Adaptive finite impulse response (FIR) filter and method

Abstract

An adaptive finite impulse response (FIR) filter for filtering an I/Q data stream having a sample period. The FIR filter comprises at least one sample-period tap delay configured to delay the I/Q data stream by an integer multiple of the sample period of the I/Q data stream, and at least one sub-sample-period tap delay configured to delay the I/Q data stream by a non-integer multiple of the sample period of the I/Q data stream. A set of adaptive weights is provided and configured to weight samples of the delayed I/Q data stream. An adder is responsive to the weighted samples and configured to combine the weighted samples of the delayed I/Q data streams to generate a filtered I/Q data stream.

Claims

1. An adaptive finite impulse response (FIR) filter for filtering an I/Q data stream, comprising: at least one sample-period tap delay configured to delay the I/Q data stream by an integer multiple of a sample period of the I/Q data stream; at least one sub-sample-period tap delay arranged in parallel with the at least one sample-period tap delay and configured to delay the I/Q data stream by a non-integer multiple of the sample period of the I/Q data stream; a set of adaptive weights configured to weight samples of the delayed I/Q data stream delayed from the at least one sample-period tap delay and the at least one sub-sample period tap delay; and an adder configured to combine the weighted samples of the delayed I/Q data streams to generate a filtered I/Q data stream.

2. The FIR filter of claim 1, wherein the at least one sample-period tap delay comprises a plurality of sample-period tap delays configured to delay the I/Q data stream by integer multiples of the sample period of the I/Q data stream.

3. The FIR filter of claim 2, wherein the at least one sub-sample-period tap delay comprises a plurality of sub-sample-period tap delays configured to delay the I/Q data stream by non-integer multiples of the sample period of the I/Q data stream.

4. The FIR filter of claim 3, wherein the sub-sample period tap delays are at least an order of magnitude less than the sample period tap delays.

5. The FIR filter of claim 1, wherein the weights are fixed real weights.

6. The FIR filter of claim 1, wherein the sub-sample-period is a non-integer multiple of the sample period.

7. The FIR filter of claim 1, further comprising a zero-delay tap arranged in parallel with the at least one sample-period tap delay and the at least one sub-sample-period tap delay and responsive to the I/Q data stream.

8. The FIR filter of claim 1, further comprising a weight computation device responsive to sampled I/Q data for computing the adaptive weights.

9. The FIR filter of claim 1, wherein a group delay of the FIR filter is ((M−1)/2)T+τ), wherein M is the number of tap delays, T is the sample period and τ is the sub-sample period delay.

10. A method of filtering an I/Q data stream with an adaptive finite impulse response (FIR) filter, comprising: delaying the I/Q data stream by an integer multiple of a sample period of the I/Q data stream with at least one sample-period tap delay of the FIR filter; delaying the I/Q data stream by a non-integer multiple of the sample period of the I/Q data stream with at least one sub-sample-period tap delay of the FIR filter arranged in parallel with the at least one sample-period tap delay; weighting the samples of the delayed I/Q data stream delayed from the at least one sample-period tap delay and the at least one sub-sample period tap delay with a set of adaptive weights; and combining the weighted samples of the delayed I/Q data streams with an adder of the FIR filter to generate a filtered I/Q data stream.

11. The method of claim 10, wherein the at least one sample-period tap delay comprises a plurality of sample-period tap delays configured to delay the I/Q data stream by integer multiples of the sample period of the I/Q data stream.

12. The method of claim 11, wherein the at least one sub-sample-period tap delay comprises a plurality of sub-sample-period tap delays configured to delay the I/Q data stream by non-integer multiples of the sample period of the I/Q data stream.

13. The method of claim 12, wherein the sub-sample period tap delays are at least an order of magnitude less than the sample period tap delays.

14. The method of claim 10, wherein the weights are fixed real weights.

15. The method of claim 10, wherein the sub-sample-period is a non-integer multiple of the sample period.

16. The method of claim 10, further comprising the step of providing the I/Q data stream to a zero-delay tap arranged in parallel with the at least one sample-period tap delay and the at least one sub-sample-period tap delay.

