: If RSSI.sub.downlink reading lies in the range to the left
of the dashed line, use RSSI.sub.downlink. Otherwise, use
WER.sub.uplink.
The above fuzzy rule is an over-simplified case, yet it illustrates
the idea behind fuzzy logic. With all parameters being used,
weighted averaging can be used to implement a combination of
parameters in the fuzzy model. Fuzzy logic is flexible in allowing
different parameters to carry different weights. In the location
system of the present invention, the weights for the fuzzy logic
averaging will be determined by the volatility of the data (used
the measure of the location band area). In the "gray" areas of
overlapping fuzzy rules, the overlapping rules are added together
(with associated weights) and then the average of the curve will be
used.
By preparing several individual parameter bands to get the smallest
volatility within a "quantization", the best solution may be
determined. Finally, the final solution may be compiled using fuzzy
logic technique values. For example, in the pseudo code above, each
of the database entries is weighted against one another such that
the database entry of minimum volatility having the strongest
predictor of distance at a particular location for particular
values is obtained where more than one rule applies.
As known to those skilled in the art, fuzzy logic is a process
where, unlike neural networks, more than one rule applies. The
rules are averages in a predetermined weighting scheme. Unlike
normal fuzzy logic rules, however, the weighting here pertains to
minimum and maximum values. In keeping with the invention,
volatility is used as an indicator of the best weight. The variable
with the least volatility is weighted the most, however, other
variables are not discounted.
In this manner, overlapping RF measurements may be utilized. Thus,
80% of WER and 20% of RSSI might be used in predicting location.
The system and method of the present invention averages the minimum
distances as well as the maximum distances which then become the
min and max boundaries for each arc segment. This process is
repeated for all other arc segments which permit a min and max
bounding polygon to be drawn around a Radio Port. The process is
thereafter repeated for neighboring Radio Ports as they are "heard"
to determine the most accurate predicted bounding contours for the
other neighboring Radio Ports. The resulting contours (i.e. the
minimum and maximum contours) are thereafter drawn around each
Radio Port, the intersections of which define the bounding polygon
where the mobile unit can be located.
Because the Radio Port data is partitioned into separate
arc-segments and then analyzed, there will be discrete jumps in the
data between arc-segments. To improve the continuity of the data
between arc-segments, a line will be added to help smooth the
jumps. The slope of this line will roughly be the magnitude of the
jump divided by some .DELTA.X (where .DELTA.X is 10-20% of the
width of the arc-segment).
In keeping with the invention, the step of modeling the determined
RF measurements as scaled contour shapes therefore requires
segmenting the coverage areas of each of the base stations into a
plurality of arc segments designated by reference numeral 120 in
FIG. 14. For each of the arc segments 120, a plurality of single or
multiple regressions must be performed so as to convert actual data
into a corresponding plurality of mathematical curve-fit equations
each representing a relationship between a predetermined measurable
variable, i.e. RSSI, WER, etc. and distance from the base station.
For each of the arc segments, the degree of fit must be determined
of the corresponding mathematical equation by comparing each of the
mathematical equations with actual data. The mathematical equations
may thereafter be optimized by determining which has the best
correlation and least standard error for a predetermined portion of
each arc segment 120.
In an alternative embodiment, a Genetic Algorithm (GA) may be used
to optimize the parameters of each of the single or multiple
regressions so as to further improve the degree of fit for greater
correlation and minimum standard error. Still further, in cases
where there is generally poor correlation between all of the
mathematical equations of an arc segment and the actual data, the
corresponding base station may be instructed along with the
receiver, i.e., the mobile unit, to each temporarily change their
transmission frequencies by 10-40 MHz. Thereafter, additional RF
measurements may be obtained for the base station at the changed
frequency, including its link budget, for the same predetermined
plurality of distances and directions. As readily seen, this will
increase the number of variables for consideration and
analysis.
The optimized mathematical equations for each arc segment are
thereafter combined so as to form the scaled contours 124 such as
that shown in the schematic of FIG. 24.
Each scaled contour 124 has minimum and maximum bounds 126 and 128.
After these boundaries have been determined for an entire base
station, minimum/maximum boundaries also define minimum/maximum
contours, based on a given set of real-time measurements in both
the uplink and downlink directions. This process is repeated for
neighboring base stations, and the resulting intersection (if any)
then define a first min/max bounding polygon 130.
