U.S. patent application number 13/663184 was filed with the patent office on 2013-05-16 for timing and time information extraction in a radio controlled clock receiver.
This patent application is currently assigned to XW LLC dba Xtendwave. The applicant listed for this patent is XW LLC dba Xtendwave. Invention is credited to Oren E. Eliezer.
Application Number | 20130121399 13/663184 |
Document ID | / |
Family ID | 46800747 |
Filed Date | 2013-05-16 |
United States Patent
Application |
20130121399 |
Kind Code |
A1 |
Eliezer; Oren E. |
May 16, 2013 |
Timing and Time Information Extraction in a Radio Controlled Clock
Receiver
Abstract
A novel and useful system and method for extracting timing, time
and additional information from a broadcast received in a radio
controlled clock (RCC) receiver. The RCC receiver extracts timing
information represented by a known synchronization sequence that is
used for acquisition and tracking purposes. The RCC receiver
extracts time information as a merged 26-bit time information word
linearly coded into 31 bits comprising the number of minutes (or
hours) since the turn of the current century. A minute counter
representing the 26 bits is converted into the date, hour, and
minute. The RCC extracts additional information including the
schedule for the next daylight saving time transition and for an
imminent leap second. The communications protocol optionally
employs error correcting codes to provide protection for data
fields in the frame, which the RCC may use to enhance reception
reliability in the presence of noise and interference.
Inventors: |
Eliezer; Oren E.; (Plano,
TX) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
XW LLC dba Xtendwave; |
Dallas |
TX |
US |
|
|
Assignee: |
XW LLC dba Xtendwave
Dallas
TX
|
Family ID: |
46800747 |
Appl. No.: |
13/663184 |
Filed: |
October 29, 2012 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
13424733 |
Mar 20, 2012 |
8300687 |
|
|
13663184 |
|
|
|
|
61559966 |
Nov 15, 2011 |
|
|
|
Current U.S.
Class: |
375/238 ;
375/324 |
Current CPC
Class: |
G04G 7/02 20130101; G04R
20/10 20130101 |
Class at
Publication: |
375/238 ;
375/324 |
International
Class: |
H04L 7/04 20060101
H04L007/04; H04L 27/22 20060101 H04L027/22 |
Goverment Interests
STATEMENT OF GOVERNMENT INTEREST
[0002] This invention was made with Government support under
National Institute of Standards and Technology under SBIR Grant No.
NB401000-11-04154. The Government has certain rights in the
invention.
Claims
1. A radio receiver method, said method comprising: receiving a
phase modulated (PM), pulse width modulated (PWM)/amplitude shift
keyed (ASK) broadcast signal encoded with phase-modulated time
information frames; extracting said time information frames from
the phase of said received signal; and wherein each said time
information frame includes a synchronization sequence field and a
current time field.
2. The method according to claim 1, wherein each said time
information frame further comprises a daylight-saving time data
field.
3. The method according to claim 1, wherein each said time
information frame further comprises a leap second data field.
4. The method according to claim 1, wherein each said time
information frame further comprises a daylight saving time data
field and leap second data field, which are jointly protected using
an error correcting code.
5. The method according to claim 1, wherein said current time field
indicates the current minute, hour, date and year expressed as a
number of hours from the beginning of the current century.
6. The method according to claim 1, wherein said current time field
indicates the current minute, hour, date and year expressed as a
number of minutes from the beginning of the current century.
7. The method according to claim 1, wherein said current time field
comprises a current time protected using an error correcting
code.
8. The method according to claim 1, further comprising avoiding
using marker symbols in a WWVB legacy protocol to convey time
information.
9. The method according to claim 1, further comprising using marker
symbols in a WWVB legacy protocol only to convey timing information
using said synchronization sequence.
10. The method according to claim 1, wherein said synchronization
sequence field comprises a 14-bit synchronization sequence.
11. A radio receiver method, said method comprising: receiving a
phase modulated (PM), pulse width modulated (PWM)/amplitude shift
keyed (ASK) broadcast signal encoded with phase-modulated time
information frames; extracting said time information frames from
the phase of said received signal; and wherein each said time
information frame includes a synchronization sequence field, a
current time field, daylight savings time data field and leap
second data field, said current time field, daylight savings time
data field and leap second data field protected using error
correcting codes.
12. The method according to claim 11, wherein said current time
field indicates the current minute, hour, date and year expressed
as a number of hours from the beginning of the current century in a
first subfield and the current minute (0-59) in a second
sub-field.
13. The method according to claim 11, wherein said current time
field indicates the current minute, hour, date and year expressed
as a number of minutes from the beginning of the current
century.
14. A radio receiver method, said method comprising: receiving a
phase modulated (PM) broadcast signal encoded with time information
frames; extracting said time information frames from the phase of
said received signal; and wherein each said time information frame
includes a synchronization sequence field and a current time
field.
15. The method according to claim 14, wherein said current time
field is split into a first sub-field indicating the current minute
(0-59) and a second sub-field indicating the hour, date and year
expressed as a number of hours from the beginning of the current
century.
16. The method according to claim 14, wherein said current time
field indicates the current minute, hour, date and year expressed
as a number of minutes from the beginning of the current
century.
17. A radio receiver, comprising: a receiver circuit operative to
receive a phase modulated (PM), pulse width modulation
(PWM)/amplitude shift keyed (ASK) broadcast signal encoded with
phase-modulated time information frames; a circuit operative to
extract said time information frames from the phase of said
received signal; and wherein each said time information frame
includes a synchronization sequence field and a current time
field.
18. The receiver according to claim 17, wherein said current time
field indicates the current minute, hour, date and year expressed
as a number of hours from the beginning of the current century in a
first subfield and the current minute (0-59) in a second
sub-field.
19. The receiver according to claim 17, wherein said current time
field indicates the current minute, hour, date and year expressed
as a number of minutes from the beginning of the current
century.
20. The receiver according to claim 17, wherein each said time
information frame further comprises a daylight saving time data
field.
21. The receiver according to claim 17, wherein each said time
information frame further comprises a leap second data field.
22. The receiver according to claim 17, wherein each said time
information frame further comprises a daylight saving time data
field and leap second data field, both of which are protected using
an error correcting code.
23. The receiver according to claim 17, wherein each said time
information frame further comprises a field to indicate whether
daylight saving time is in effect or not, said indication provided
with an error correcting code.
24. The receiver according to claim 17, wherein said current time
field comprises a current time protected using an error correcting
code.
25. The receiver according to claim 17, wherein marker symbols in a
WWVB legacy protocol are not used to convey time information.
26. The receiver according to claim 17, wherein marker symbols in a
WWVB legacy protocol are used only to convey timing information
using said synchronization sequence.
27. The receiver according to claim 17, wherein said
synchronization sequence field comprises a 14-bit synchronization
sequence.
28. The receiver according to claim 17, wherein the
least-significant bit (LSB) of said current time field is extracted
during a tracking operation.
Description
REFERENCE TO PRIORITY APPLICATION
[0001] This application claims priority to U.S. Provisional
Application Ser. No. 61/559,966, filed Nov. 15, 2011, entitled
"Reception of Time Information and Synchronization Information in a
Radio Controlled Clock," incorporated herein by reference in its
entirety.
FIELD OF THE INVENTION
[0003] The present invention relates to the field of wireless
communications, and more particularly relates to a timing, time and
additional information broadcast protocol for use in a radio
controlled clock receiver.
BACKGROUND OF THE INVENTION
[0004] Radio-controlled-clock (RCC) devices that rely on time
signal broadcasts have become widely used in recent years. A
radio-controlled-clock (RCC) is a timekeeping device that provides
the user with accurate timing information that is derived from a
received signal, which is broadcast from a central location, to
allow multiple users to be aligned or synchronized in time.
Colloquially, these are often referred to as "atomic clocks" due to
the nature of the source used to derive the timing at the
broadcasting side. In the United States, the National Institute of
Standards and Technology (NIST) provides such broadcast in the form
of a low-frequency (60 kHz) digitally-modulated signal that is
transmitted at high power from radio station WWVB in Fort Collins,
Colo. The information encoded in this broadcast includes the
official time of the United States.
[0005] Similar services operating at low frequencies exist in other
regions worldwide, including Europe and Japan. Many consumer-market
products exist, including watches, alarm clocks and wall clocks,
that are capable of receiving one or more of these broadcasts and
which can display the correct time to within approximately one or
two seconds of accuracy. While the broadcast may be active
continuously, a typical radio-controlled clock may be set to
receive the broadcast only once a day. Such reception, if
successful (depending on the condition of the wireless link and
potential interference), is typically used to reset the timekeeping
device, such that if it were set incorrectly or has drifted away
from the correct time, it will be set in accordance to the time
communicated through the broadcast signal.
[0006] Reception of the time signal, however, is being challenged
by a growing number of sources of electromagnetic interference. In
particular, the on-frequency interference from the MSF radio
station in the United Kingdom has been identified as a particularly
challenging jammer for receivers on the East Coast.
[0007] There is thus a need for an improved protocol for time
signal broadcasts, such as that provided by WWVB in the United
States and radio stations in other countries, that attempts to
cost-effectively address the reception challenges. Such a new
protocol should preserve existing amplitude modulation properties
of the transmitted signal, in order to maintain backwards
compatibility and not impact existing devices.
SUMMARY OF THE INVENTION
[0008] A novel and useful method for extracting timing, time and
additional information from a broadcast communications protocol for
use in a radio controlled clock (RCC) receiver. Information is
transmitted in 60 second frames. The RCC receiver extracts timing
information represented by the phase of a known synchronization
sequence that is used both for time-acquisition and for
timing-tracking purposes in the RCC. The RCC receiver also extracts
time information including hour, minute, date and year information
that are represented in a merged time field comprising the number
of minutes (or hours) since the turn of the current century. The
RCC also extracts additional information from the received frame,
including daylight-saving time (DST) schedule and leap second
information. The communications protocol also optionally uses error
correcting codes to provide protection for one or more data fields
of the 60 second frame.
[0009] There is thus provided, in accordance with the invention, a
radio controlled clock (RCC) method comprising (1) receiving a
phase modulated (PM), pulse width modulated (PWM)/amplitude shift
keyed (ASK) broadcast signal encoded with time information frames,
(2) extracting the time and timing information frames from the
phase of the received signal, and wherein each of the time
information frame includes a synchronization sequence field and a
current time field, and (3) responding to the extracted information
by appropriately setting the time and date information in the
time-keeping device and by adjusting its timing.
