U.S. patent application number 11/778550 was filed with the patent office on 2008-01-17 for combined parallel and sequential detection for gps signal acquisition.
Invention is credited to Charles P. Norman, Stephen F. Rounds.
Application Number | 20080015776 11/778550 |
Document ID | / |
Family ID | 22799163 |
Filed Date | 2008-01-17 |
United States Patent
Application |
20080015776 |
Kind Code |
A1 |
Norman; Charles P. ; et
al. |
January 17, 2008 |
Combined Parallel and Sequential Detection for GPS Signal
Acquisition
Abstract
A two part signal acquisition process includes a parallel signal
detection process and signal verification/false alarm rejection
process. A massively parallel architecture of acquisition
correlators search a large region of the time-frequency uncertainty
during the parallel signal detection process to identify the most
likely detections for each search dwell. Concurrent with the
parallel signal detection process performed by the acquisition
correlators, the current list of most likely detections is examined
with additional search dwells in the verification/false alarm
rejection process. The verification/false alarm rejection process
is performed by a plurality of independent correlators or tracking
channels. Under software control, the tracking channels perform
repeated dwells on the most likely detections until they can be
dismissed as false alarms or verified as the desired signal.
Inventors: |
Norman; Charles P.;
(Huntington Beach, CA) ; Rounds; Stephen F.;
(Irvine, CA) |
Correspondence
Address: |
David S. Sarisky, Esq.;FULWIDER PATTON LEE & UTECHT, LLP
Tenth Floor
6060 Center Drive
Los Angeles
CA
90045
US
|
Family ID: |
22799163 |
Appl. No.: |
11/778550 |
Filed: |
July 16, 2007 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11549511 |
Oct 13, 2006 |
7246011 |
|
|
11778550 |
Jul 16, 2007 |
|
|
|
09886427 |
Jun 20, 2001 |
7127351 |
|
|
11549511 |
Oct 13, 2006 |
|
|
|
60214462 |
Jun 27, 2000 |
|
|
|
Current U.S.
Class: |
701/469 ;
342/357.68; 342/357.69; 375/E1.011 |
Current CPC
Class: |
G01S 19/37 20130101;
G01S 19/30 20130101; H04B 1/709 20130101; G01S 19/29 20130101; H04B
1/70775 20130101 |
Class at
Publication: |
701/213 |
International
Class: |
G01C 21/26 20060101
G01C021/26 |
Claims
1. A signal acquisition process comprising: a) performing an
acquisition dwell on a plurality of cells within a time/frequency
uncertainty range to detect a set of cells having the largest
correlation peaks; b) performing an initial verification dwell on
the set of cells detected in step a by comparing the peak of each
cell to a threshold and retaining those cells having a peak at
least as great as the threshold; c) performing an acquisition dwell
on another plurality of cells within the time/frequency uncertainty
range to detect another set of cells having the largest correlation
peaks; and d) performing a subsequent verification dwell on the
cells retained in step b and an initial verification dwell on the
set of cells detected in step c by comparing the peak of each cell
to the threshold and retaining those cells having a peak at least
as great as the threshold.
Description
RELATED APPLICATIONS
[0001] This application claims the benefit of co-pending
provisional application Ser. No. 60/214,462, filed Jun. 27,
2000.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The invention relates generally to an apparatus and method
for acquiring satellite signals and more particularly, to an
apparatus and method for acquiring GPS signals that combines
parallel and sequential detection processes.
[0004] 2. Description of the Related Art
[0005] The requirements for fast acquisition in weak signal and
high jamming environments has led the GPS industry to massively
parallel architectures for both commercial and military
applications. These massively parallel architectures comprise
hundreds of thousands of correlators and dozens of FFT taps. This
architecture allows searches of large regions of the time-frequency
uncertainty domain in a single dwell of the acquisition engine.
However, since all elements of the parallel architecture are
treated equally, it has not been possible to realize the advantages
of a sequential detector in minimizing the dwell time as a function
of the dynamically observed signal and noise environment.
[0006] Typical GPS signal acquisition processes comprise two steps.
In the first step of the process, the largest signal detected in an
array of thousands of time/frequency cells is compared against a
threshold. In the second step of the process, if the largest
detected signal exceeds the threshold, a verification/false alarm
rejection algorithm attempts to detect the signal again, multiple
times to verify that the signal does indeed exceed the software
threshold.
