STANDALONE GNSS ANTI-JAM NULLER-BEAMFORMER COMBINING SFAP AND STAP

Abstract

A method and apparatus for providing a standalone anti-jamming (AJ) nuller-beamformer. Signals from an antenna array include a sum of Global Navigation Satellite System (GNSS) signals and jamming signals from a plurality of spatial sources. A front end is configured to amplify, filter, down-convert, and sample the input signals which are then filtered, downconverted, and decimated prior to frequency-domain and time-domain partitioning. Weights are computed and applied for spatial nulling of jamming signals in each frequency bin for the partitioned signals. Frequency and time-domain reconstruction generates a reconstructed signal with suppressed jamming.

Claims

1. A standalone anti-jamming (AJ) nuller-beamformer, comprising: an antenna array comprising a plurality of antennas configured to receive an input signal that includes a sum of Global Navigation Satellite System (GNSS) signals and jamming signals from a plurality of spatial sources; an analog and mixed signal front end to amplify, filter, down-convert, and sample the input signals of the antennas of the array; a digital front end to filter, down-convert, and decimate the input signals of the antennas of the array; a frequency and time-domain partitioning module configured to perform frequency-domain and time- domain partitioning of the input signals of the antennas of the array; a weight calculator module to calculate weights for spatial nulling of jamming signals in each frequency bin for the partitioned signals; a weight application module to apply the weights to the partitioned signals from the antennas and time taps in each frequency bin; a frequency and time-domain reconstruction module to sum the weighted signals from the antennas and the time taps in each frequency bin and perform frequency-domain reconstruction to generate a reconstructed signal with suppressed jamming; and a sampling rate conversion module to convert a sampling rate of the reconstructed signal to a sampling rate for the digital input(s) of one or more GNSS receivers.

2. The standalone AJ nuller-beamformer of claim 1, further including a delay buffer block, wherein the signals from the frequency and time-domain partitioning module are delayed in the delay buffer block before being sent to the weight application module.

3. The standalone AJ nuller-beamformer of claim 1, further including an adaptive digital filtering module, wherein the signals from the digital front end, prior to the frequency and time-domain partitioning, are processed in the adaptive digital filtering module for suppression of jamming that has low spectral overlap with the GNSS signal.

4. The standalone AJ nuller-beamformer of claim 1, wherein the weight calculator module is configured to compute a separate set of weights for each GNSS satellite whose signals are being processed, with the weights in each set of weights configured to maximize the signal-to-jamming ratio for the corresponding satellite, and wherein the weight application and frequency and time-domain reconstruction modules perform the weight application and signal reconstruction separately for each of the GNSS satellites.

5. The standalone AJ nuller-beamformer of claim 4, wherein the sampling rate conversion module is configured to perform sampling rate conversion of the reconstructed signals for different satellites to generate a plurality of digital outputs.

6. The standalone AJ nuller-beamformer of claim 4, wherein one of the digital output signals with suppressed jamming is converted to an analog RF signal and sent to an analog RF output for compatibility with GNSS receivers.

7. A method, comprising: for a standalone anti-jamming (AJ) nuller-beamformer, comprising: receiving, by an antenna array comprising a plurality of antennas, an input signal that includes a sum of Global Navigation Satellite System (GNSS) signals and jamming signals from a plurality of spatial sources; amplifying, filtering, down-converting, and sampling the input signals of the antennas of the array by an analog and mixed signal front end; filtering, down-converting, and decimating the input signals of the antennas of the array by a digital front end; frequency-domain and time- domain partitioning of the input signals of the antennas of the array by a frequency and time-domain partitioning module; calculating weights for spatial nulling of jamming signals in each frequency bin for the partitioned signals by a weight calculator module; applying the weights to the partitioned signals from the antennas and time taps in each frequency bin by a weight application module; summing the weighted signals from the antennas and the time taps in each frequency bin and performing frequency-domain reconstruction to generate a reconstructed signal with suppressed jamming by a frequency and time-domain reconstruction module; and converting a sampling rate of the reconstructed signal to a sampling rate for the digital input(s) of one or more GNSS receivers by a sampling rate conversion module.