[0010] There is also provided, in accordance with the invention, a
radio receiver method comprising (1) receiving a phase modulated
(PM), pulse width modulated (PWM)/amplitude shift keyed (ASK)
broadcast signal encoded with time information frames, (2)
extracting the time information frames from the phase of the
received signal, wherein each the time information frame includes a
synchronization sequence field, a current time field, daylight
savings time data field and leap second data field, and wherein the
current time field, and the field designated to daylight saving
time and leap second scheduling are protected using an error
correcting code, (3) optionally employing error detection and
correction in the received fields in order to enhance the reception
reliability, and (4) providing a time-keeping device with the
extracted information.
[0011] There is further provided, in accordance with the invention,
a radio receiver method comprising (1) receiving a phase modulated
(PM) broadcast signal encoded with time information frames, (2)
extracting the time information frames from the phase of the
received signal, wherein each the time information frame includes a
synchronization sequence field and a current time field, and (3)
providing a time-keeping device with the extracted information for
use in a time-based control system, such as an irrigation or
heating system, or in a wrist-watch, wall-clock or night-table
alarm-clock, or in any other device where there is need for either
the absolute official time or for timing synchronization amongst
multiple devices, or for both.
[0012] There is also provided, in accordance with the invention, a
radio receiver, comprising a receiver circuit operative to receive
a phase modulated (PM), pulse width modulation (PWM)/amplitude
shift keyed (ASK) broadcast signal encoded with time information
frames, a circuit operative to extract the time information frames
from the phase of the received signal, wherein each time
information frame includes a synchronization sequence field and a
current time field.
BRIEF DESCRIPTION OF THE DRAWINGS
[0013] The invention is herein described, by way of example only,
with reference to the accompanying drawings, wherein:
[0014] FIG. 1 is a high level block diagram illustrating an example
timing and time information transmitter of a system operating in
accordance with the present invention;
[0015] FIG. 2 is a high level block diagram illustrating an example
timing and time information receiver constructed in accordance with
the present invention;
[0016] FIG. 3 is a diagram illustrating an example embodiment of
phase modulation, shown at baseband, added to a pulse width
amplitude modulated carrier;
[0017] FIG. 4 is a diagram illustrating the signal space
representation of the prior art AM/pulse-width `0` and `1` signals,
as well as that of an example embodiment of the present invention,
where Phase-Reversal-Keying (PRK) is added onto the AM/pulse-width
modulation;
[0018] FIG. 5A is a diagram illustrating a first example waveform
of phase modulation added to an amplitude/phase modulated carrier
in an example communication protocol;
[0019] FIG. 5B is a diagram illustrating a second example waveform
of phase modulation added to an amplitude/phase modulated carrier
in an example communication protocol;
[0020] FIG. 6 is a diagram illustrating a first example time
information frame structure incorporating timing, time and
additional information;
[0021] FIG. 7 is a diagram illustrating a second example time
information frame structure incorporating timing, time and
additional information;
[0022] FIG. 8 is a diagram illustrating a third example time
information frame structure incorporating timing, time and
additional information;
[0023] FIG. 9 is a diagram illustrating a fourth example time
information frame structure incorporating timing, time and
additional information;
[0024] FIG. 10 is a diagram illustrating a fifth example time
information frame structure incorporating timing, time and
additional information;
[0025] FIG. 11 is a graph illustrating the word error rate
comparison for coded and uncoded versions of the time data
field;
[0026] FIG. 12 is a graph illustrating the word error rate
comparison for coded and uncoded versions of the DST and leap
second indicators;
[0027] FIG. 13 is a graph illustrating the word error rate
comparison for coded and uncoded versions of the DST schedule word;
and
[0028] FIG. 14 is a block diagram illustrating an example parallel
to serial conversion of AM and PM bits.
DETAILED DESCRIPTION OF THE INVENTION
[0029] The system and method of the present invention is a receiver
that receives, demodulates and decodes a broadcast signal, whose
modulation and encoding of time and timing information,
representing the novel communications protocol of the present
invention, allow for reliable and power-efficient operation. The
communication protocol of the present invention is adapted to allow
for prior art devices, operating in accordance with an historical
communication protocol, to be unaffected by the changes introduced
by the protocol of the present invention. Devices that are adapted
to operate in accordance with the present invention will benefit,
however, from various performance advantages. These advantages
include greater robustness of the communication link, allowing
reliable operation at a much lower
signal-to-noise-and-interference-ratio (SNIR), greater reliability
in providing the correct time, and reduced energy consumption,
resulting in extended battery life in battery-operated devices.
[0030] In accordance with one embodiment of the present invention,
only phase modulation is applied to the carrier thus allowing
existing devices that operate in accordance with the legacy
communication protocol and rely on envelope-detection based AM
demodulation to continue to operate with the modified protocol
without being affected. Although this backward compatibility
property of the communication protocol of the present invention may
represent a practical need when upgrading an existing system, the
scope of the invention is not limited to the use of this combined
modulation scheme and to operation in conjunction with an existing
communication protocol.
[0031] Legacy receivers are typically adapted to receive the legacy
pulse width modulated/amplitude modulation through relatively
simple envelope detection, which is unaffected by the phase
modulation of the present invention. This is because there are no
phase transitions introduced during the high-amplitude portion
pulses, which could result in bandwidth expansion and consequently
in reduced power passing through the narrowband filtering used in
these receivers.
[0032] The enhanced robustness offered by the present invention,
resulting in reliable reception at lower SNIR values with respect
to those required for proper operation of prior art devices, is a
result of the use of (1) a known synchronization sequence having
good autocorrelation properties; (2) coding that allows for error
detection and correction within the fields of information bits that
are part of each data frame; and (3) a superior phase modulation
scheme (e.g., binary phase shift keying (BPSK)). The BPSK
modulation, in particular phase reversal keying (PRK), representing
an antipodal system, is known to offer the largest distance in the
signal space with respect to the signal's power, whereas the
historical modulation schemes that are used for time broadcasting
worldwide are based on pulse width modulation that relies on
amplitude demodulation, requiring a higher SNIR to achieve the same
decision error probability or bit error rate (BER) in the presence
of additive white Gaussian noise (AWGN).
[0033] Furthermore, the AM/pulse-width based prior art system is
highly susceptible to on-frequency interference of the type that is
experienced as a result of intentional and non-intentional
emissions that may be received by the receiver and interfere with
its operation. In contrast, the modulation scheme and method of
reception of the present invention allow for operation in the
presence of relatively strong on-frequency interference.
[0034] Further enhanced performance in particularly low SNIR
conditions is achieved in the system of the present invention
through the accumulation of multiple frames, each of which spans
one minute. For example, in one embodiment of the present
invention, the contents of 60 consecutive frames in one hour differ
only in the 6-bit field dedicated to the encoding of the minutes
and the 5-minute field dedicated to the parity of the time word.
The results from demodulating all other bits in the frame may be
averaged to arrive at a 60.times. (18 dB) improvement factor in the
SNR.
[0035] The time information recovered from the signal received
through the antenna is typically used in a radio-controlled-clock
(RCC) device to either initialize or track the time in the
application based on the RCC. Note that this information, at least
initially, may be made available to the application from sources
other than the broadcast, such as through factory setting, manual
setting by the user, or some form of conveying the time to the
application either wirelessly or through a physical connection that
alleviates the need to receive this information from the broadcast
signal. Such means for initializing the time in the application are
particularly beneficial in scenarios where the reception of the
broadcast may be marginal and the recovery of such information from
it may be difficult, whereas time-tracking, based on correlating
against the known synchronization word in the frame, is possible
even at much lower SNIR values.
[0036] Information may also need to be recovered from the broadcast
signal in order to verify the schedule for the upcoming
daylight-saving-time (DST) transition or for other information that
may be embedded in the frame. This may, however, be very infrequent
or not necessary at all, if the application does not need such
additional information and only needs to maintain synchronicity
with other devices and/or the broadcast which represents the
accurate time.
[0037] The timing information embedded in the broadcast may be
extracted from it through a correlation operation against a known
synchronization sequence. Such operation is more robust in nature
and may be successful at SNIR values that are considerably lower
than those necessary for reliable recovery of individual
information bits. Therefore, the extraction of the timing
information from the signal, which the application may require
periodically in order to compensate for natural drifts in its
internal timing source, can often be accomplished even in scenarios
where the recovery of new information may not be possible. The
frequency at which the application may turn on the receiver and
attempt to extract the timing information depends on its needs and
can vary from multiple times per day to once a week or even more
infrequently.
[0038] The timing information may be extracted through the
reception of a portion of the frame, rather than a whole frame,
allowing the device to minimize the energy consumption associated
with this operation.
[0039] Once the time information has been acquired and is known at
the receiving end, the information fields may be used in addition
to the known synchronization fields for the purpose of determining
timing through the use of correlation. If, for example, a device
that is already set to the correct time (including the minute)
initiates a reception for the purpose of adjusting for a few
seconds in possible timing drift experienced by it, it may
correlate the received signal against the contents of an entire
frame, such that if the synchronization sequence is allocated 12
seconds, for example, and the entire frame is of 60 second
duration, a gain of 5 (7 dB) is obtained through such extended
correlation. This allows for reasonably accurate timing corrections
to be performed at extremely low SNIR values, for which the
probability of error in the recovery of individual 1-second bits
would be intolerably high.
[0040] Note that an RCC receiver incorporating the system and
methods of the present invention may be implemented in any type of
timekeeping device. The timekeeping device may comprise a watch; an
alarm clock; a wall clock, a utility meter; a microwave oven; a car
radio that can display the time; a timekeeping device that acquires
its initial timing (sets its time) based on the reception of a
phase-modulated data frame where coding is employed to allow for
the detection and correction of errors in the time information;
wherein the device can correct at least one bit that is received in
error; or where the detection of one error or more in the received
frame results in a reception of at least part of the next frame, in
order to verify that the time information extracted from the first
frame is correct; a timekeeping device that determines anytime
during the daylight-saving time (DST) period when the DST period is
about to end, based on extracting that data from a coded field the
broadcast, which serves to support several possible schedules, and
schedules a minus one hour (or plus one hour) correction to the
appropriate instance (typically 2 AM on a particular Sunday),
without having to receive the time around the instance of
transitioning out of or into DST.