[0007] In the first step, the architecture of the GPS receiver,
specifically the acquisition correlators, form a correlation output
over the desired coherent and non-coherent integration period and
report the time and frequency coordinates of the peak detected
signal that exceeds a software-controlled threshold. Once these
coordinates are reported, the architecture uses the same
acquisition correlators previously used to detect the peak signal
to attempt to detect the signal in multiple, repeated dwells to
confirm the presence of the signal.
[0008] One drawback with this technique is the large sensitivity to
the selected threshold. If the threshold is too high there is a
large probability of a missed detection, if it is too low the
search rate will be slowed due to false alarms. Another significant
drawback to this GPS acquisition process is that it requires the
reported signal to be the largest detected out of thousands of
signals. Assuming that the desired probability of acquisition is
98%, this means that the probability that a noise sample exceeds
the signal plus noise must not exceed a 2% missed detection
probability. For a 511 correlator, 64-tap FFT architecture having
32,704 cells, this requires a signal-to-noise (S/N) ratio of 6.95.
Achieving such a S/N ratio implies either a lower jam immunity/weak
signal sensitivity, or an increase in the time-to-first-fix (TTFF)
while the signal is integrated to the required S/N.
[0009] Hence, those skilled in the art have recognized the need for
a GPS signal acquisition process having signal detection capability
that is not so heavily dependent on the setting of a threshold
level, as are current processes. The need has also been recognized
for a process that retains the wideband search capabilities of
existing massively parallel architectures, but also includes many
of the advantages of a sequential detector.
SUMMARY OF THE INVENTION
[0010] Briefly, and in general terms, the present invention
provides an apparatus and a process for acquiring GPS signals that
employs an algorithm which may be characterized as a combined
parallel/sequential detector. The first step of the process screens
out the most likely detections using the massively parallel
architecture of a GPS receiver. The screening step is followed by
what is essentially a sequential detector--a verification/false
alarm rejection algorithm that rejects a potential detection after
any one of a number of steps unless it passes a threshold at each
step.
[0011] Traditional acquisition algorithms have been shown to be
sub-optimal for use with the massively parallel architectures that
are required for acquisition of the GPS signal in a high jamming or
weak signal environment. The present invention provides an improved
acquisition algorithm which combines the best features of the
parallel acquisition with the strengths of a sequential detector
algorithm. A detailed implementation of this algorithm is described
for two exemplary GPS receiver architectures. One architecture
includes 511 acquisition correlators and a 64-tap FFT with six
independent correlators for verification. A second architecture
includes 240 acquisition correlators which can be configured as
twelve independent correlators for verification. A mathematical
analysis quantifies the advantages of the algorithm over more
conventional designs. These analyses show that the required
signal-to-noise for successful acquisition can be substantially
reduced.
[0012] In a first aspect, the invention relates to a signal
acquisition process that include a) performing an acquisition dwell
on a plurality of cells within a time/frequency uncertainty range
to detect a set of cells having the largest correlation peaks and
b) performing an initial verification dwell on the set of cells
detected in step a by comparing the peak of each cell to a
threshold and retaining those cells having a peak at least as great
as the threshold. The process further includes c) performing an
acquisition dwell on another plurality of cells within the
time/frequency uncertainty range to detect another set ofcells
having the largest correlation peaks and d) performing a subsequent
verification dwell on the cells retained in step b and an initial
verification dwell on the set of cells detected in step c by
comparing the peak of each cell to the threshold and retaining
those cells having a peak at least as great as the threshold.
[0013] In another aspect, the invention relates to a signal
acquisition device that includes a plurality of acquisition
correlators and a plurality of independent correlators. The
acquisition correlators are adapted to perform an initial
acquisition dwell and a series of subsequent acquisition dwells on
a plurality of cells within a time/frequency uncertainty range. A
set of cells having the largest correlation peaks is detected
during each acquisition dwell. Each of the independent correlators
is adapted to receive a detected cell from the acquisition
correlators, perform an initial verification dwell on the detected
cell by comparing the peak of the detected cell to a threshold and
retaining the detected cell only if it has a peak at least as great
as the threshold. The independent correlators are further adapted
to perform at least one subsequent verification dwell on the
retained cell.
[0014] In another aspect, the invention relates to a system for
tracking the location of an object using signals transmitted by GPS
satellites. The system includes an antenna associated with the
object for receiving GPS signals and a signal acquisition device
similar to the one described in the preceding paragraph. The signal
acquisition device is in operative communication with the antenna
and acquires and processes the GPS signals received by the
antenna.