It must thereafter be determined which lobes or bandpass filters
are communicating with the mobile unit 94. Next, the orientation of
the communicating lobes/bandpass filters must be determined in
terms of azimuth angles so as to define a second bounding polygon
area 132 similar to that shown, for example, in FIG. 25. Finally,
by determining where the first and second bounding polygon areas
intersect, a location polygon 134 may be defined, as shown in FIGS.
26-27, which more precisely describe the position of the mobile
unit 94 in terms of minimum and maximum error estimate. Once the
location polygon 134 has been defined, its center may be calculated
in terms of latitude and longitude. Finally, the exact street
addresses contained with the location polygon area may be
determined in cooperation with a run-time database. A flow chart of
the method steps is provided in FIG. 28.
FIG. 25 of the drawings depicts a system with 3 SDMA-equipped base
stations, simultaneously providing location estimates for two E-911
users. In normal operation, a caller is served by a single base
station as in conventional systems. Upon detection of an incoming
call, however, network-level software instructs the serving base
station along with its surrounding base stations to supply
estimates of the callers (E-911) relative azimuths. These azimuth
estimates can then be combined with the known locations of the base
stations to produce an estimate of the caller's position, i.e. to
provide the second bounding polygon area referenced above.
As known to those skilled in the art, common frequency planning
strategies deliberately assign distinct channels to neighboring
base stations. SDMA base stations, however, are typically equipped
with frequency agile multi-channel receivers that can rapidly tune
to any of the frequencies in use by the system. An SDMA-equipped
base station can thereby provide a location estimate for an
adjacent base station's user. In contrast, conventional base
stations typically have a single-channel receiver for each of the
frequencies in which they operate, and are therefore incapable of
monitoring an adjacent base station's channels.
SDMA-equipped base stations also have the ability to accurately
locate a mobile unit even though the signal received by the base
station may be too weak to be successively demodulated. This
ability to accurately locate weak signals is inherent in the
advanced signal processing techniques employed at SDMA-equipped
base stations.
To provide a realistic numerical example, suppose that the accuracy
of each base station's azimuth estimate is plus or minus
0.3.degree., and that the three base stations depicted in FIG. 12
lie on the vertices of an equilateral triangle which is 10 km on a
side (i.e. a nominal 5 km range for each base station). The maximum
distance that a caller inside of the triangle could be from any one
of the base stations is then 10 km, and the associated position
estimate would cover an area of roughly 0.008 km.sup.2
(corresponding to a horizontal position uncertainty of roughly 0.1
km or 100 m). The uncertainty is reduced by a factor of nearly 400
from the TDMA example provided above.
Where a smart antenna having floating lobes is utilized, it is
understood that each of the lobes is operative to track and
communicate with the mobile unit on a corresponding communication
channel by changing its orientation within a predetermined
direction range. Thus, after the first bounding polygon area 130
has been determined, it must further be determined which lobes of
the neighboring base stations are in communication with the mobile
unit 94. Armed with this information, the orientation ranges of the
communicating lobes of the neighboring base stations may thereafter
be determined in terms of azimuth angles so as to define the second
bounding polygon area 132 that describes the relative position of
the mobile unit. As above, it may thereafter be determined where
the first and second polygon areas intersect so as to define a
location polygon 134, as shown in FIGS. 26 and 27, which describes
the position of the mobile unit in terms of minimum and maximum
error estimate.
FIG. 26 illustrates the location polygon 134 formed by the
intersection of the first bounding polygon area 130 obtained from
RF measurements and a second bounding polygon area 132 obtained
from bidirectional measurements from a smart antenna lobe. As
indicated above, directional information (azimuth) is obtained to
determine where the lobe is centered. The intersection of the lobe
coverage and the contour is where the mobile unit is located. Thus,
for the example shown of an SDMA-equipped base station, the area
around 45.degree. (specifically, 45.degree. plus or minus half the
width of the lobe) is the location of the mobile unit.
FIG. 27 illustrates a similar location polygon this time formed by
the intersection of the contour and the reception angle of a
bandpass filter. Again, the intersection area, in the example of a
directed bandpass filter, is approximately 180.degree.
(specifically 180.degree. plus or minus half the reception angle of
the bandpass filter).
While the best modes for carrying out the invention have been
described in detail, those familiar with the art to which this
invention relates will recognize various alternative designs and
embodiments for practicing the invention as defined by the
following claims.
* * * * *