[0041] The timekeeping device may comprise a timekeeping device
that determines before the DST period when the DST is about to
start, based on extracting that data from a coded field in the
broadcast, which serves to support several possible schedules, and
schedules a plus one hour correction to the appropriate instance
(typically 2 AM on a particular Sunday becomes 3 AM), without
having to receive the time around the instance of transitioning
into DST.
[0042] The timekeeping device may schedule its reception window to
an instance in time where a known sequence is to be received,
preferably with good autocorrelation properties, allowing the
receiver to synchronize with it, i.e. determine the timing of the
received signal through a correlation operation and adjust the
device's time accordingly, thereby compensating for drifts that may
have been experienced in it since the last synchronization
opportunity, where the width of the reception window and the
corresponding duration of the correlation operation are limited
based on the estimated drift that is to be compensated for, such
that power consumption associated with the reception is minimized;
or where the width of the reception window and the corresponding
duration of the correlation operation are limited based on the
signal-to-noise-and-interference (SNIR) conditions, such that
sufficient signal energy is involved in the correlation operation
to allow for adequate synchronization, while avoiding overly
extended durations that may result in excessive power consumption
in the receiver.
[0043] A high level block diagram illustrating an example timing
and time information transmitter system operating in accordance
with the present invention is shown in FIG. 1. The equipment at the
transmitter end, generally referenced 10, comprises a high accuracy
clock source (frequency source) 12 from which a clock signal
(timing information) is derived, a time-code-generator 14 having
user-interface 16, a source of time data 13, a transmitter 18
generating a TX signal 19 and coupled to transmitting antenna
11.
[0044] The time code generator 14 keeps track of time based on the
high-accuracy frequency source input to it from source 12,
constructs the frames of data representing the time information
received from time data source 13 and other information that is to
be transmitted, modulates the data frames onto the RF carrier in
accordance to a defined protocol and allows time initialization and
other controls to be set in it through its user interface 16. The
transmitter 18 amplifies the modulated signal to generate an output
TX signal 19 at the desired levels, e.g., 50 kW, and drives the
antenna 11 that is used for the wide-coverage omnidirectional
broadcasting of the signal.
[0045] A high level block diagram illustrating an example
timekeeping device constructed in accordance with the present
invention is shown in FIG. 2. Typically, the timekeeping device is
incorporated into low cost consumer market products, but may be
implemented in any device that requires a precision time reference.
The timekeeping device, generally referenced 20, comprises
receiving antenna 21, receiver module 24 operative to receive RX
signal 22, processor and controller 26, timekeeping function 30,
internal or external clock source 31, display 32 and user interface
34.
[0046] In one embodiment, the timekeeping RCC device, whose
receiver demodulates one or more of the phase-modulation schemes
BPSK/QPSK/MSK, is fabricated using CMOS technology and may be
incorporated into a larger SoC that could comprise functionality
beyond that of the RCC and timekeeping functions.
[0047] The receiver module 24 extracts timing and time information
from the received signal 22, in accordance with the modulation
scheme and protocol in use (described in more detail infra), and
provides the processing and control function 26 with the extracted
timing and time information. Controller function/processor 26
appropriately enables/disables the operation of the receiver module
through control line 28 such that it is limited to the intervals of
interest to minimize energy consumption in those applications where
it may be critical to do so (e.g., wrist watches). The timekeeping
function 30 keeps track of the time based on pulses provided by
clock source 31 having limited accuracy. Note that the clock source
31 may comprise any suitable clock source or clock signal such as a
crystal oscillator and may be provided internal to the timekeeping
device 20 or supplied from a source external to the timekeeping
device.
[0048] The timekeeping may be adjusted by the processor/controller
in accordance with an estimated drift at a specific instant, which
is either measured or calculated or a combination of the two. The
display function 32 may be used to display the time as well as
various indications to the user, including reception quality,
estimated bound for error in displayed time, battery status, etc.
The user interface function 34, based on pushbuttons,
slide-switches, a touch-screen, keypad, computer interface, a
combination therefrom, or any other form of human interface, may be
used to set the initial time, define the maximal allowed timing
error, the time-zone according to which time is to be calculated,
the use of daylight saving time, etc.
[0049] In one embodiment of the invention, the timekeeping device
is operative to extract timing and time information conveyed in a
broadcast signal. Timing information denotes information related to
synchronization and tracking and is also used for bit and frame
synchronization. Time information denotes information related to
the current time being communicated, such as the date and the time
of day (hours and minutes), as well as scheduled events, such as an
upcoming DST transition, leap second, etc.
[0050] Typical currently available time-broadcast signals employ
some form of amplitude modulation combined with some form of pulse
width modulation (PWM) to send binary data bits. As an example
consider the WWVB signal broadcast from Fort Collins, Colo. in the
United States of America. The WWVB signal comprises a 60 second
frame consisting of 60 bits. Each bit, of one second duration, is
sent as a pulse width modulated signal wherein the carrier is
transmitted at a low amplitude or a high amplitude for different
portions of the bit. The frame also consists of several marker bits
spread out evenly through the frame, which serve only to indicate
timing and do not convey time information. The existing WWVB system
transmits a pulse-width modulated amplitude-shift keyed waveform on
a 60 kHz carrier. The one-second duration `0` and `1` symbols are
represented by a power reduction of -17 dB at the start of the
second for 0.2 s and 0.5 s, respectively.
[0051] The invention also comprises a digital phase modulation (PM)
receiver operative to perform phase demodulation on a signal that
comprises both phase modulation and amplitude modulation, said
amplitude modulation resulting in a portion of the symbol being
transmitted at a low level while the remaining portion is
transmitted at a higher level.
[0052] The low level may be zero, thus reducing the amplitude
modulation to on-off-keying (OOK). The symbol time may be one
second and the low-level portion of the transmitted symbol is its
first 0.1 seconds, 0.2 seconds, 0.5 seconds, or 0.8 seconds.
[0053] The phase demodulation operation may be limited to a fixed
portion of the symbol that is expected to have the high-amplitude.
The fixed portion may be the second half of the symbol, thus
simplifying the receiver implementation. The duration of the
portion of high-amplitude may be predicted based on the knowledge
of the exact time, allowing the receiver to time its demodulation
duration to the portion of the symbol that is of high amplitude,
thus maximizing the energy involved in the phase demodulation
operation. Further, the phase modulation may comprise, for example,
binary phase shift keying (BPSK), minimum-phase-shift-keying (MSK),
phase reversal keying (PRK), quadrature phase shift keying (QPSK),
and frequency shift keying (FSK).
[0054] The invention also comprises the optimal scheduling of
reception instances whereby the RCC establishes, through a learning
process, which instances work best for timing adjustments (e.g., 2
AM at night) and at some point moves into steady-state mode, where
it only receives at those instances. If receptions deteriorate over
time, the RCC may return to the learning mode and attempt other
instances too.
[0055] The invention also comprises the extraction of an advance
schedule for DST transition from the bits designated to that in the
frame by using a look-up-table that converts the 6-bit word
representing the selected schedule into its actual meaning (e.g.,
mapping to the first Sunday in November at 2 AM, or whatever the
meaning may be).
[0056] A timekeeping device that wirelessly acquires and tracks the
time provided by a digital broadcast and the protocol of that
broadcast, defined by its data frame structure and modulation
scheme, is operative to allow for superior performance of the
timekeeping devices in terms of range of operation, immunity to
interference, ability to operate with lower cost antennas and
reduced energy consumption.
[0057] The protocol is designed to allow for adaptive operation in
the receiver wherein the acquisition and tracking operations may
extend over different durations in accordance with reception
conditions such that in very low signal-to-noise-and-interference
(SNIR) conditions the receiver may extend its duration of reception
to accumulate greater amounts of signal energy, thereby allowing it
to reliably extract time and timing information from the broadcast
signal.
[0058] In one embodiment, error correction coding is employed to
allow for various levels of error detection and correction in the
various fields of a transmitted packet. A dedicated field in the
broadcast signal is decoded in a nonlinear fashion, providing the
receiver with a reliable indication of when the next
daylight-saving time (DST) transition is to take place. This allows
it to employ very infrequent periodic receptions of the full frame
of the broadcast signal (e.g., once every few months), whenever
energy savings are needed, and to limit its more frequent periodic
operation to a short known portion of the frame for the purpose of
fine timing corrections.
[0059] In one embodiment, the radio controlled clock (RCC) receiver
is operative to receive and decode a signal transmitted in
accordance with the protocol described supra. Such a receiver is
operative, inter alia, to extract the DST schedule from the DST
related specific bits within the frame, decode the bits using the
look-up-table provided supra and schedules the one-hour adjustment
accordingly.
[0060] In one embodiment of the invention, phase modulation is
added to an amplitude modulated carrier. A diagram illustrating
phase modulation added to an amplitude modulated carrier in an
example communication protocol is shown in FIG. 3. This diagram
describes the amplitude/pulse width modulation (PWM) used in the
historical WWVB broadcast as well as the phase modulation
introduced in accordance with an embodiment of the present
invention. The diagram shows the baseband representation of the `0`
and `1` symbols in both the historical WWVB modulation and in one
that is modified in accordance with an example embodiment of the
present invention. It is noted that the enhancement in the
communication protocol offered by the present invention, in the
form of independently defined phase modulation and the use of a
known synchronization sequence, is not limited to the broadcast of
WWVB and may be applied to other timing/time information broadcast
systems such as those in other countries around the world where
similar AM/pulse-width schemes are used or where no AM/pulse-width
modulation needs to be supported, allowing for continuous BPSK to
be used. Note that the receiver may be operative to receive and
decode any phase over amplitude modulated transmitted signal that
has time synchronization and time information conveyed therein.
[0061] In one embodiment, the additional phase modulation added to
the signal is binary phase shift keying (BPSK) having an
180.degree. difference in the carrier's phase between the `0` and
`1` symbols, also known as antipodal phase modulation or Phase
Reversal Keying (PRK). Hence, the modulated waveforms representing
these symbols may be expressed as the products of the sinusoidal 60
kHz carrier (in the case of WWVB) and the baseband waveforms
s.sub.0(t)=x.sub.0(t) (waveform 80) and s.sub.1(t)=-x.sub.1(t)
(waveform 84), respectively, as shown in FIG. 3. Waveform 82
represents the original `1` symbol s.sub.1(t)=x.sub.1(t) that is
replaced by its inverse waveform 84 in one example embodiment of
the present invention. As is shown in FIG. 3, the enhanced
modulation scheme can be accomplished through simple sign inversion
for the waveform representing the `1` symbol. It is noted that
since the existing envelope detector based receivers designed to
receive and decode the current WWVB AM/PWM based broadcast signal
do not consider the carrier's phase, they are not impacted by the
modification of phase inversion of the `1` symbol.