[0015] In another aspect, the invention relates to a signal
acquisition process that includes performing a plurality of
acquisition dwells on a plurality of cells within a time/frequency
uncertainty range. Each acquisition dwell detects "x" number of
cells having the largest correlation peaks. The process also
includes comparing the cells detected during the acquisition
dwells, retaining "x" number of cells having the largest
correlation peaks and performing a verification dwell on the "x"
number of cells by comparing the peak of each cell to a threshold
and retaining only those cells having a peak that exceeds the
threshold.
[0016] In another aspect, the invention relates to a signal
acquisition device including a plurality of acquisition
correlators, a processor implemented software list and a plurality
of independent correlators. The plurality of acquisition
correlators are adapted to perform acquisition dwells on a
plurality of cells within a time/frequency uncertainty range to
detect "x" number of cells having the largest correlation peaks.
The software list is adapted to compare the cells detected during
the acquisition dwells and retain "x" number of cells having the
largest correlation peaks. Each of the plurality of independent
correlators is adapted to receive a detected cell from the software
list, perform an initial verification dwell on the detected cell by
comparing the peak of the detected cell to a threshold and
retaining the detected cell only if it has a peak at least as great
as the threshold. The independent correlators are further adapted
to perform a subsequent verification dwell on the retained
cell.
[0017] In another aspect, the invention relates to a system for
tracking the location of an object using signals transmitted by GPS
satellites. The system includes an antenna associated with the
object for receiving GPS signals and a signal acquisition device
similar to the one described in the preceding paragraph. The signal
acquisition device is in operative communication with the antenna
and acquires and processes the GPS signals received by the
antenna.
[0018] Benefits of the present invention include the performance of
signal detection in the absence of a software-controlled threshold.
As a result, there is no sensitivity to errors in threshold
settings. The present invention also maximizes the use of the GPS
receiver architecture by operating the tracking channels, i. e.,
independent correlators, to perform signal verification/false alarm
in parallel within the signal detection correlators. The process of
the present invention also looks at more possible detections than
conventional processes. This provides improved probability of
detection.
[0019] These and other aspects and advantages of the invention will
become apparent from the following detailed description and the
accompanying drawings which illustrate byway of example the
features of the invention.
BRIEF DESCRIPTION OF THE DRAWINGS
[0020] FIG. 1 is a block diagram of an acquisition process
performed in accordance with one embodiment of the present
invention;
[0021] FIG. 2 is a graph showing the probability that less than six
noise hits (out of 32,000) are larger than the signal as a function
of the single cell probability; and
[0022] FIG. 3 is a block diagram of an acquisition process
performed in accordance with another embodiment of the present
invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0023] Referring now to the drawings, wherein the reference
numerals denote like or corresponding parts throughout the figures,
and particularly to FIG. 1, there is shown a block diagram of an
acquisition process 8 performed by an acquisition engine configured
in accordance with one embodiment of the present invention. The
acquisition engine is part of a GPS receiver which may be used in
commercial or military application to track and monitor the
location of moving objects such as automobiles and aircraft.
Details of such tracking systems are well known to those skilled in
the art and will not be described herein.
[0024] With reference to FIG. 1, the acquisition process 8
performed by the acquisition engine is divided into two steps:
parallel signal detection 10 and signal verification/false alarm
rejection 12. These two steps are distinct in both the software and
the hardware implementations.
[0025] The parallel signal detection process 10 uses a massively
parallel architecture of acquisition correlators 14 to quickly
search a large contiguous region of the time-frequency uncertainty.
The acquisition correlators 14 identify the most likely detections
16 for each search dwell. The dwell times for both coherent and
non-coherent integration are chosen to achieve the desired
sensitivity and/or jamming resistance. However, it is not necessary
to extend the dwell times of this parallel process to unambiguously
identify the signal, as this step is only the first step of a
sequential detection process. Instead, it is only necessary that
the true signal be included in a short list of most likely
detections.
[0026] Concurrent with the parallel signal detection process 10
performed by the acquisition correlators 14, the current list of
most likely detections 16 is examined with additional search dwells
in the verification/false alarm rejection process 12. The
verification/false alarm rejection process 12 is performed by a
plurality of independent correlators 18, i. e., tracking channels.