8. The method of claim 7, further including employing a delay buffer block, wherein the signals from the frequency and time-domain partitioning module are delayed in the delay buffer block before being sent to the weight application module.

9. The method of claim 7, further including employing an adaptive digital filtering module, wherein the signals from the digital front end, prior to the frequency and time-domain partitioning, are processed in the adaptive digital filtering module for suppression of jamming that has low spectral overlap with the GNSS signal.

10. The method of claim 7, wherein the weight calculator module computes a separate set of weights for each GNSS satellite whose signals are being processed, with the weights in each set of weights maximizing the signal-to-jamming ratio for the corresponding satellite, and wherein the weight application and frequency and time-domain reconstruction modules perform the weight application and signal reconstruction separately for each of the GNSS satellites.

11. The method of claim 10, wherein the sampling rate conversion module performs sampling rate conversion of the reconstructed signals for different satellites to generate a plurality of digital outputs.

12. The method of claim 10, wherein one of the digital output signals with suppressed jamming is converted to an analog RF signal and sent to an analog RF output for compatibility with GNSS receivers.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The foregoing features of this disclosure may be more fully understood from the following description of the drawings in which:

[0012] FIG. 1 is a high level block diagram of example signal flow in a GNSS unit combining array-based AJ nuller-beamformer with a subsequent GNSS receiver.

[0013] FIG. 2 is a block diagram illustrating example adaptive array processing.

[0014] FIG. 3 is a block diagram illustrating example SFAP.

[0015] FIG. 4 is a block diagram illustrating example STAP.

[0016] FIG. 5 is a block diagram illustrating an example implementation of a combination of SFAP and STAP known as SFAP-T.

[0017] FIG. 6 is a block diagram illustrating another example implementation of a combination of SFAP and STAP known as SFAP-T.

[0018] FIG. 7 is a block diagram of a nuller-beamformer illustrating the example embodiments of the disclosure.

DETAILED DESCRIPTION

[0019] FIG. 1 shows example signal flow in an AJ GNSS unit combining array-based AJ nuller-beamformer with a subsequent GNSS receiver in accordance with example embodiments of the disclosure. An input signal, which is a sum of desired signals, jamming, and receiver noise, is received by M antenna elements of the antenna array 111. The M-channel analog signal 113 is filtered, down-converted, and digitized in the analog and mixed-signal front end 115. The M-channel digital signal 117 can be further down-converted to baseband, filtered, and decimated in the digital front end 119. The M-channel baseband complex-valued digital signal 121 undergoes adaptive array AJ processing 123 that performs weighted combining of the antenna channels. This combining is optimized for spatial nulling of the jamming while maintaining an appropriate level of desired signals. The computation of proper weights for the combining is also performed as part of AJ processing 123 using the samples of signal 121 and optionally some external information. The output signal 125 of the nuller-beamformer with increased signal-to-jamming ratio is sent to the GNSS receiver 127. The number of channels of signal 125 can vary from one for a nuller to the number of beams for a nuller-beamformer. The standalone GNSS nuller-beamformer described in this disclosure can contain, in addition to block 125, also blocks 119, 115, and 111.

[0020] FIG. 2 shows an example embodiment of adaptive array processing. The M-channel digitized, down-converted, and decimated input signal 222, which can be an equivalent of signal 121 in FIG. 1, is sent to the input of weight calculator 228 for periodically calculating M optimal or near-optimal weights for the linear weighted combining of the antenna channels of signal 222 with the purpose of nulling the jamming, possibly with beamforming to the GNSS signal sources (typically satellites). Example algorithms for weight calculation include, but are not limited to, minimum variance (MV), minimum variance distortionless response (MVDR) algorithms, and the like. The output weights 229 are applied to incoming samples of the corresponding channels of signal 222 in a weight application module 230. The weighted signal 232 enters the signal combining module 234 where the weighted components of signal 232 are summed, creating a combined signal 236 in which the jamming power is reduced compared to the desired signal power.