[0062] A diagram illustrating the signal space representation of AM
only and PM over AM `0` and `1` symbols is shown in FIG. 4. As
shown in the diagram, the new pair of waveforms, x.sub.0
(referenced 88) and -x.sub.1 (referenced 86), having the same
amount of energy (corresponding to their distances from origin),
exhibit a much greater distance between the `0` and `1` symbols (as
compared to waveform pair x.sub.0 and x.sub.1 (referenced 90),
thereby allowing for more robust reception in the presence of
additive noise. Note that the existing symbols x.sub.0 and x.sub.1
are strongly correlated, i.e. they have a very short distance
between them in the signal space with respect to their
energies.
[0063] The Euclidean distance between the two amplitude modulated
waveforms x.sub.0 and x.sub.1 is shown to be 0.47, whereas the
Euclidean distance for the two phase modulated waveforms x.sub.0
and -x.sub.1 increases to 1.55. Therefore, the modulation gain
(denoted m.sub.g) representing the power ratio by which the
detection capability in the presence of additive noise is improved,
is given by
m g = 20 log 10 ( 1.55 0.47 ) = 10.36 dB ( 1 ) ##EQU00001##
[0064] Thus, by simply adding such phase modulation, an order of
magnitude of improvement may be achieved when assuming additive
white Gaussian noise (AWGN). This analysis implicitly assumes that
the receivers for both schemes would be optimal, i.e. based on
correlation or matched filtering. In practice, the BPSK receiver
may be implemented digitally in a near-optimal fashion, whereas the
receivers for the existing AM/pulse-width scheme found in
consumer-market products, not designed as a classical
digital-communications system, are based on envelope detection, as
previously noted. This adds an additional gap of 2 to 4 dB between
the two when only AWGN is considered. In the presence of
on-frequency interference, however, the gain offered by realizing a
near-optimal BPSK receiver may be arbitrarily higher. Furthermore,
additional gains can be offered, such as (1) through encoding of
the information, and (2) use of a known synchronization
sequence.
[0065] In an embodiment of the present invention, the information
represented by the phase modulation in each bit is independent from
that represented by the existing (legacy) AM/pulse-width
modulation, such that an inverted phase would not necessarily be
tied to the shorter waveform 82, represented by inverted waveform
-x.sub.1(t) 84 in FIG. 3. In an example embodiment, with
independent data being communicated through the carrier's phase, a
phase inverted bit, which may represent a "1", for example, may be
combined with either a "0" or a "1" in the AM/PWM signal.
[0066] The receiver extracting the information from the phase may
limit the phase demodulation operation to the last 0.5 sec of each
bit, where both the "0" and "1" symbols of the AM/PWM scheme shown
in this example are at high amplitude. Alternatively, in order to
gain from the additional energy in the longer "0" pulses (0.8 sec
in this example), the receiver may extend the demodulation of phase
during those symbols to 0.8 sec when the content is of the AM/PWM
modulation is known to be "0". In the existing WWVB protocol, for
example, there are several such bits fixed at "0". Additionally,
when a device operating in accordance with the present invention
has already acquired the time and is tracking it, its reception of
the phase modulated information may consider the predicted
durations of the time-information bits as they are defined by the
particular AM/PWM protocol, thereby further optimizing
reception.
[0067] In an alternative embodiment, non-antipodal phase modulation
can be used to modulate the PWM signal. For example, the magnitude
of phase modulation applied may be set at any value less than
180.degree., e.g., .+-.45.degree., .+-.25.degree., .+-.13.degree.,
etc. Use of a lower value such as .+-.13.degree. ensures that the
modulated signal, even if the rate of phase modulation were
significantly increased, is contained within a narrow bandwidth and
does not escape the narrow filtering in typical existing AM
receivers, which is on the order of 10 Hz. Note that such
narrowband PM is not comparable in performance to antipodal BPSK,
where the two symbols are 180.degree. apart exhibiting a
correlation factor of -1.
[0068] A diagram illustrating a first example phase modulation
added to an amplitude modulated carrier in an example communication
protocol is shown in FIG. 5A. The waveform illustrates three
consecutive example bits in the transmission as a time-domain
waveform 150. The three bits 152, 154 and 156 each span a duration
of one second. Each of the one second bits is divided into a first
portion 160 for which the carrier power is low and a second portion
162 for which the carrier power is high. In the WWVB protocol, the
information in each bit depends on the durations of these two
portions with an even 0.5/0.5 sec partition representing a "1" bit,
and the uneven 0.2/0.8 sec partition representing a "0" bit. A
0.8/0.2 sec partition represents a `marker` bit, which may be used
for timing identification, but does not carry information. The bits
represented under the legacy PWM/AM modulation are indicated at the
top portion of the diagram. For example, the three PWM/AM bits
shown are "1", "0" and "1".
[0069] In accordance with an embodiment of the present invention,
information is added to the existing modulation using BPSK
modulation. A "1" is represented by a carrier having an inverted
phase, with the phase inversion 158 occurring at the beginning of
the bit, as shown for the third bit 156 at t=2 sec. It is noted
that the phase inversion may also be performed at any other
instance, e.g., during the low amplitude portion of the carrier, if
the receiver's phase demodulation operation is limited to the
high-amplitude duration and disregards the low amplitude portion.
While the information represented by the pulse widths is shown to
be "1", "0", "1", the information that is sent in parallel, in
accordance with the example BPSK (or PRK) protocol of the present
invention, is shown to be "0", "0", "1" (as shown along the bottom
portion of the diagram). Note that there is not necessarily any
relationship between the bit pattern transmitted using PWM/AM and
that transmitted using PM as they can be completely independent. It
is noted that the carrier frequency is not shown to scale in the
figure to enhance clarity, but it is preferable for the phase
transitions to occur at zero crossing instances of the carrier.
[0070] A diagram illustrating a second example phase modulation
added to an amplitude modulated carrier in an example communication
protocol is shown in FIG. 5B. The waveform illustrates four
consecutive example bits in the transmission as a time-domain
waveform 151. The four bits each span a duration of one second.
Each of the one second bits is divided into a first portion for
which the carrier power is low and a second portion for which the
carrier power is high. The bits represented under the legacy PWM/AM
modulation are indicated in waveform 153 in the middle portion of
the diagram. For example, the three PWM/AM bits shown are "0", "1",
"0" and "1". The same bit pattern is represented in the phase
modulation over PWM/AM waveform 155 shown in the lower portion of
the diagram. As shown, the phase of the carrier is inverted for the
"1" bits. Note that the data transmitted using legacy PWM/AM may be
completely independent of the data transmitted using PM. In this
example they are the same.
[0071] The diagram in FIGS. 5A and 5B describe the
amplitude/pulse-width modulation used in the historical WWVB
broadcast, as well as the phase-modulation introduced in accordance
with the present invention. It is noted that the enhancement in the
communication protocol offered by the present invention, in the
form of independently defined phase modulation and the use of a
known synchronization sequence, is not limited to the broadcast of
WWVB and may be applied worldwide, where similar AM/pulse-width
schemes are used or where no AM/pulse-width modulation needs to be
supported, allowing for continuous BPSK to be used.
[0072] The phase modulation added to the amplitude modulation may
comprise any suitable type of phase modulation including, for
example, BPSK, DBPSK, PRK, PM, MSK, and FSK. In addition, the
underlying amplitude modulation may comprise any type of amplitude
modulation, including, for example, ASK, AM, SSB, QAM, pulse
position modulation (PPM), pulse width modulation (PWM), OOK, and
ASK.
[0073] A diagram illustrating a first example time information
frame structure incorporating timing, time and additional
information is shown in FIG. 6. The time information frame,
generally referenced 100, comprises a synchronization sequence
field 104, a current time data field 106, other time related data
field 108 and optional error correction code (ECC) field 110. In
one embodiment, each time information frame spans 60 seconds.
Frames having other durations are possible as well.
[0074] The synchronization sequence field 104 comprises a known
synchronization sequence (e.g., barker code, modified barker code,
pseudo random sequence, or any other known word or bit/symbol
sequence) at a known timing within the one minute time information
frame of 60 bits that is transmitted every 60 seconds. Note that in
alternative embodiments the synchronization sequence may be placed
within a frame N such that it overlaps or straddles the frame N-1
before it or frame N+1 after it.
[0075] The current time data field 106 may consist of a merged date
and time of day field or may be broken down into individual
sub-fields used to indicate date, year, hours, minutes, etc. The
other time related data field 108 may include zero or more fields
used to indicate, daylight savings time start, leap second
information, etc.
[0076] A diagram illustrating a second example time information
frame structure incorporating timing, time and additional
information is shown in FIG. 7. The time information frame,
generally referenced 120, comprises a synchronization sequence
field 122, an hour data field 124, minute data field 126, optional
hour/minute ECC field 128, daylight savings time (DST) data field
130, leap second data field 132 and DST/leap second ECC field 134.
In one example, the synchronization sequence field 122 spans
14-bits; the combined hour data field 124, minute data field 126
and hour/minute ECC field 128 span 31-bits in a merged time data
field; and the DST data field 130, leap second field 132 and
DST/leap second ECC field 134 comprising an additional information
field, spans 11-bits in total.
[0077] A diagram illustrating a third example time information
frame structure incorporating timing, time and additional
information is shown in FIG. 8. The time information frame,
generally referenced 200, comprises a synchronization sequence
field 202, a minute data field 204, optional minute ECC field 206,
daylight savings time (DST) data field 208, leap second data field
210 and DST/leap second ECC field 212. In one example, the
synchronization sequence field 202 spans 14-bits; the minute data
field 204 and minute ECC field 206 span 31-bits in a merged time
data field; and the DST data field 208, leap second field 210 and
DST/leap second ECC field 212 comprising an additional information
field, spans 11-bits in total.
[0078] In the above two versions of the time information frame, the
synchronization sequence comprises a 14-bit known sequence. The
receivers use this sequence to acquire initial synchronization such
as when first powering on. Receivers also use the sequence to track
the synchronization timing signal broadcast in the frames in order
to adjust their internal timekeeping to maintain synchronization
with the transmitter.