Under software control, the tracking channels 18 perform repeated
dwells on the most likely detections 16 until they can be dismissed
as false alarms, or verified as the desired signal.
[0027] By segregating the acquisition process 8 in this manner, the
search of the large time-frequency uncertainty performed by the
acquisition correlators 14 is limited to identifying likely
detections via the parallel signal detection process 10. Only for
these candidates is a signal verification/false alarm rejection
process 12 performed. This combination utilizes the features of an
acquisition engine according to its particular strengths; the
massively parallel acquisition correlators 14 are used to quickly
scan the large time-frequency uncertainty and the tracking channels
18 of the sequential detector accurately identifies the signal from
the false alarms.
[0028] An ancillary, but not insignificant, benefit of this
implementation is the elimination of the requirement for carefully
chosen thresholds during the detection process 10. In fact, the
parallel signal detection process 10 performed by the acquisition
correlators 14 generates a fixed-length list of the most likely
detections 16 without the use of thresholds. During the signal
verification/false alarm rejection process 12 performed by the
software controlled tracking channels 18, thresholds may be used.
However, the algorithm used in the signal verification/false alarm
rejection process 12 is inherently strong and threshold selection
may be performed with minimal sensitivity to variation in the value
of the threshold.
[0029] As described above, the acquisition process 8 is separated
into a parallel signal detection process 10 and a
verification/false alarm rejection process 12. These two processes
are described separately below. Also provided below is an analysis
of the algorithm used in the exemplary GPS receiver
architecture.
Acquisition Process
[0030] The following descriptions are presented in the context of a
GPS receiver architecture having 511 correlators and a 64-tap FFT,
herein referred to as "the 511 architecture", which has the
capability to simultaneously search 32,704 cells in the
time-frequency uncertainty domain. The invention, however, is not
limited to this exemplary GPS receiver architecture.
[0031] With continued reference to FIG. 1, the unique feature of
the signal detection process 10 of the present invention is related
to the most likely detections 16 reported to the software, i. e.,
software controlled tracking channels 18, by the signal processing
hardware, i. e., acquisition correlators 14. Traditional
architectures form a correlation output over the desired coherent
and non-coherent integration period and report the peak detection
that exceeds a software-controlled threshold. In accordance with
the present invention, a GPS receiver may be configured to report a
number of detections 16, any of which may be a GPS signal.
[0032] In a preferred embodiment, the acquisition correlators 14
may be configured to return the time/frequency coordinates of the
six maximum detections 16 to provide the potential detections from
the hardware to the software. The first advantage of this technique
is that there is no threshold used at all, and therefore no
sensitivity to an improperly calculated threshold. For the
probability of detection, it is only required that the true signal
be included in the list of the six strongest signals 16. This
technique does imply that there will be false alarms that must be
properly discriminated and therefore a strong false alarm rejection
algorithm is required.
[0033] With continued reference to FIG. 1, the GPS receiver
architecture may also include six tracking channels 18. These
tracking channels 18 are controlled separately from the acquisition
correlator 14, and are used for signal verification/false alarm
rejection 12 in parallel with continued signal detection 10.
[0034] To implement the desired architecture, the time/frequency
coordinates of the six largest signals 16.sub.1 or potential
detections, are read from the GPS receiver upon completion of an
initial acquisition dwell 20. The six sets of time/frequency
coordinates of the potential detections 16.sub.1 from the initial
acquisition dwell 20 are passed to the six tracking channels 18.
The GPS receiver is then commanded to immediately resume the signal
detection process 10 with a subsequent acquisition dwell 22 by
searching the next 32,704 cells in the time/frequency uncertainty
range using the acquisition correlators 14 for a subsequent set of
most likely detections 16.sub.2 This guarantees that the
acquisition process 8 continues, unperturbed by the signal
verification/false alarm rejection process 12, which is performed
in parallel.
[0035] Each tracking channel 18 is used as an independent
correlator. Each of the six correlator power outputs 16.sub.1
(outputs 1-6) is compared to a software controlled threshold during
an initial verification process 24, and those that fail to meet the
threshold are discarded as false alarms while those that meet the
threshold are retained in the tracking channels 18. With reference
to FIG. 1, an exemplary initial verification process 24 yielded no
outputs which met the threshold requirements thus the six most
likely detections 16.sub.2 from the subsequent acquisition dwell 22
are passed to the set of six tracking channels 18.