17. The method of claim 10, further comprising the step of computing the adaptive weights with a weight computation device responsive to sampled I/Q data.

18. The method of claim 10, wherein a group delay of the FIR filter is ((M−1)/2)T+τ), wherein M is the number of tap delays, T is the sample period and τ is the sub-sample period delay.

19. The FIR of claim 1, wherein the set of adaptive weights comprises a first weight to be applied to the delayed I/Q data stream delayed from the at least one sample-period tap delay and a second weight to be applied to the delayed I/Q data stream delayed from the at least one sub-sample-period tap delay.

20. The method of claim 10, further comprising the step of receiving the I/Q data stream from an auxiliary antenna of a radar system, wherein the auxiliary antenna receives one or more jamming signals to be removed from an I/Q data stream received by a main antenna of the radar system.

Description

BRIEF DESCRIPTION OF THE FIGURES

(1) Understanding of the present invention will be facilitated by consideration of the following detailed description of the preferred embodiments of the present invention taken in conjunction with the accompanying drawings, in which like numerals refer to like parts and in which:

(2) FIG. 1 shows a block diagram of an adaptive radar system having one or more sub-sample-period filters according to an embodiment of the present invention;

(3) FIG. 2 shows a block diagram of functional components of a sub-sample period adaptive filter of FIG. 1 where T is the I/Q sample period and τ.sub.1 through τ.sub.N1 are sub-sample-period time delays that are less than T and the W.sub.1, W.sub.2, . . . , are complex adaptive weights;

(4) FIG. 3 shows a block diagram of functional components of an exemplary all-pass 7-tap FIR filter configured to delay a received data stream by 3T+τ according to an embodiment of the present invention where τ is a sub-sample-period time delay and α.sub.1 through α.sub.7 are real weights;

(5) FIG. 4 shows a block diagram of functional components of an exemplary adaptive weight computation device for the adaptive radar system of FIG. 1 having a single auxiliary channel (i.e. N=1) according to an embodiment of the present invention; and

(6) FIG. 5 shows a process flow for a sub-sample-period adaptive FIR filter suitable for use by the adaptive radar system of FIG. 1 according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

(7) It is to be understood that the figures and descriptions of the present invention have been simplified to illustrate elements that are relevant for a clear understanding of the present invention, while eliminating, for purposes of clarity, other elements found in typical adaptive processing systems and methods (e.g. radar systems). However, because such elements are well known in the art, and because they do not facilitate a better understanding of the present invention, a discussion of such elements is not provided herein.

(8) FIG. 1 shows an adaptive radar system 100 according to an embodiment of the present invention. Adaptive radar system 100 includes a main antenna 102, such as a phased array antenna, configured to transmit and receive radar signals and to output received radio frequency (RF) echo signals to a downstream main channel receiver 104. Main channel receiver 104 may be a typical radar receiver configured to receive RF signals from main antenna 102, amplify the received signals, and output intermediate frequency (IF) signals to downstream signal processing components illustrated collectively in FIG. 1 as main processing block 108. Main processing block 108 may include one or more typical radar processing components/devices configured to receive IF signals from the main receiver, perform analog-to-digital (ND) conversion, digital demodulation (DDM), and equalization (i.e. channel matching), and to output I/Q samples at an I/Q sample rate of (f.sub.S) hertz (Hz) to a downstream main digital beamformer 110. Main digital beamformer 110 may be a typical digital beamformer configured to receive I/Q samples at f.sub.S Hz and to output an I/Q data stream from the main beam or subarray beam to a downstream noise cancellation node 112. In the exemplary embodiment, the I/Q sample rate f.sub.S Hz may be about 25% greater than the bandwidth of the channel (e.g. f.sub.S=5 megahertz (MHz) for a radar system with a 4 MHz bandwidth).