[0079] As described supra, the addition of phase modulation to the
legacy amplitude/pulse-width modulation provides significant
performance improvements in the presence of AWGN and RFI. The
system benefits even further by representing different information
in the phase modulation from that conveyed in the historical
amplitude/pulse-width modulation. In order to maximize such
benefits, common usages of the received signal have been
considered, with the following assumptions being made.
[0080] In a first function, which may be called time-acquisition or
more simply acquisition, the received signal functions to convey
the time of day (and date) to those devices that have not yet
acquired it, such as a new wall-clock which has just had its
batteries installed. In this scenario, the greatest amount of
completely unknown data is assumed to be conveyed, for which the
greatest receiver effort may be expected. This data is referred to
time information. Once the current time is acquired, the RCC device
uses its own timekeeping capability and does not need to repeat the
acquisition process.
[0081] In a second subsequent function, which may be called
timing-tracking or more simply tracking, devices that have already
acquired the current time periodically rely on the received signal
to compensate for whatever time-drifts that may have accumulated
due to the inherent frequency error in their internal or external
clock sources (e.g., crystal based oscillator with frequency
accuracy typically on the order of .+-.10 ppm). This would,
therefore, be the most common use of the WWVB signal, since an RCC
device may require very few acquisition operations in its lifetime,
but would regularly depend on the periodic time-adjustments based
on the WWVB signal (i.e. tracking). This information is the
broadcast signal is referred to as timing information.
[0082] In a third function, which may be called event-scheduling,
advance notification of the next daylight saving time (DST)
transition, i.e. either when entering or exiting DST, is extracted
from the broadcast signal, allowing the receiving device to perform
the one-hour time shift at the correct instance without having to
receive the WWVB signal around the time of the scheduled
transition. In devices that display the actual time and in control
systems that operate in accordance with it (e.g., pool controllers,
irrigation systems, heating/air-conditioning controllers), it is
important for the correct time to be considered and hence useful to
accommodate this third function. Devices that simply need to
maintain synchronicity with one another, however, may not need
this. Other information that falls under this category includes the
advance notification of an imminent leap second, wherein an
indication of the possible presence of a leap second at the end of
the current half-year is extracted from the signal. The DST
schedule and leap second notification information are referred to
as additional or other information.
[0083] For each of the three functions (or types of information,
i.e. timing, time and additional information) described above an
efficient and robust way for representing the information is
provided. A diagram illustrating a third example time information
frame structure incorporating timing, time and additional
information is shown in FIG. 9. The time information frame,
generally referenced 140, comprises 60-bits, where each bit is
allocated one second of the 60-second frame. The bit designations
for both historical amplitude/pulse-width modulation 142 and phase
modulation 144 are shown. In this embodiment, the time information
is divided into a 26 bit hour and minute word that is encoded into
a 31-bit field using a linear block code.
[0084] Note that the term phase modulation may denote either (1)
phase modulation over (i.e. combined with) the historical
amplitude/pulse-width modulation or (2) pure phase modulation
alone. Note also that the data fields assigned to each bit in the
frame for both amplitude and phase modulation are essentially
completely independent. In one embodiment, aspects of the
historical amplitude/pulse-width modulation, however, are taken
into consideration. For example, the assignment of data to the
marker bits of the amplitude/pulse-width modulation is avoided due
to the shorter high power duration (0.2 s) of these symbols.
[0085] The representation of the information is provided under the
constraint of the same 60-second frame defined in the historical
WWVB protocol, while also considering the marker symbols in it,
which are of lesser energy, having a duration of only 0.2 s of high
power. For this reason, the use of the markers is avoided entirely
in conveying bits of information and is limited in their use to
known components of a synchronization sequence or to redundant
information. While three of the seven markers are within the fixed
synchronization sequence, the remaining four marker symbols are
shown to be reserved (denoted `R` 146) for future use.
[0086] Extracting timing from a digitally modulated received signal
is best accomplished when a known sequence, having good
autocorrelation properties, is embedded within it. This allows for
a correlation operation in the receiver to reveal the timing of the
received signal even in low SNIR conditions, for which the recovery
of individual bits within the sequence might have involved high
error probabilities. The successful identification of the known
sequence does not require the recovery of the individual bits
comprising it, and directly corresponds to the total energy in the
known sequence, which is proportional to its duration. Therefore,
the duration of the known sequence in the frame is maximized, while
weighing this against the need to send the time information in a
robust fashion, i.e. with redundancy. As is shown in the time
information frame 140 a total duration of 14 seconds is allocated
to the known sequence, starting from the last second of the
previous frame and ending 13 seconds into the current frame. Hence,
the amount of energy invested in the timing information is on the
order of a quarter of the total energy in a 60-second frame.
[0087] A diagram illustrating a fourth example time information
frame structure incorporating timing, time and additional
information is shown in FIG. 10. The bit allocation for the time
information frame, generally referenced 170, comprises 60-bits,
where each bit is allocated one second of the 60-second frame. The
bit designations for both historical amplitude/pulse-width
modulation 172 and phase modulation 174 are shown. In this
embodiment, the time information is conveyed as a 26-bit minute
field that is encoded into a 31 bit field using a linear block
code.
[0088] Note that the term phase modulation may denote either (1)
phase modulation over (i.e. combined with) the historical
amplitude/pulse-width modulation or (2) pure phase modulation
alone. Note also that the data fields assigned to each bit in the
frame for both amplitude and phase modulation are essentially
completely independent. In one embodiment, aspects of the
historical amplitude/pulse-width modulation, however, are taken
into consideration. For example, the time(0) bit (LSB of the minute
indication) is repeated over several (e.g., four) marker bits.
Non-repetitive information in the marker bits is avoided due to the
short high power duration (0.2 s) of these symbols.
[0089] Since a device that is in tracking mode typically uses the
received signal only to compensate for drift, the LSB of the minute
counter may be useful in resolving a timing drift whose magnitude
approaches 30 seconds, for which there may be uncertainty in the
correct minute. For example, if a timing drift of +27 seconds is
established based on the correlation operation against the known
sequence word, an actual drift of -33 seconds may be suspected.
These two possibilities will differ in terms of the correct minute
that is to be assumed, and the ambiguity may be resolved through
the reading of the minute's LSB, which may be found in five
different locations in the example frame 170, allowing for reliable
determination of the value of this bit even in harsh condition of
noise and interference.
[0090] For example, a receiver may either add the energies of all
five minute LSB fields, or may select one or multiple instances of
this bit in the frame for which a sufficiently high SNIR has been
established. It is noted that successful tracking relies on the
result of the correlation operation that the receiver performs
against the known synchronization sequence, but this operation
benefits from the extended duration of this sequence and from a
priori knowledge of its contents. Contrarily, the LSB of the minute
counter, denoted TIME[0] in example frame 170, must be resolved
through demodulation of the phase during one or more of its
instances in the frame. Hence, it is beneficial to allocate
multiple instances to this bit of information, providing both an
increase in total energy, as well as time-diversity, which is
effective against non-stationary noise/interference, as realized in
this embodiment of the present invention.
Merged Current Time Word
[0091] Rather than allocate a separate field to each element of
time (e.g., hour, minute, etc.) and encode this information in BCD
format as is done in the historical WWVB protocol where a total of
31 bits are consumed, the current time is represented more
efficiently and robustly in a single merged word as defined
below.
[0092] With reference to FIG. 9, the entire current time
information, which includes the minute, hour and date, is allocated
a total of 26 bits, to which 5 redundant parity bits are added,
which is, by coincidence, the same of number of bits dedicated to
the representation of time in the historical WWVB protocol. In this
example embodiment of the present invention, within the 26-bit time
word, 6 bits are used for a minute counter (0-59) and the remaining
20 bits represent the number of hours that have elapsed since the
beginning of the current century.
[0093] The number of hours in a century is limited to
100.times.365.25.times.24=876,600. Thus, the 20-bit field is used
efficiently since log.sub.2(876600)=19.74. The 6-bit portion
representing the minute is also used efficiently since
log.sub.2(60)=5.9, which is very close to 6 (60 of the possible 64
combination are used).
[0094] Alternatively, in another embodiment, with reference to FIG.
10, the minute within the century is specified in the 26-bit time
word. The number of minutes in a century is limited to
100.times.365.25.times.24.times.60=52,596,000. Thus, the 26-bit
field is used efficiently since log.sub.2(52,596,000)=25.64.
[0095] In both embodiments, the 26-bit time word is encoded into a
31-bit code-word using an error correcting code such as a
well-known Hamming systematic block code. This linear block code
has the capability to correct a single error that may occur in any
of the 31 bits and the capability to detect up to two errors. If
three errors were to occur, the received 31-bit word may appear
like a legitimate one, resulting in erroneous decoding. Such a
scenario is considered intolerable and the system is thus designed
to minimize the probability of such an event occurring. If, for
example, the BER is on the order of 10.sup.-2 (1%), the probability
for three errors is on the order of 10.sup.-6, which is
considerably lower.
[0096] In one embodiment, the block code used is a systematic code,
which means that the input data is embedded in the encoded output
and may be read directly from the output without requiring any
decoding. This property does not come at the cost of performance,
while it allows for simplified testing and reception (i.e.
elimination of the decoding procedure, particularly for high
SNR).
[0097] A block code may be denoted as a (n, k) code, where n and k
denote the code-word size and the number of information bits
respectively. For the time word in these two embodiments n=31 and
k=26. The equations below specify how each of the five parity bits,
denoted "timepar[i]" (i=0, 1, 2, 3, 4), is to be calculated using
the 26 bits comprising the time word denoted "time[k]" (k=0, 1 . .
. , 25).
timepar[0]=sum.sub.(modulo
2){time[23,21,20,17,16,15,14,13,9,8,6,5,4,2,0]}
timepar[1]=sum.sub.(modulo
2){time[24,22,21,18,17,16,15,14,10,9,7,6,5,3,1]}
timepar[2]=sum.sub.(modulo
2){time[25,23,22,19,18,17,16,15,11,10,8,7,6,4,2]}
timepar[3]=sum.sub.(modulo
2){time[24,21,19,18,15,14,13,12,11,7,6,4,3,2,0]}
timepar[4]=sum.sub.(modulo
2){time[25,22,20,19,16,15,14,13,12,8,7,5,4,3,1]}
[0098] Syndrome based decoding, wherein a syndrome vector is
calculated based on the received word in a linear fashion, can be
used in the receiver. A non-zero syndrome indicates that at least
one error occurred in the received word. For this Hamming code, the
syndrome can correctly indicate the error location if only one
error occurs in the received word. In order to guarantee the
reliability of the recovered information, the receiver does not
necessarily have to correct the received word when the syndrome is
non-zero. Instead, a second reception, confirming the contents of
the first, may be used, resulting in increased reliability at the
cost of delayed acquisition. Thus, the error detection capability
may, in some instances, be considered more important than the
correction capability.