[0036] The GPS receiver is then commanded to immediately resume the
signal detection process 10 with a second subsequent acquisition
dwell 26 while a subsequent verification process 28 is performed on
the subsequent set of detections 16.sub.2 (outputs 7-12). In the
exemplary subsequent verification process 28 shown in FIG. 1, one
of the six outputs 16.sub.2 (output "11") has met the threshold and
is retained in the tracking channels 18. Because output "11" has
been retained, only the five most likely detections 16.sub.3 from
the second subsequent acquisition dwell 26 are passed to the
tracking channels 18. The GPS receiver again is commanded to
immediately resume the signal detection process 10 with a third
subsequent acquisition dwell 30 while a subsequent verification
process 32 is performed on output "11" and the five new detections
16.sub.3. In the exemplary process of FIG. 1, previously retained
output "11" has met the threshold again and is retained in the
tracking channels 18 while each of the five new detections (outputs
13-17) have been discarded. Again, because output "11" has been
retained, only the five most likely detections 16.sub.4 (outputs
18-22) from the third subsequent acquisition dwell 30 are passed to
the tracking channels 18. In the last signal verification process
34 previously retained output "11" once again satisfies the
software controlled threshold requirement and is declared the
signal detection. By insisting that the detector power output, e.
g., output "11", exceed the software threshold two or three times
in a row the reliability of this process is increased.
[0037] Two drawbacks are immediately apparent from this false
alarm/signal verification algorithm. The first is the return to
reliance on a software computed threshold. While this is true, the
analysis that follows demonstrates that the sensitivity to
perturbations in this threshold is very small. Conceptually, this
is because it is only necessary to detect the signal in the
presence of five false alarms, rather than the original detection
process which is trying to detect the signal in the presence of
32,703 potential false alarms.
[0038] The second drawback is the requirement for multiple
detections prior to declaring a detection. Assuming that the signal
verification/false alarm rejection dwell 24, 28, 32, 34 is of the
same length as the acquisition dwell 20, 22, 26, 30, if a signal
exceeds the threshold in the tracking channel 18, requiring that it
be detected a second or third time allows only five remaining
tracking channels to be used for subsequent signal
verification/false alarm rejection processes. While this is true,
thresholds can be set sufficiently high to limit this occurrence to
negligibly small rates without compromising the probability of
detection significantly. The foregoing is quantified in the
probability analysis that follows.
Analysis of Algorithm
[0039] The probability of the success of the total algorithm
includes the probability that the signal was detected, the signal
verification process was successful, and that there were no false
alarms. This can be represented as:
P.sub.success=P.sub.dct*P.sub.ver*Q.sub.FA where P is used to
denote the probability of an event being successful, and Q
represents the intrinsically-defined complementary probability that
the event is not successful, where P+Q.ident.1.
[0040] In performing the subsequent calculations, all probabilities
are assumed to be Gaussian in nature. While some of these processes
are more accurately defined by other distributions, the variations
due to this simplification are felt to be small.
Analysis of Probability of Detection
[0041] The analysis begins with the assumption that the desired
probability of acquisition for the first satellite is 98%. This is
consistent with a desired probability of acquisition of enough
satellites for a navigation solution of 95%. Obviously, the
techniques identified below are valid for other probabilities, and
it is a simple matter to repeat the analyses for alternate desired
probabilities.
[0042] Given the architecture of the GPS chip, it is necessary to
determine the probability that there are not six or more cell
locations in which the measured power (due strictly to noise) is
greater than the measured power in the cell which contains the
signal (together with its noise). Given the 98% overall
requirement, this determines the probability that the noise does
not exceed the signal plus noise in any single cell.
[0043] Given this latter probability, it is straightforward to
compute the S/N ratio that is required to meet or exceed this
probability. The result is then the S/N ratio required to insure a
single-satellite detection probability of 98%.
Probability that the Signal is in the Top Six
[0044] The probability that the signal is within the top six is the
probability that there are exactly 0 cells which have noise which
exceeds the signal plus noise, plus the probability of exactly 1,
plus the probability of exactly 2, plus the probability of exactly
3, plus the probability of exactly 4, plus the probability of
exactly 5: P det = i = 0 i = 5 .times. P .function. ( n = i )
##EQU1##
[0045] The probability that there are exactly i detections, is
given by the binomial probability function: P .function. ( n = i )
= ( N i ) .times. p i .times. q N - i ##EQU2## where N=32703,
q=1-p, p is the single-cell probability that the noise exceeds the
signal plus noise, and ( N i ) = N ! ( N - i ) ! .times. i !