(9) Adaptive radar system 100 also includes one or more auxiliary channels (1, 2, . . . , N) useful for cancelling out noise or jamming signals received on the main channel. Each auxiliary channel includes an auxiliary antenna 122.sub.i (where i=1, . . . , N) configured to receive target free and clutter free signals and to output RF signals to a downstream auxiliary receiver 124.sub.i (where i=1, . . . , N). Auxiliary antennas 122.sub.i may be separate antennas from main antenna 102 or they may be subarrays of the main antenna. In an exemplary embodiment, each subarray could be the entire antenna aperture with the only difference between the main and auxiliary channels being the beam shaping and steering weights applied to the array elements (i.e. one set of weights is applied to form the main beam and another set is applied in parallel to form an auxiliary beam). In an alternative exemplary embodiment, each subarray may comprise only a single array element. Each auxiliary receiver 124.sub.i may be a typical radar receiver configured to receive RF signals from an auxiliary antenna 122.sub.i, amplify the received signals, and output IF signals to downstream signal processing components, illustrated collectively in FIG. 1 as auxiliary processing block 126.sub.i (where i=1, . . . , N). Auxiliary processing block 126.sub.i may include one or more typical radar processing components/devices configured to receive IF signals from the main receiver, perform ND conversion, DDM, and equalization (i.e. channel matching), and to output I/Q samples at a sample rate of f.sub.S Hz to a downstream auxiliary digital beamformer 128; (where i=1, . . . , N). Auxiliary digital beamformer 128.sub.i may be a typical digital beamformer configured to receive I/Q samples at f.sub.S Hz and to output an I/Q data stream downstream to both an adaptive weight computation device 132 and a sub-sample-period (SSP) adaptive filter such as finite impulse response (FIR) filter 130.sub.i (where i=1, . . . , N).

(10) Adaptive weight computation device 132 is a processing component configured to receive a subset of the total available I/Q data stream selected to contain minimal or no target returns and clutter signal returns and to output a plurality of adaptive weights to each of N downstream SSP adaptive FIR filters 130.sub.i. By way of example, adaptive weight computation device 132 may be configured to receive I/Q data streams only during relatively short “listening windows” occurring immediately before each transmit pulse, thereby minimizing or eliminating target returns and clutter signal returns caused by transmissions from the main channel. The frequency at which adaptive weight computation device 132 computes adaptive weights depends on aspects of the radar system. For example, in a fixed radar adaptive weights may be calculated relatively infrequently (e.g. once per coherent integration period (CIP)). Alternatively, in a rotating radar, adaptive weights may require updating during a CIP (e.g. at the pulse repetition frequency (PRF)). The processing performed by adaptive weight computation device 132 is discussed below with reference to FIG. 4.

(11) Each SSP adaptive FIR filter 130.sub.i is a processing component configured to receive both an I/Q data stream at a frequency of f.sub.S Hz and one or more adaptive weights. Each SSP adaptive FIR filter adaptively matches the frequency response of the I/Q data stream to that of the main channel I/Q data stream using a number of sample period delays and a number of sub-sample period delays (less than the sample period delays), for output to downstream noise cancellation node 112. The processing performed by SSP adaptive FIR filter 130.sub.i is discussed below with reference to FIG. 2.

(12) Cancellation node 112 may be a typical adder configured to receive a main channel I/Q data stream contaminated with components of jamming signal noise, and one or more auxiliary channel I/Q data streams configured to output the various jamming signal components of the main channel. The auxiliary channel I/Q data streams from the main channel I/Q data stream, and cancellation node 112 outputs to downstream components a jam-free I/Q data stream 113. Further downstream components may include typical radar components, such as by way of example only a coherent signal processor 114, a detection and data processor 116, a tracker 118, and a command and control processor 120.

(13) FIG. 1 shows SSP adaptive FIR filters 130.sub.i implemented in a digital beamforming radar. It should be clear to those of skill in the art, however, that the SSP adaptive FIR filters 130.sub.i may be similarly implemented in a conventional radar (e.g. a radar having main digital beamformer 110 upstream in the I/Q sample flow from main processing block 108). Additionally, while the adaptive radar system 100 of FIG. 1 performs equalization in the main processing block 108 prior to digital beamforming in main digital beamformer 110, alternative embodiments of the present invention may omit equalization prior to digital beamforming and include a more general SSP adaptive FIR filter 130.sub.i in each auxiliary channel. In such an embodiment, SSP adaptive FIR filter 130.sub.i may comprise at least several sample-periods and SSP adaptive weights.