[0021] In the diagram shown in FIG. 2, each output sample s is a weighted sum of the corresponding samples in all antenna channels, where s can be computed as set forth below as:

[00001]$s={\text{w}}^H\text{s}$

where s = [s.sub.1 ... s.sub.M].sup.T is the M×1 input sample vector and

[00002]${\text{w}}^H=\left[\begin{array}{ccc}{w}_1^{\ast }& .Math.& w_M^{\ast }\end{array}\right]$

is the 1×M weight vector.

[0022] An optimal weight vector, as set forth below, can be calculated according to the aforementioned MVDR algorithm minimizes the output jamming power while satisfying the linear constraint

[00003]${\text{w}}^H\text{v}=1$

where v is a M×1 steering vector towards a GNSS satellite, with its elements representing the gains of the corresponding array elements in the satellite’s direction. This constraint ensures the unit gain in the direction specified by v. An MVDR weight vector solution is

[00004]${\text{w}}^H={\text{v}}^H{\text{R}}^{-1}/\left({\text{v}}^H{\text{R}}^{-1}\text{v}\right)$

where R is the M×M spatial covariance matrix of input signal samples in the antenna channels.

[0023] Different beams require different steering vectors v (and therefore different resulting weight vectors w), whereas the covariance matrix R can be the same.

[0024] Other weight optimization algorithms besides MVDR may also be used.

[0025] FIG. 3 provides an example embodiment of SFAP. The incoming signal 222 enters the frequency-domain partitioning module 324 where it is separated into N frequency subbands, also known as frequency bins. In embodiments, module 324 has N×M output channels to accommodate components of the output signal 326. Each of N frequency components of signal 326 independently undergoes the adaptive processing illustrated in FIG. 2, with weight calculator 328 computing an M-element weight vector according to equation (3) for each of N frequency bins during each weight update period. A signal combining/reconstruction module 334 performs summation of the weighted antenna channels in each frequency bin and frequency-domain reconstruction of the whole signal from N frequency components, creating the output signal 336.

[0026] In embodiments, SFAP increases the AJ performance for all types of frequency-dependent mismatch among the array antennas and their analog channels. In particular, it is effective against frequency-domain gain and phase ripple mismatch.

[0027] FIG. 4 provides as example embodiment of STAP. In a partitioning module 424, the incoming signal 222 channel passes through a tapped delay line with K taps, which will be referred to as time-domain partitioning. In example embodiments, the partitioning output has K×M channels to accommodate components of output signal 426, which undergoes adaptive processing that may be similar to that illustrated in FIG. 2 but with higher dimensionality. A weight calculator module 428 computes KM weights 429 during each weight update period. The algorithms for STAP weight calculation include, but are not limited to, linear constrained minimum variance (LCMV) algorithm and the like. The combining/reconstruction module 434 sums the weighted taps in the antenna channels, and performs signal combining and time-domain reconstruction.

[0028] Signal combining according to STAP in FIG. 4 is still described by equation (1), but in this case the input sample vector s = [s.sub.11 ... s.sub.M1 ... s.sub.1K ... s.sub.MK].sup.T has dimensions KM×1 and weight vector

[00005]${\text{w}}^H=\left[\begin{array}{ccccccc}{w}_{11}^{\ast }& .Math.& w_{M1}^{\ast }& .Math.& w_{1K}^{\ast }& .Math.& w_{MK}^{\ast }\end{array}\right]$

has dimensions 1×KM.

[0029] An example optimal weight vector calculated according to the aforementioned LCMV algorithm minimizes the output jamming power while satisfying the linear constraint

[00006]${\text{w}}^H\text{C}={\text{g}}^H$

[0030] This constraint can maintain a desired impulse response g.sup.H in the direction of a GNSS satellite specified by steering vector v. In this case, g.sup.H is 1×K vector and C is the KM×K matrix calculated as C = I .Math. v where v the M×1 steering vector, I is the K×K identity matrix, and .Math. designates Kroneker product.

[0031] An example LCMV weight vector solution can be computed as:

[00007]${\text{w}}^H={\text{g}}^H\left({\text{C}}^H{\text{R}}^{-1}\text{C}\right){\text{C}}^H{\text{R}}^{-1}$

where R is the KM×KM space-time covariance matrix of input signal samples in all taps of all antenna channels.