##EQU3## Combining these equations gives: P det = 0.98 ##EQU4## P
det = i = 0 i = 5 .times. ( N i ) .times. p i .times. q N - i
##EQU4.2##
[0046] A curve depicting P.sub.det vs. p is presented in FIG. 2.
This shows that for the desired value of P.sub.det=0.98, the
required value of p is 0.0000639 (63.9 ppm).
Single Cell Probability of Detection
[0047] The remaining step is to determine the S/N ratio required to
support the single-cell probability of high noise determined above.
This can be stated as: Prob {S+X.sub.1<X.sub.2} where X.sub.2
and X.sub.2 are independent samples of a unit-variance, zero-mean,
normal distribution.
[0048] This can be rewritten as: Prob{S<X.sub.3}, where
X.sub.3.ident.X.sub.2-X.sub.1, and is a sample of zero-mean normal
distribution with a variance of 2. With this simplification, a
signal-to-noise ratio of 5.42 is required to achieve the desired
single cell probability of 0.0000639. Verification Algorithm
[0049] With the requirement imposed by the detection process that
the S/N ratio must be 5.42, or higher, it is straightforward to
design a strict verification algorithm. As described above, the
algorithm to be used is to require multiple consecutive detections
using a software-controlled threshold. The probability that the
signal plus noise will exceed the threshold is a straight
cumulative Gaussian probability: Prob{X>T/N-S/N}, where X is a
zero mean, unit variance random variable, T/N is the
threshold-to-noise ratio, and S/N is the signal-to-noise ratio.
[0050] The probability of false alarm is similarly straightforward:
Prob{X>T/N}.
[0051] Using a value of 2.7 for T/N gives a 99.65% probability of
correctly verifying the true signal, and only a 0.35% probability
of falsely verifying a false alarm. Requiring two consecutive
verification dwells using a lower threshold provides an even more
robust verification algorithm. For example, using a T/N ratio of
2.0, requiring two consecutive verifies would correctly confirm
99.93% of true signals, and would falsely verify only 0.052% of
false alarms. Increasing the number of consecutive verifications
that are required can provide an even more robust algorithm,
although this rapidly reaches a point of diminishing returns.
[0052] While requiring a false alarm probability as low as 0.052%
may seem unnecessarily restrictive, it must be remembered that in
each dwell there are six false alarms which must be rejected, and
the number of dwells may be very large if a large time/frequency
uncertainty must be searched. For example, if the total time
uncertainty is .+-.10 msec, and the frequency uncertainty is .+-.1
ppm, it will require about 1900 search dwells to span the entire
time/frequency uncertainty (assuming a 20 msec coherent integration
time), and therefore about 11,400 successful false alarm
rejections. In fact if the total false alarm rate is meant to be
kept to less than 1%, the individual false alarm rate is given by:
Q.sub.fa.sup.1400>99% P.sub.fa<0.88 ppm
[0053] This false alarm rate would require a T/N ratio setting of
greater than 3.1, which would compromise the probability of
detection (it would degrade by about 1%). Therefore, the selected
strategy is to require three consecutive verifications, using a T/N
ratio value of 2.35. This results in a probability of verification
of 99.9%, and a probability of false alarm of <1% (assuming a
S/N of >5.42, and measuring the false alarm probability over the
entire .+-.10 msec/.+-.1 ppm time/frequency uncertainty range).
Total Probability of Success
[0054] Combining the separate probabilities, the total probability
of success is given as: P.sub.success=P.sub.det*P.sub.verQ.sub.FA
P.sub.success=0.98*0.999*0.99 P.sub.success=0.969
[0055] These probabilities are for successfully acquiring the first
satellite in a S/N environment of 5.42, while searching the
time/frequency uncertainty range of .+-.10 msec/.+-.1 ppm.
[0056] The advantages of the algorithm are best viewed by
contrasting it with the acquisition algorithm of a conventional
receiver. In such a receiver, the largest detection in the array of
the same 32,704 time/frequency cells is compared against a
threshold, If it exceeds the threshold, the signal
verification/false alarm rejection algorithm uses the same
acquisition correlators that are used to attempt to detect the
signal in multiple dwells to confirm the presence of the
signal.