(14) FIG. 2 shows an SSP adaptive FIR filter 130.sub.i discussed above with reference to FIG. 1. SSP adaptive FIR filter 130.sub.i is configured to receive at input 1302 I/Q samples at f.sub.S Hz from a corresponding auxiliary digital beamformer 128.sub.i (shown in FIG. 1). A tapped delay line 1304 delays the I/Q sample stream before the I/Q sample stream is multiplied by each of complex adaptive weights w.sub.i indicated generally as 1306 (where i=1, 2, 3, . . . , N.sub.1+N.sub.2+1). Complex adaptive weights w.sub.i are trained by adaptive weight computation device 132 (discussed below with reference to FIG. 4). The products of multiplying the I/Q sample stream by each w.sub.i are then summed together by a summer 1308. Summer 1308 outputs to the downstream noise cancellation node 112 (FIG. 1) the I/Q sample streams whose channel passband is matched in the sense that the complex frequency response (i.e., amplitude and phase) across the passband has substantially the same shape as the main channel's frequency response over the same passband to within a predetermined tolerance.

(15) The tap delay line may include a zero delay tap, N.sub.1 SSP taps (τ), and N.sub.2 sample-period taps (T=1/f.sub.S). The SSP taps τ may be significantly less in duration than T. By way of example, in one embodiment of SSP adaptive FIR filter 130.sub.i, T may be 200 nanoseconds (ns) and accordingly 2T may be 400 ns. However, τ.sub.1 may be about one nanosecond (ns) and τ.sub.2 may be two nanoseconds. Thus, a sub-sample-period τ may be hundreds of times smaller than sample-period T. An all-pass M-tap f.sub.S Hz FIR filter 1310.sub.i whose frequency response closely approximates unit gain and linear phase slope across the radar's instantaneous bandwidth may be implemented for delaying the sampled I/Q data stream by τ. An exemplary FIR filter 1310.sub.i is discussed with reference to FIG. 3 below.

(16) It should be understood by one of ordinary skill in the art that the delays τ and T shown in FIG. 2 (and in FIGS. 3 and 4 below) are relative delays. Each SSP adaptive FIR filter 130.sub.i additionally has a bulk delay, and all I/Q data streams are delayed by this bulk delay to render delays of T and T delays relative to the main channel.

(17) Various applications of the SSP adaptive FIR filter 130.sub.i may require a mix of SSP taps and sample-period taps as shown, while other applications may require only SSP taps. By way of example only, an SSP adaptive FIR filter 130.sub.i configured to replace fixed channel equalization and support adaptive processing, may need a wide range of tap delays to accommodate both slow and fast variations of the multiple channel responses. By way of alternative example, an SSP adaptive FIR filter 130.sub.i may be configured in an adaptive radar system 100 having fixed equalization to remove most of the frequency-response variations from the multiple channels, leaving only time delay differences due to the changing aspect angle of the radar antenna relative to a set of jammers as the radar rotates. Such an SSP adaptive FIR filter 130.sub.i may need only a single SSP tap to adaptively match the changing time delays.

(18) FIG. 3 shows an exemplary all-pass 7-tap f.sub.S Hz FIR filter 1310.sub.i whose frequency response closely approximates unit gain and linear phase slope across the radar's instantaneous bandwidth. FIR filter 1310.sub.i is configured to receive at an input a sampled I/Q data stream Y. The sampled I/Q data stream Y is representative of an analog signal y(t) sampled at a given sample period T. The FIR filter receives the sampled I/Q data stream and generates at its output a sampled “version” Y′ of the analog signal y (t−3T−τ). This sampled version Y′ has the same sample period T, but the underlying analog signal has been shifted in time by the bulk delay (e.g. 3T) and the targeted timed delay (e.g. τ). In the illustrated embodiment, the system is configured to output the same sampled I/Q data stream delayed by 3T+τ, with 3T being the bulk delay and τ being the delay relative to the main channel, where τ is on the order of hundreds of times smaller than the sample period (τ<<T).

