US2020116851A1PendingUtilityA1
Robust Constant False Alarm Rate (CFAR) Detector for Interference-Plus-Noise Covariance Matrix Mismatch
Est. expiryOct 10, 2038(~12.1 yrs left)· nominal 20-yr term from priority
Inventors:Ramachandran S. Raghavan
G01S 7/414G01S 13/582G01S 13/584G01S 7/418G01S 7/023G01S 13/5246
40
PatentIndex Score
0
Cited by
0
References
0
Claims
Abstract
Detection of a radar target from a received radar signal includes computing a vector of filter weights dependent upon a steering vector and determining a threshold value dependent upon a designated probability of false alarm. The vector of filter weights is applied to samples of the received radar signal at a test cell, corresponding to a test range, to provide a filtered test signal and a test power of the filtered test signal is computed. The weights are also applied to samples of the received radar signal at a number of reference cells, to produce filtered reference signals. A reference power is computed from the filtered reference signals and the radar target is detected at the test range when a ratio of the test power to the reference power exceeds the threshold value.
Claims
exact text as granted — not AI-modifiedWhat is claimed is:
1 . A method for detecting a radar target, the method comprising:
computing a signal transformation matrix T −1 from a matrix Y of training vectors as m T − =(YY H ) † , denotes a conjugate transpose of the matrix Y and the symbol ‘†’ denotes a matrix inverse or pseudo-inverse; computing a vector w of filter weights dependent upon a steering vector s as w=T −1 s; determining a threshold value η dependent upon a designated probability of false alarm P FA ; filtering a received radar signal using the vector of filter weights to provide a plurality of filtered radar signals as w H z n , where w H denotes a conjugate transpose of the vector w of filter weights and z n is a vector of samples of the received radar signal at an n th range/Doppler cell; computing a test power of the filtered radar signal at a test cell corresponding to a test range for the radar target as |w H z 0 | 2 ; computing a reference power G from a power of the filtered radar signals |w H z k | 2 at a plurality M of reference cells, indexed by k, corresponding to reference ranges for the radar target other than the test range, where the reference cells are indexed by k; and detecting the radar target at the test range when a ratio of the test power to the reference power exceeds the threshold value according to
w
H
z
0
2
G
(
w
H
z
1
2
,
w
H
z
2
2
,
…
,
w
H
z
M
2
)
>
η
,
where the threshold value η is determined from the probability P FA of false alarm.
2 . The method of claim 1 , where the radar target is detected at the test range when |w H z 0 | 2 >ηG(|w H z 1 | 2 , |w H z 2 | 2 , . . . |w H z M | 2 ), which comprises a comparison of the test power |w H {tilde over (z)} 0 | 2 to the reference power G scaled by the threshold value η.
3 . The method of claim 1 , where the reference power is computed from the power of the filtered radar signals |w H z k | 2 at M reference cells as
G
(
w
H
z
k
2
)
=
∑
k
=
1
M
{
w
H
z
k
2
}
,
and where the threshold value η is determined from the probability P FA of false alarm. through the equality
P
FA
=
1
(
1
+
η
)
M
.
4 . The method of claim 1 , where computing the reference power comprises computing an order statistic of the power of the filtered radar signals |w H z k | 2 at M reference cells.
5 . A non-transitory computer readable medium containing programming instructions that, when executed on a processor, perform the method of claim 1 .
6 . A method for detecting a radar target from a received radar signal, the method comprising:
computing a vector of filter weights dependent upon a steering vector; determining a threshold value dependent upon a designated probability of false alarm; applying the vector of filter weights to samples of the received radar signal at a test cell, corresponding to a test range, to provide a filtered test signal; computing a test power of the filtered test signal; applying the vector of filter weights to samples of the received radar signal at a plurality of reference cells, to produce a plurality of filtered reference signals; computing a reference power from the plurality of filtered reference signals; and detecting the radar target at the test range when a ratio of the test power to the reference power exceeds the threshold value.
7 . The method of claim 6 , where computing the vector of filter weights comprises computing a product of the steering vector with a signal transformation matrix.
8 . The method of claim 7 , further comprising computing the signal transformation matrix as an inverse covariance matrix of a set of training vectors prior to receiving the samples at the test cell and the reference cells.
9 . The method of claim 8 , where the inverse covariance matrix of the set of training vectors is mismatched with an inverse covariance matrix of interference and noise in the samples of the received radar signal at the test cell.
10 . The method of claim 7 , where the signal transformation matrix is independent of the samples at the test cell and the samples at the reference cells.
11 . The method of claim 6 , where computing the reference power comprises summing powers of the plurality of filtered reference signals.
12 . The method of claim 6 , where computing the reference power comprises ordering powers of the plurality of filtered reference signals.
13 . The method of claim 6 , where detecting the radar target at the test cell when the ratio of the test power to the reference power exceeds the threshold value comprises:
scaling the reference power by the threshold value to provide a scaled reference power; and detecting the radar target when the test power is greater than the scaled reference power.
14 . A radar system for detecting a target, where the target is detected with a specified probability of false alarm, the system comprising:
a filter weight computation engine configured to compute a vector of filter weights dependent upon a steering vector and a transformation matrix; a memory for storing the vector of filter weights; a filter configured to produce a filtered radar signal by applying the vector of filter weights to samples of a received radar signal; and a processor configured to:
compute a threshold value dependent upon the specified probability of false alarm;
compute a test power from a power of the filtered radar signal at a test cell corresponding to a test range;
compute a reference power from a power of the filtered radar signal at a plurality of reference cells other than the test cell; and
detect the radar target at the test range when a ratio of the test power to the reference power exceeds a threshold value.
15 . The radar system of claim 14 , where the vector of filter weights is computed as a product of the steering vector with a signal transformation matrix.
16 . The radar system of claim 15 , where the signal transformation matrix is computed as an inverse covariance matrix of a set of training vectors prior to the radar signal being received.
17 . The radar system of claim 16 , where the inverse covariance matrix of the set of training vectors is mismatched with an inverse covariance matrix of interference and noise in the test cell.
18 . The radar system of claim 14 , where the signal transformation matrix is independent of the received radar signal.
19 . The radar system of claim 14 , where the processor is configured to compute the reference power as a sum of the power of the filtered radar signal at the plurality of reference cells.
20 . The radar system of claim 14 , where the processor is configured to compute the reference power as an order statistic of the power of the filtered radar signal at the plurality of reference cells.
21 . The radar system of claim 14 , where the processor is configured to detect the radar target at the test cell when the ratio of the test power to the reference power exceeds the threshold value by:
scaling the reference power by the threshold value to provide a scaled reference power; and detecting the radar target when the test power is greater than the scaled reference power.Join the waitlist — get patent alerts
Track US2020116851A1 — get alerts on status changes and closely related new filings.
We store only your email — no account needed. See our privacy policy.