[0099] Since it is probable to have more than one error in the
received word in low SNR scenarios, an erasure may occur and the
receiver can make a second acquisition attempt. In contrast, in
high SNR scenarios, where the likelihood of having two or more
errors is very low, a correction may be made, which is more
energy-efficient than repeated reception.
[0100] An analysis of linear versus nonlinear coding schemes for
the phase modulation based protocol in the presence of AWGN is
provided below.
[0101] The 26-bit time word represents the timing information
including the minute, hour and date. It is encoded into a 31-bit
codeword using a Hamming systematic block code. A Hamming code is a
linear block code that has single error correcting and double error
detecting capabilities. Hamming codes can be systematic or
non-systematic. A systematic code indicates that the input data is
embedded in the encoder output. For any positive integers
n-k=m.gtoreq.3, there exists a Hamming code with the following
parameters:
[0102] Code length: n=2.sup.m-1
[0103] Number of information bits: k=2.sup.m-m-1
[0104] Number of parity-check bits: n-k=m
[0105] Minimum Hamming distance: d.sub.min=3
[0106] Where n denotes the block size and k denotes the number of
information bits and the resulting block code is usually denoted as
a (n, k) code. In the system described herein, n=31 and k=26.
[0107] The encoder of a Hamming code uses vector matrix
multiplication. If u=(u.sub.0, u.sub.1, . . . , u.sub.k-1) is the
message to be encoded, the corresponding codeword v=(v.sub.0,
u.sub.1, . . . , v.sub.n-1) is given as follows
v=uG (2)
[0108] Where matrix G is called a generator matrix with dimension
k.times.n. There is a one-to-one mapping relationship between the
generator matrix G and the parity check matrix H used by the
decoder.
[0109] Similar to the encoder, the decoder of a Hamming code also
uses vector matrix multiplication, except that the matrix used is
called parity check matrix and is usually denoted H. The parity
check matrix H for the Hamming code is a n.times.(n-k) matrix
consisting of all the non-zero m-tuples as its columns. In
systematic form, the columns of H are arranged in the following
form:
H=[I.sub.mQ] (3)
[0110] where I.sub.m is a m.times.m identity matrix, and the
submatrix Q consists of k=2.sup.m-m-1 columns that are the m-tuples
of Hamming weight two or more. The Hamming weight of a binary
codeword refers to the number of `1`s in it. The parity check
matrix is shown in Table 1 below.
TABLE-US-00001 TABLE 1 Parity Check Matrix H for the (31,26)
Systematic Hamming Code ##STR00001##
[0111] For a systematic Hamming code, the generator matrix G can be
expressed as
G=[Q.sup.TI.sub.k] (4)
[0112] The generator matrix G is shown in Table 2 below.
TABLE-US-00002 TABLE 2 Generator Matrix G for the (31,26)
Systematic Hamming Code ##STR00002##
[0113] Syndrome based decoding is a general decoding scheme for
linear block codes where the syndrome S can be computed by
S=rH.sup.T (5)
[0114] Where r is the received word. A non-zero syndrome indicates
that at least one error occurred in the received word. For the
Hamming code, the error location is the index of the column in H
which is identical to the syndrome S.
[0115] In order to guarantee the correctness of the retrieved
information, however, the receiver does not necessarily attempt to
correct the received word when the syndrome is non-zero. The reason
is that the correction based on syndrome decoding is reliable only
when there is one error in the received word, otherwise the
receiver will mistakenly correct the received word to the wrong
codeword. Since it is probable to have more than one error in the
received word in low SNR scenarios, an erasure may occur and the
receiver can make a second acquisition attempt. In contrast, a
correction can be made in high SNR scenarios, where the likelihood
of having two or more errors is very low. The above correction
mechanism involves an SNR threshold and a trade-off between the
block error rate and the cost of making a second acquisition
attempt. In the analysis provided herein, it is assumed that the
receiver only makes one acquisition attempt and tries to correct
the error whenever the syndrome is non-zero.
[0116] Given a signal-to-noise-ratio (SNR), the probability of a
bit error for BPSK is given by
P.sub.bit.sub.--.sub.error=Q( {square root over (2.times.SNR)})
(6)
[0117] Where the noise power N is measured in bandwidth B=R.sub.b,
which is equivalent to
S N R = E b N 0 ##EQU00002##
with E.sub.b being the energy per bit,
N 0 2 ##EQU00003##
being the power spectral density of the AWGN, and Q(x) being the
tail probability of the normal distribution, i.e.
Q ( x ) = 1 2 .pi. .intg. x .infin. exp ( - u 2 2 ) u ( 7 )
##EQU00004##
[0118] The word error rate (WER) of the uncoded minute counter can
be calculated using
P.sub.word.sub.--.sub.error=1-(1-P.sub.bit.sub.--.sub.error).sup.k
(8)
[0119] When coded using a linear block code, a decoding error
occurs when there are two or more errors in the received word.
Therefore, the WER for the coded time word using the Hamming code
can be calculated as follows
P.sub.decoded.sub.word=1-(1-P.sub.bit.sub.error).sup.n-nP.sub.bit.sub.er-
ror(1-P.sub.bit.sub.error).sup.n-1 (9)
[0120] Whenever at least one bit in a word is recovered in error
(despite the coding) this represents a time word error event, for
which the word error probability (WER) is defined. The WER
comparison for the coded and uncoded time word (merged time data
field) is shown in FIG. 11. It is noted that a WER of 10.sup.-3
corresponds to one error in 100 years even if time acquisition is
performed 10 times in a year. The demodulation SNR for this WER is
shown to be about 8.9 dB for the uncoded word and 6.4 dB for the
coded word representing a coding gain of 2.5 dB.
Additional Information Field
[0121] The additional information field comprises seven bit (before
any error correcting codes are applied). The 7-bit additional
information field comprises one bit indicating whether DST is in
effect or not, one bit indicating whether a leap-second is to be
added at the end of the current half-year and a 5-bit DST schedule
word, which serves to indicate the time and day for the next DST
transition. With error correcting codes applied, the additional
information field consists of 11 bits {dston, leap, dlpar[0],
dlpar[1], dlpar[2]} and {dst[0], dst[1], . . . , dst[5]}. The
additional information field is primarily used to communicate the
DST state and the schedule for the upcoming DST transition (i.e.
either entering or exiting DST). This field is based on two
information words of 2-bits and 5-bits that are encoded into 5 bits
and 6 bits, respectively, for error detection and correction.
[0122] The additional information that may be communicated through
this field includes a leap second notification, emergency messages,
and whatever other information that is desired to be conveyed.
Table 3, Table 4, Table 5 and Table 6 specify the bit designations
and how the information in this field is used and interpreted.
TABLE-US-00003 TABLE 3 DST State (Bit 6) dston Description 0 DST
not in effect (OFF); next transition is into DST (use Column C in
Table 6) 1 DST in effect (ON); next transition is out of DST (use
Column B in Table 6)
TABLE-US-00004 TABLE 4 Leap Second Notification (Bit 5) leap
Description 0 no leap second scheduled at the end of this half-year
1 leap second scheduled for this half-year (the earlier of June
30.sup.th or Dec 31.sup.st)
TABLE-US-00005 TABLE 5 Bits 4:3 in 5-bit DST Schedule Word Bit 4
Bit 3 Description 0 0 use Column A to decode bits 2:0 (special
messages) 0 1 next DST transition hour is 1AM, day is in bits 2:0 1
0 next DST transition hour is 2AM, day is in bits 2:0 1 1 next DST
transition hour is 3AM, day is in bits 2:0
TABLE-US-00006 TABLE 6 Bits 2:0 in 5-bit DST Schedule Word Column B
Column C Bit 2 Bit 1 Bit 0 Column A (end of DST) (start of DST) 0 0
0 DST permanently 3.sup.rd Sunday 6.sup.th Sunday off before "O"
since "M" 0 0 1 DST permanently 2.sup.nd Sunday 7.sup.th Sunday on
before "O" since "M" 0 1 0 DST out of 1.sup.st Sunday 8.sup.th
Sunday range before "O" since "M" 0 1 1 Reserved last Sunday
1.sup.st Sunday of Oct. = "O" of March = "M" 1 0 0 Reserved
1.sup.st Sunday 2.sup.nd Sunday of Nov. since "M" 1 0 1 Reserved
2.sup.nd Sunday 3.sup.rd Sunday of Nov. since "M" 1 1 0 Reserved
3.sup.rd Sunday 4.sup.th Sunday of Nov. since "M" 1 1 1 Reserved
4.sup.th Sunday 5.sup.th Sunday of Nov. since "M"
[0123] Note that the first Sunday in April could be either the
5.sup.th or 6.sup.th Sunday since the beginning of March. Note also
that the last Sunday in October or in March may be either the
4.sup.th or the 5.sup.th Sunday of that month.
[0124] Bit "dston" (dst-on) indicates whether DST is in effect. In
the United States, up until 2007, under legislation enacted in
1986, Daylight Savings Time began at 2:00 am on the first Sunday of
April and ended at 2:00 am on the last Sunday of October. In 2007
this has been changed, such that DST starts on the second Sunday of
March until the first Sunday of November. Note that the dston field
is potentially represented by more than one bit in the frame for
the purpose of redundancy (i.e. robustness and error correction
capabilities). Bit "leap" indicates whether a leap second is
scheduled at the end of a predefined period (e.g., end of this
month, half-year, etc.). A leap second may be added to the last
minute of June 30.sup.th, or December 31.sup.st.
[0125] If DST is in effect (e.g., in July), then the interpretation
of the 5-bit DST schedule word refers to when it is to end. If DST
is not in effect (e.g., in December), the interpretation of this
word refers to when it is to start again. The start date and end
date options are listed in Table 6 above. A total of eight specific
options are supported for each, and an "out of range" possibility
is defined, in case the DST schedule is changed in the future to a
time that is not within those covered by the table. Additional
options are defined to allow for DST to be implemented permanently
or to be cancelled altogether. With three possible values for the
time at which the DST transition is to occur (1 AM, 2 AM or 3 AM)
and a 4.sup.th option to be used for special messages, as shown in
Table 5, a total of 32 combinations may exist for the DST 5-bit
schedule word.