(19) Referring still to FIG. 3, FIR filter 1310.sub.i includes seven sample period (T) delays 1330. The delayed sampled I/Q data stream is multiplied by fixed real weights 1332 (details of selection of fixed real weights be α.sub.1-α.sub.7 designated as 1332 is discussed with reference to FIG. 5 below). The results of the multiplication with the fixed real weights 1332 are added by summer 1334, which outputs a sampled I/Q data stream delayed from the input by 3T+τ. By delaying the main channel I/Q data stream by 3T (i.e. the bulk delay of the exemplary all-pass 7-tap f.sub.S Hz FIR filter 1310.sub.i), a delay in the auxiliary channel relative to the main channel of τ is thereby achieved.

(20) Alternative embodiments of the present invention may implement any M-tap I/Q f.sub.S Hz FIR filter 1310.sub.i whose frequency response closely approximates unit gain and linear phase slope across the radar's instantaneous bandwidth. M must be chosen to be an odd integer. Thus, if the phase slope dφ/df across the radar's instantaneous bandwidth I is 2π, the group delay of the filter will be ((M−1)/2)T+τ). The linear phase in the frequency domain is equivalent to a time delay in the time domain. Thus, the main channel may be delayed by (M−1)/2 multiples of T, thereby achieving a delay in the auxiliary channel relative to the main channel of τ. It is appreciated by one of ordinary skill in the art that negative delays may be produced by an M-tap I/Q f.sub.S FIR filter 1310.sub.i since τ represents a relative time increment.

(21) Referring now to FIG. 4, an exemplary adaptive weight computation device 132 is shown for an adaptive radar system of FIG. 1 having a single auxiliary channel (i.e. N=1). Adaptive weight computation device 132 is configured to receive at an input an I/Q data stream from the main digital beamformer 110 as well as an I/Q data stream from the auxiliary digital beamformer 128.sub.i. The streams from the main digital beamformer 110 and from auxiliary digital beamformer 128.sub.i are configured to receive target-free and clutter-free interference-plus-noise I/Q samples. The input I/Q data stream 410 from the main channel may be input into an adaptive weight computation processor 1322 without any time delay. The input I/Q data stream 420 from the auxiliary channel may be input into a tap delay line 1320 which may include a zero-delay tap, SSP taps (τ), and sample-period taps (T). The delay tap τ may be significantly less than period T (e.g. on the order of hundreds of times smaller than period T). In embodiments of adaptive weight computation devices 132, tap delay line 1320 may be the same as tap delay line 1304 discussed with reference to FIG. 2 above. Tap delay line 1320 outputs time delayed I/Q data streams into adaptive weight computation processor 1322. It should be understood by those of skill in the art that each I/Q data stream is additionally delayed by a bulk delay (greater than T), as discussed with reference to FIG. 2 above. This bulk delay is an integer multiple of T and is a consequence of the generation of the subsample period delays (τ), wherein the main channel I/Q data stream is delayed by the bulk delay so as to provide a net or relative delay of (T), between the channels (relative delay of the auxiliary channel I/Q data streams with respect to the main channel I/Q data stream).

(22) Adaptive weight computation processor 1322 may be a conventional processor comprising hardware, software, firmware, or any combination thereof. Adaptive weight computation processor 1322 may implement one of numerous known algorithms selected to train the adaptive weights prior to interference cancellation. Adaptive weight computation is well known in the field of adaptive processing, the details of which are omitted herein for purposes of brevity. Adaptive weight computation processor 1322 outputs one or more adaptive weights w.sub.i to SSP adaptive FIR filter 130.sub.1.

(23) In alternative embodiments of the present invention, adaptive weight computation device 132 may be configured to receive inputs from main digital beamformer 110 and from N auxiliary digital beamformers 128.sub.(1-N). Such an adaptive weight computation device 132 would be configured to compute one or more adaptive weights for each respective input from an auxiliary digital beamformer 128.sub.i and output to a respective SSP adaptive FIR filter 130.sub.i the one or more adaptive weights.