[0057] The biggest drawback to such a scheme is that it requires
the signal to be the largest detection. Mathematically, this means
that the probability that a noise sample exceeds the signal plus
noise must not exceed the allowable 2% missed detection
probability. For the same 32,704 cells, this requires a S/N ratio
of 6.95 (as compared with 5.42 for the algorithm described above).
Achieving such a S/N ratio implies either a lower jam immunity/weak
signal sensitivity, or an increase in the TTFF while the signal is
integrated to the required S/N ratio.
[0058] This points out the qualitative differences between the
detection algorithm for a massively parallel architecture as
compared with algorithms which have proven successful in the past
for more limited architectures. While previous detection algorithms
have had to be concerned with setting thresholds for signal
detection and less concerned with false alarms masking the signal,
the existence of a large number of time/frequency cells being
searched in parallel increases the importance of false alarms.
[0059] The algorithm described herein addresses this issue in two
ways. First, instead of the hardware reporting only a single
potential detection, it reports multiple detections. This vastly
reduces the complexity of the problem to verifying the existence of
the true signal from among a handful of potential detections. The
second significant aspect is that the tracking channels are used
for verification/false alarm rejection in parallel with the main
acquisition correlators. Given the importance that has been
transferred to this second half of the algorithm, it is imperative
that this be performed in parallel, thereby maintaining the rapid
acquisition capabilities of the massively parallel
architecture.
[0060] With reference to FIG. 3, in an alternate embodiment of the
invention, the acquisition process is applied to a GPS receiver
having 240 correlators and 16 FFT taps. This GPS receiver is
referred to herein a "the 240 architecture." Similar to the 511
architecture, the 240 architecture returns the largest correlation
peaks following each search dwell. A second aspect of the 240
architecture is that it does not include separate tracking
channels. Instead, the 240 acquisition correlators 40 can be
reconfigured as twelve 20-tap tracking channels 42. Normally these
are used to track twelve satellites. However, following the
strategy described above, these tracking channels 42 are used as
twelve independent acquisition channels for the purpose of signal
verification/false alarm rejection.
[0061] Following the first search 441, the hardware list of the
twelve largest correlations peaks are placed on a software list 46
of the most likely detections. With 240 chips at 1/2-chip spacing,
it takes nine searches 48 with the chip to search the entire 1023
chips of the C/A code. For the second through ninth searches
442-449, the newly determined largest correlation peaks are merged
with the pre-existing list, with only the top twelve peaks being
retained. The result is that following the search of the complete
C/A code in nine dwells, the list contains the twelve largest
correlation peaks over the 2046 code positions.
[0062] Without delay, the acquisition correlator is then
reconfigured as twelve independent acquisition channels 42, each
assigned one of the most likely detections from the list of twelve.
A verification dwell 50, typically implemented as twice the dwell
length of the acquisition dwell, is then performed. The output of
the verification dwell 50 replaces the software list 46.
[0063] The hardware is then reconfigured as the 240-tap acquisition
correlator 40 and the acquisition process continues. Signals from
the verification dwell 50 that correspond to false alarms will
naturally be replaced by stronger detections from the subsequent
acquisition dwells. Signals from the verification dwell 50 that
correspond to the true signal will persist on the list. The
software requires that a detection be verified twice (although this
is a software controlled constant) before final validation as the
true signal.
[0064] This implementation retains most of the advantages described
above for the 511 architecture implementation. The massively
parallel architecture is used to screen the large number of
time/frequency cells to provide a small list of most likely
detections. A sequential detector process is then used on this
small list for the verification/false alarm rejection algorithm.
One difference with this implementation is that the parallel and
sequential detectors are run serially, and this does represent an
increase in TTFF of 22%. However, this is more than offset by the
time reduction associated with the reduced S/N requirements for
this implementation.
[0065] The mathematics for calculating the probability of
acquisition are similar to the algorithms used for the 511
architecture. For the 240 architecture, in order for the signal to
be contained in the twelve most likely detections of 240 code
positions, the S/N must be at least 4.42. In another GPS receiver
architecture having 240 separate taps and an 8-point FFT the S/N
must be at least 5.22 to insure the same 98% probability of
detection. A conventional detection algorithm would require S/N
values of 5.32 and 6.02 for these architectures, respectively.
[0066] It will be apparent from the foregoing that while particular
forms of the invention have been illustrated and described, various
modifications can be made without departing from the spirit and
scope of the invention. Accordingly, it is not intended that the
invention be limited, except as by the appended claims.
* * * * *