[0126] The Sundays in Column C of Table 6, indicating the start
date of DST, are not in chronological order, since it is
advantageous to designate the same word to the 1.sup.st Sunday of
November and to the 2.sup.nd Sunday of March, being the most
recently enacted end and start dates respectively. This allows for
more efficient representation of the information under the
assumption that this DST schedule, which is currently in use, will
likely remain the schedule for many years to come. The other
optional schedules are defined to allow some margin around what
appears to be a possible schedules that could be instated in the
future. It is to be noted that the last Sunday in October or in
March may be either the 4.sup.th or the 5.sup.th Sunday of that
month.
[0127] The 2-bit word comprising the DST status bit ("dston") and
the leap-second notification bit ("leap") may be used immediately
upon reception and is of high importance. These bits are relatively
unpredictable and thus have high information content. Hence, in one
embodiment, three parity bits are allocated to increase the
robustness of the 2-bit word. The 2-bit word is encoded into 5 bits
using a shortened Hamming systematic linear block code that
provides relatively high robustness, as detailed infra.
[0128] In contrast, due to the highly disparate a priori
probability for the DST transition schedule options, a nonlinear
code is used to encode the 5-bit DST schedule word into a 6-bit
codeword, providing non-uniform distancing for the various
codewords, with the most probable one having the highest
protection, i.e. the greatest Hamming distance from all other
codewords.
[0129] The 2-bit word comprising the DST status bit ("dston") and
the leap-second notification bit ("leap") and the 5-bit information
word of scheduled DST time and date are encoded separately. This is
because the 5-bit DST schedule word has specific content with high
probability, which would represent less information, whereas the
contents of the 2-bit word are more unpredictable. Due to the high
uncertainty and importance of the 2-bit word, three parity bits are
allocated to increase its robustness.
[0130] The Hamming block code used for encoding the 2-bit word in
the additional information field is a systematic code, which means
that the 2-bit information word input into the code also appears at
its output and may be read there directly without decoding. The (5,
2) shortened Hamming systematic code that is used to encode these
two information bits into a 5-bit code-word is derived from a (7,
4) Hamming systematic code.
[0131] The three additional parity bits are calculated according to
the three "dlpar" equations below, which are derived from the
generating matrix in Table 7 below. These parity bits are appended
to the 2-bit word from the MSB side (i.e. are transmitted before
the two bits). This code provides the receiver with the capability
to correct one error and to detect up to two errors in the 5-bit
word.
TABLE-US-00007 TABLE 7 DST Status/Leap Second 2-bit to 5-bit
Encoding ##STR00003##
[0132] The equations below specify how each of the three parity
bits, denoted "dlpar[i]" (i=0, 1, 2), is to be calculated using the
two input bits denoted dston and leap. They are equivalent to
multiplying a two element row vector, representing the 2-bit
information word, by the 2.times.5 coding matrix, resulting in a
five element row vector representing the codeword to be
transmitted.
[0133] dlpar[0]=dston
[0134] dlpar[1]=sum(modulo 2){leap, dston}
[0135] dlpar[2]=leap
[0136] A lookup table showing the 5-bit to 6-bit encoding of the
DST schedule is shown in Table 9 below. Due to the unequal a priori
probability of each value for the 5-bit field for the DST schedule,
the 5-bit to 6-bit encoder is nonlinear resulting in non-uniform
distancing of the codewords, thus the encoder is based on a lookup
table as specified in Table 9.
[0137] The mapping of the 2.sup.5=32 codewords to the information
shown in Tables 5 and 6 is shown in the rightmost column. The
codewords and their corresponding d.sub.min are shown. The first
codeword, having a maximum d.sub.min of 3, is mapped to the most
probable DST schedule, which is the one instated most recently
(i.e. the DST period starting on the 2.sup.nd Sunday of March and
ending on the 1.sup.st Sunday of November). Since the transition
will most likely remain at 2 AM, and will be implemented on the
same Sundays that it has been recently moved to, this combination
was selected as the most probable one, for which the maximal coding
protection was assigned in the non-linear code.
[0138] Thus, the most likely message, indicating the regular (i.e.
current) schedule for the DST transition is designated codeword #0
(the first row in Table 9), for which the minimum Hamming distance
(d.sub.min) is the highest (d.sub.min=3). Other messages, having
secondary likelihood, which have been used historically, are
designated codewords having minimum Hamming distance d.sub.min=2,
while those of least probability or importance are designated the
codewords having minimum Hamming distance d.sub.min=1.
[0139] The minimum Hamming distance of each codeword reflects the
error protection capability for that codeword. A codeword with
greater hamming distance has higher error protection capability and
is therefore assigned to an information word with higher a priori
probability or greater importance. In the example embodiment shown
in Table 9, codeword #0 (the first row), being the most probable
word and having the strongest protection, has been assigned the
value 111001 as these are the first six bits of the 7-bit barker
code 1110010. The marker at the end of the 60-minute frame, being
permanently assigned 0 in this embodiment, can serve as the last
(7.sup.th) bit. The addition of this 7-bit barker code can further
enhance the synchronization capabilities, as it is likely that this
word will be present most of the time.
TABLE-US-00008 TABLE 9 Lookup Table for DST Schedule Word 5-bit to
6-bit Encoder ##STR00004##
[0140] In Table 9, "O" represents the last Sunday of October and
"M" represents the first Sunday of March. Since the DST schedule
word is followed by the synchronization sequence, it is desirable
to configure the most probable DST schedule word, in conjunction
with the synchronization sequence, to have good autocorrelation
properties. Therefore, in one embodiment, an offset word C is added
to all codewords in Table 9 in order to improve the synchronization
performance while maintaining the minimum Hamming distance of the
codewords.
[0141] As described supra, the additional information includes a
2-bit word for DST status and leap second and a 5-bit word
representing the schedule for the next DST transition. The 2-bit
word is encoded into 5-bits using a shortened Hamming systematic
code.
[0142] The (5, 2) shortened Hamming systematic code used is derived
from a (7, 4) Hamming systematic code by deleting two columns in
the submatrix Q of the parity check matrix H. The encoder and
decoder structure is the same as the ones used for the time word.
The parity check matrix H and the generator matrix G are shown in
Tables 10 and 11, respectively, below.
TABLE-US-00009 TABLE 10 Parity Check Matrix H for (5, 2) Shortened
Systematic Hamming Code ##STR00005##
TABLE-US-00010 TABLE 11 Generator Matrix G for (5, 2) Shortened
Systematic Hamming Code ##STR00006##
[0143] The assumption and evaluation of the WER performance is the
same as with the (31, 26) Hamming code for the time word. A graph
illustrating the performance with and without coding is shown in
FIG. 12. At a WER of 10.sup.-3, corresponding to one error in 100
years if time acquisition is made 10 times in a year, the
demodulation SNR is approximately 7.3 dB for the uncoded word and
4.3 dB for the coded word, thereby representing a coding gain of 3
dB. Note that the WER performance for this 2-bit field is better
than that of the 26-bit field time word with or without coding,
simply because the field is shorter.
[0144] The nonlinear code used for the 5-bit DST schedule field is
designed such that at least one particular codeword will have
maximum protection, corresponding to a maximal minimum Hamming
distance, denoted as d.sub.min. The remaining codewords are
selected to have a maximal number of codewords of maximum
d.sub.min.
[0145] With the number of information bits denoted as k and the
coded block size as n, the design of the codeword follows these
steps: First, one n-length word, denoted as c.sub.M, is chosen that
has maximum protection against error. Once the code is constructed,
an offset may be added to all codewords to obtain good
autocorrelation properties for c.sub.M. Second, all the n-length
words are eliminated that have Hamming distance of i with the
codeword c.sub.M, where i=1, 2, . . . , m, and m is the largest
integer that could satisfy the following conditions
2.sup.n-2.sup.k-.SIGMA..sub.i=1.sup.m(.sub.m.sup.n).gtoreq.0
(10)
[0146] The equality in Equation 10 above ensures that the number of
words that are eliminated is less than or equal to the total number
of invalid codewords. For this (6, 5) non-linear code, we obtain
m=2 implying d.sub.min=3. The remaining set of words, denoted as R,
would have r=42 elements.
[0147] Third, the set R is divided into two sets: set S.sub.1 and
set S.sub.2, which together contain n-length words with d.sub.min=1
and 2 within the set respectively. Note that each word in R must be
in set S.sub.1 or in set S.sub.2 but cannot be in both sets, i.e.
R=S.sub.1.orgate.S.sub.2 and S.sub.1.andgate.S.sub.2=O. This
problem is formulated as a linear program to ensure the maximum
number of words in set S.sub.2. The cardinality of set S.sub.1 and
set S.sub.2 are represented by s.sub.1 and s.sub.2. The framework
of the linear program is further described infra.
[0148] If the number of words in set S.sub.2 is no less than the
codeword needed, i.e. s.sub.2.gtoreq.2.sup.k-1, the design is
performed by arbitrarily picking 2.sup.k-1 words in set S.sub.2.
Otherwise, which is the case in the (6, 5) nonlinear code of the
example embodiment of Table 9, a second linear program is needed to
choose 2.sup.k-1-s.sub.2 words from set S.sub.1, such that a
minimal number of words in set S.sub.2 has a reduced d.sub.min. The
framework of the linear program is further described infra.
[0149] The first linear program comprises selecting a minimum
number of words from set R to set S.sub.1, such that set S.sub.2
will have maximum cardinality. Let us denote the set of remaining
n-length words as R, with cardinality r. Parameter A is a r.times.r
indicator matrix, where A.sub.i,j=1 if-and-only-if (iff) the
Hamming distance between word i.epsilon.R and word j.epsilon.R is
one. Binary variable x.sub.i=1 iff word i.epsilon.R is chosen as a
member of set S.sub.1. For the convenience of notation, we also
define binary variable y complement to x, i.e. y.sub.i= x.sub.i
where i=1, 2, . . . , r. The r.times.r matrix B is a binary
variable, which represents the resulting indicator matrix after
eliminating words of d.sub.min=1 and putting them in set
S.sub.1.
[0150] The objective function is to minimize the number of words
eliminated and put them in set S.sub.1, i.e.
min.sub.x,y,B.SIGMA..sub.i=1.sup.rx.sub.i, subject to constraints
11, 12 and 13.