(24) Referring now to FIG. 5, there is shown a process flow for generating the fixed weights be α.sub.1-α.sub.7 suitable for use by the adaptive FIR filter in FIG. 2 to create the SSP delays τ.sub.1, τ.sub.2, . . . τ.sub.n. This architecture produces an M-sample all-pass time-delay FIR with real coefficients and a group delay equal to (M−1)/2 I/Q sample periods plus τ, where τ need not be an integer multiple of the I/Q sample period (i.e. τ may be a SSP). A relative time delay of τ is created by filtering the I/Q data stream to be delayed through the time delay FIR and delaying all other relevant channels by (M−1)/2 I/Q sample periods (i.e. the bulk delay of (M−1)/2 I/Q samples).

(25) At step 502, an M-sample all-pass prototype FIR filter A is created where M is an odd integer. The prototype FIR filter is an all-pass filter, meaning that the I/Q data stream is not modified, but only delayed. The prototype FIR filter A includes (M−1)/2 real weights of zero, a real weight of one, and (M−1)/2 additional real weights of zero. Accordingly, the prototype FIR filter only functions to delay the I/Q data stream by (M−1)/2 samples (i.e. the group delay). At step 504, an M-point inverse discrete Fourier transform (IDFT) is taken of the M samples, thereby creating a complex output from step 504. At step 506, a linear phase is applied across the M samples. The slope of the linear phase is proportional to the desired or target sub-sample time delay τ (the linear phase slope is a function of f.sub.S and of the time delay τ, thereby creating a phase with slope −2πτ). At step 508, an M-point discrete Fourier transform (DFT) is taken to re-enter the time domain. At step 510, an M-sample amplitude taper is applied to remove discontinuities toward the ends of the sample streams (e.g. the first and last samples), thereby smoothing out the time-delay ripple. At the output of step 510, all of the samples are complex but have a common phase which is irrelevant to the design of the FIR filter. At step 512, the common phase is removed (e.g. by taking only the real components or parts), resulting in all of the samples (i.e. coefficients) being real. At step 514, the M samples are normalized (e.g. a common amplitude adjustment is applied to all of the samples), thereby ensuring that the noise level at the output is the same as the input noise level. The result of the process flow is a filter having a delay of (M−1)/(2f.sub.S)+τ (i.e. ((M−1)/2)T+τ). Thus, the process flow of FIG. 5 provides for design of both a bulk delay filter (i.e. the prototype FIR filter) and for the sub-sample-period relative delay filter (i.e. the final product of the process flow providing a delay equal to the bulk delay+the subsample period τ).

(26) Additional benefits of implementing an adaptive radar having SSP adaptive FIR filters include significant reduction in latency and increased numerical stability. Still further, fewer training samples are required because the number of training samples needed to achieve a given level of channel match is proportional to the number of adaptive weights. Finally, enhanced channel match is obtained because a small set of weight training samples is more likely to be statistically stationary than a larger set.

(27) In embodiments of the present invention, such as the adaptive radar system 100 of FIG. 1, N auxiliary channels may be chosen to meet a particular set of requirements. For example, multiple auxiliary channels may be advantageous when the adaptive radar system 100 may encounter multiple simultaneous barrage noise jammers.

(28) Thus, there is disclosed an embodiment of a method for adaptively matching the frequency response of multiple channels of a received radar signal, comprising: receiving a main RF signal on a main antenna; sampling the main RF signal at a given sample rate; delaying the sampled main RF signal by a given bulk delay equal to an integer number of sample periods; receiving at least one other RF signal on at least one other auxiliary antenna associated with the main antenna; sampling each of the at least one other RF signals at the given sample rate; filtering each of the other at least one sampled RF signals utilizing an adaptive finite impulse response (FIR) filter having at least one sub-sample-period time delay less than the sample period of the filtered signal; and combining the sampled main RF signal with each of the filtered other sampled RF signals. The sample period is the inverse of the sample rate. The sub-sample-period delay is on the order of hundreds of times smaller than the sample period delay. The sub-sample-period may be a non-integer multiple of the sample period. Each adaptive FIR filter may further have at least one sample-period delay. Each of the sub-sample-period time delays is generated by one or more fixed weight FIR filters.

(29) In another embodiment a system for adaptively matching the frequency response of multiple channels, comprises: a main channel configured to receive a main data stream and to output a main in-phase/quadrature-phase (I/Q) data stream at a given sample rate; one or more auxiliary channels, each auxiliary channel configured to receive one or more jamming signals to be removed from the main I/Q data stream and to output an auxiliary I/Q data stream at the given sample rate; an M-tap adaptive finite impulse response (FIR) filter for each auxiliary channel configured to receive the auxiliary I/Q data stream and to output an I/Q data stream; the FIR filter having a first set of taps separated by integer multiples of the I/O sample rate, and a second set of taps separated by sub-sample-period time delays that are less than the I/Q sample period, and wherein the M-tap I/Q sample rate FIR contains M fixed real weights; and an adder configured to sum the output data streams of the main channel and each of the auxiliary channels to cancel the one or more jamming signals from the main I/Q data stream. The matched I/O data stream has a passband matched to the main channel passband over the instantaneous bandwidth. The sub-sample-period delay is on the order of hundreds of times smaller than the sample period delay.

(30) Furthermore, each of the main and auxiliary channels includes a receiver configured to receive RF signals from the corresponding main and auxiliary antennas, amplify the received signals, and output intermediate frequency (IF) signals. Each of the main and auxiliary channels further includes a digital beamformer downstream of the corresponding receiver.

(31) In a different embodiment, an adaptive finite impulse response (FIR) filter for filtering an I/Q data stream having a sample period, comprises: a plurality of sample-period tap delays configured to delay the I/Q data stream by integer multiples of the sample period of the I/Q data stream; a plurality of sub-sample-period tap delays configured to delay the I/Q data stream by non-integer multiples of the sample period of the I/Q data stream, the sub-sample period tap delays being at least an order of magnitude less than the sample period tap delays; a set of adaptive weights configured to weight samples of the delayed I/Q data stream; and an adder configured to combine the weighted samples of the delayed I/Q data streams to generate a filtered I/Q data stream. The weights of the FIR filter are fixed real weights. The sub-sample-period is a non-integer multiple of the sample period.

(32) In another embodiment, a time delay circuit delays a received I/Q data stream sampled at a sampling rate, comprises a FIR filter comprising an odd integer number of sample period tap delays configured to delay the I/Q data stream, wherein the sample period is the reciprocal of the sampling rate; an odd integer number of fixed real weights configured to weight samples of the delayed I/Q data stream; and an adder configured to combine the weighted samples of the delayed I/O data stream. The group delay of the FIR filter is ((M−1)/2)T+τ) where T is the sample period and t is the sub-sample period delay.

(33) In a further embodiment, a method for determining a set of fixed real weights for a time delay circuit in a computer system including an M-sample all-pass FIR filter, wherein the M samples include M real weights of zero and a center real weight of one, comprises: performing via a computer processor an M-point inverse discrete Fourier transform of the M samples; applying a liner phase slope corresponding to a time delay; performing via a computer processor an M-point discrete Fourier transform of the M samples; applying an M-sample amplitude taper to the M samples; and removing a common phase offset from the M samples.

(34) While the embodiments of the present invention discussed herein generally refer to adaptive radar systems, the present invention can be applied generally to other applications of adaptive processing requiring channel matching. By way of non-limiting example, the present invention could beneficially be implemented in wireless communications systems, including systems using code division multiple access (CDMA) technology and frequency hopping.

(35) It will be apparent to those skilled in the art that modifications and variations may be made in the method and system of the present invention without departing from the spirit or scope of the invention. It is intended that the present invention cover the modification and variations of this invention provided they come within the scope of the appended claims and their equivalents.