[0151] The first constraint is the complementary relationship
between variables x and y.
x.sub.i=1-y.sub.i,.A-inverted.i=1,2, . . . , r (11)
[0152] The second constraint is the relationship between indicator
matrix A and B. In other words, by eliminating codewords with
d.sub.min=1, variable y will zero out the ones in matrix A
resulting in matrix B.
B.sub.i,j.gtoreq.A.sub.i,jy.sub.iy.sub.j,.A-inverted.i=1,2, . . .
,r and .A-inverted.j=1,2, . . . ,r (12)
[0153] The third constraint is that set T should not have codewords
with d.sub.min=1.
.SIGMA..sub.i=1.sup.r.SIGMA..sub.j=1.sup.rB.sub.i,j=0 (13)
[0154] Considering the (6, 5) nonlinear code of this example
embodiment, the cardinalities of set S.sub.1 and S.sub.2 are 16 and
26, respectively. Since the cardinality of set S.sub.2 is less than
the requirement number of codewords 31, a second linear programming
procedure is used.
[0155] The second linear programming procedure is to choose
2.sup.k-1-S.sub.2 words from set S.sub.1 to set S.sub.2 such that
the resulting set has a maximum number of codewords with
d.sub.min=2. Set S.sub.1 and set S.sub.2 represent the sets of
n-length words that have a d.sub.min of one and two within the set
respectively, with cardinalities s.sub.1 and s.sub.2 respectively.
Binary parameter D is defined as a s.sub.1.times.s.sub.2 matrix,
such that D.sub.i,j=1 iff word i.epsilon.S.sub.2 and word
j.epsilon.S.sub.1 have Hamming distance of one. Parameter
n.sub.needone denotes the number of codewords needed to be taken
from set S.sub.1, in this case n.sub.needone=2.sup.k-1-s.sub.2=5.
Binary variable g.sub.i=1 iff word i.epsilon.S.sub.1 is selected.
Decision variable h.sub.i denotes the number of words picked from
set S.sub.1 to set S.sub.2 that have Hamming distance of one with
word i.epsilon.S.sub.2. Binary variable z is the indicator of
variable h, i.e. z.sub.i=1 iff h.sub.i>0.
[0156] The goal is to minimize the words in set S.sub.2 that will
have a decreased d.sub.min due to the codewords selected from set
S.sub.1, which can be written as
min.sub.g,h,z.SIGMA..sub.i=1.sup.s.sup.2 z.sub.i subject to
constraints 14, 15 and 16.
[0157] The first constraint is the relationship between variable h
and z, which is given as
z i .gtoreq. h i s 1 .A-inverted. i .di-elect cons. S 2 ( 14 )
##EQU00005##
[0158] Where the division serves to ensure the binary variable
z.sub.i is always greater than the right-hand-size, by letting the
right-hand-side be less than or equal to one.
[0159] The second constraint is the relationship between variable g
and h. For a given word in set S.sub.2, the number of codewords
selected from set S.sub.1 that have a Hamming distance of one
associated with them is given by summing the rows of the chosen
codeword in indicator matrix D.
h.sub.j=.SIGMA..sub.i=1.sup.S.sup.1g.sub.iD.sub.i,j.A-inverted.j.epsilon-
.S.sub.2 (15)
[0160] The third constraint ensures that the number of words picked
from set S.sub.1 is no less than needed.
.SIGMA..sub.i=1.sup.S.sup.1g.sub.i.gtoreq.n.sub.needone (16)
[0161] Considering the (6, 5) nonlinear code, five codewords are
selected from set S.sub.1 to set S.sub.2, and there are 16 words
from set S.sub.2 that have a reduced d.sub.min=1. In summary, there
is one codeword that has d.sub.min=3; 10 codewords that have
d.sub.min=2; and the remaining 20 codewords have d.sub.min=1.
[0162] Regarding receiver performance, for a given
signal-to-noise-ratio (SNR), the calculation of probability of bit
error for BPSK and the uncoded block error rate are given by
Equations 6 and 8 infra, respectively. To simplify the performance
analysis, the same assumption is made that the receiver only makes
one acquisition attempt and tries to correct any errors. Given a
received word, the decoder compares the Hamming distance between
the received word and all codewords and chooses the codeword that
has minimum Hamming distance with the received word as its
estimated output. If there are two or more codewords that yield the
same Hamming distance, the receiver chooses the codeword listed
earliest in the codebook.
[0163] The codewords in the codebook can be listed according to the
a priori probability of each codeword. A graph illustrating the
average word error rate (WER) of codewords that have different
d.sub.mins is shown in FIG. 13. The performance at the WER of
10.sup.-3 is also compared, which corresponds to one error in 100
years if 10 acquisitions are made every year. The demodulation SNR
is approximately 4.5 dB for the most protectively coded word and 8
dB for the rest, including uncoded words and coded word with
d.sub.min of 1 and 2. Hence, FIG. 13 shows a 3.5 dB coding gain on
the most protective codeword and almost zero coding gain for other
codewords.
[0164] The performance of using the most protective codeword is
superior due to the following reasons: (1) it has the maximum
d.sub.min; and (2) it is listed first in the codebook since it has
the largest a priori probability. In other words, whenever a word
is received having the same Hamming distance to the most
protected/probable codeword as to some other codeword, the receiver
always chooses the most protected/probable word. This is why
codewords with d.sub.min=2 only have marginal coding gain. The
performance of the codewords that have d.sub.min=2, however, could
potentially be much higher than codewords that have d.sub.min=1
since the decoder can still detect when a single error occurs,
although it might correct it to the wrong codeword. The detection
capability of the codeword with d.sub.min=2 can reduce the
occurrence of wrong correction by considering the SNR estimate
information.
[0165] To increase the reliability of the extracted information, a
device operating in accordance with the present invention may
repeat the reception of the DST schedule word, at the cost of
energy consumption, whenever an error in it is detected, rather
than attempting to correct it,
Frame Generator
[0166] A block diagram illustrating an example parallel to serial
conversion of AM and PM bits is shown in FIG. 14. The frame
generator, generally referenced 180, comprises a 60-bit AM register
182, 60-bit PM register 184, serializers 186, 190 and delay 188.
The frame generator and modulator in the time code generator 14
(FIG. 1) are operative to output the 60-bit frame for both the
amplitude and phase-modulated data. The bit allocation for the AM
and PM data are shown in FIGS. 9 and 10 as described supra.
[0167] The input signals to the phase modulation frame generator
include the following: The CLOCK IN signal comprises a 1 pulse per
second (pps) signal derived from an accurate frequency reference
that is input to the TCG, e.g., 5 MHz). The FRAME STROBE input is a
pulse provided every minute to trigger the generation of a new
60-bit frame. A PM ENABLE IN input is used to enable/disable the
phase modulation. It is gated with the FRAME STROBE IN signal in
order to ensure that is takes effect only on frame boundaries.
[0168] The time information, which includes the minute, hour and
date, is initially input by the user or through a user interface
such as a touch screen. In one embodiment (FIG. 10, for example),
this information is converted into a 26-bit minute counter
representing the number of minutes that have elapsed since the
beginning of the century. The time keeping operation requires
incrementing this counter and detecting once the century is over,
at which time it will transition to zero. As described supra, the
26-bit time word is encoded into a 31-bit code word which is
conveyed in the frame locations {time[0] . . . time[25], timepar[0]
. . . timepar[4]} as shown in FIG. 10.
[0169] Aside from in the time information, an additional
information field comprising 11-bits is provided, as described
supra. This field includes information such as the schedule for DST
transition and leap second. This field is based on two information
words of 2-bits and 5-bits that are encoded into 5-bit and 6-bits,
respectively, for error detection and correction. The contents of
this field are conveyed in frame locations {dston, leap, dlpar[0],
dlpar[1], dlpar[2]} and {dst[0] . . . dst[5]} as shown in FIGS. 9
and 10.
[0170] The output generated by the PM frame generator is a 60-bit
word that is serialized in the time code generator in parallel with
the serialization of the 60-bit AM word where each is input to the
appropriate modulator (i.e. amplitude or phase) as shown in FIG.
14.
[0171] The frame also comprises a 14-bit synchronization word that
is placed in locations {sync[0] . . . sync[13]} of the 60-bit time
frames. A different synchronization word {sync_M[0] . . .
sync_M[13]} is used for message frames.
[0172] In one embodiment, regarding the relative timing for the
phase modulation, the 60-bit word that determines the phase
modulation for the carrier is synchronized to the 60-second
frame-timing dictated by the existing AM/pulse-width modulation and
is delayed 100 ms (delay 188, FIG. 14) with respect to it, such
that the boundaries for each PM bit of 1-second duration do not
coincide with those of the AM bits. This delay is introduced in
order to avoid 180.degree. phase transitions when the amplitude is
transitioning, which effectively creates a greater transient for
the transmitter.
[0173] The terminology used herein is for the purpose of describing
particular embodiments only and is not intended to be limiting of
the invention. As used herein, the singular forms "a", "an" and
"the" are intended to include the plural forms as well, unless the
context clearly indicates otherwise. It will be further understood
that the terms "comprises" and/or "comprising," when used in this
specification, specify the presence of stated features, integers,
steps, operations, elements, and/or components, but do not preclude
the presence or addition of one or more other features, integers,
steps, operations, elements, components, and/or groups thereof.
[0174] The corresponding structures, materials, acts, and
equivalents of all means or step plus function elements in the
claims below are intended to include any structure, material, or
act for performing the function in combination with other claimed
elements as specifically claimed. The description of the present
invention has been presented for purposes of illustration and
description, but is not intended to be exhaustive or limited to the
invention in the form disclosed. As numerous modifications and
changes will readily occur to those skilled in the art, it is
intended that the invention not be limited to the limited number of
embodiments described herein. Accordingly, it will be appreciated
that all suitable variations, modifications and equivalents may be
resorted to, falling within the spirit and scope of the present
invention. The embodiments were chosen and described in order to
best explain the principles of the invention and the practical
application, and to enable others of ordinary skill in the art to
understand the invention for various embodiments with various
modifications as are suited to the particular use contemplated.
[0175] It is intended that the appended claims cover all such
features and advantages of the invention that fall within the
spirit and scope of the present invention. As numerous
modifications and changes will readily occur to those skilled in
the art, it is intended that the invention not be limited to the
limited number of embodiments described herein. Accordingly, it
will be appreciated that all suitable variations, modifications and
equivalents may be resorted to, falling within the spirit and scope
of the present invention.
* * * * *