[0032] Different beams require different steering vectors v (and therefore different matrices C and resulting weight vectors w), whereas the covariance matrix R can be the same.

[0033] LCMV algorithm may be modified to include other types of hard and soft constraints. Other weight optimization algorithms besides LCMV may also be used.

[0034] In embodiments, STAP increases the AJ performance for all types of frequency-dependent mismatch among the array antennas and their analog channels, being particularly effective against delay mismatch (which manifests as a frequency-domain phase slope), even if K is small.

[0035] FIG. 5 provides an example illustration of SFAP-T processing combining SFAP and STAP as illustrated in FIGS. 3 and 4, respectively. The input signal 222 undergoes both frequency-domain and time-domain partitioning in a partitioning module 524, and the subsequent processing is essentially STAP (illustrated in FIG. 4) in each frequency bin. Weight calculator 528 computes N weight vectors (one per frequency bin), each consisting of KM elements. Weight application module 530 applies the weights 529 to the partitioned signals, as described above. A combining/reconstruction module 534 performs signal combining of the weighted signals together with time-domain and frequency-domain reconstruction to generate the output signal 536.

[0036] In FIG. 5, frequency-domain partitioning in block 524 precedes time-domain partitioning. For practical purposes, both these operations can be considered linear and time-invariant. Therefore, they may be performed in any order (same is true about reconstruction in block 534). The choice of the order depends on design considerations. For example, frequency-domain partitioning is often combined with signal decimation in each frequency bin. While this decimation reduces the throughput requirements, it can make time-domain partitioning more granular. If fine precision of time-domain partitioning is desired, it is often better to perform it before frequency-domain partitioning. Such an implementation of SFAP-T is illustrated in FIG. 6. The reconstruction order is also selected based on practical considerations and does not necessarily depend on the partitioning order.

[0037] In the system 600 of FIG. 6, as can be seen in the partitioning module 624, time-domain partitioning is performed prior to frequency-domain partitioning. A weight calculator module 628 generates weights 629 and a weight application module 630 applies the weights to the partitioned signals 626. A combining/reconstruction module 634 performs signal combining of the weighted signals together with time-domain and frequency-domain reconstruction to generate the output signal 636.

[0038] In embodiments, SFAP-T is very effective against both delay and ripple mismatch (as well as any other type of frequency-dependent mismatch among the array antennas and their analog channels) due to combining the advantages of SFAP and STAP.

[0039] FIG. 7 shows an example of a standalone GNSS AJ nuller-beamformer 700 in accordance with example embodiments of the disclosure. In embodiments, the nuller-beamformer is self-contained and independent of, and separable from, the GNSS receiver. In embodiments, SFAP and STAP processing are combined to provide SFAP-T processing. In embodiments, SFAP processing is augmented with STAP processing in each frequency bin.

[0040] Antenna array 742 is coupled to an analog and mixed signal front end 746, whose outputs are processed in the digital front end 750. In some embodiments, the output signals of the digital front end 750 are filtered in an optional adaptive digital filter 754, which suppresses jamming with low spectral overlap with GNSS signals. Subsequently, the signals enter a frequency- and time-domain partitioning module 758, which performs frequency-domain and time-domain partitioning of the signals, as illustrated by block 524 of FIG. 5 or block 624 of FIG. 6, for example. The partitioned signals 760 are used in the weight calculator 762 to compute the optimal weights for jamming nulling in each frequency bin. In some embodiments, these weights are also optimized for beamforming. In some embodiments, separate sets of such weights are computed for different GNSS satellites to provide optimal beamforming for each satellite. The partitioned signals 760 are also sent to the weight application module 770, in some embodiments through a delay buffer 766. In some embodiments, separate weight sets are applied to the partitioned signals for processing in different GNSS satellite channels. The subsequent signal combining, frequency and time-domain reconstruction module 774 reconstructs the whole signal with suppressed jamming. In some embodiments, separate reconstruction is performed for signals in different GNSS satellite channels. The reconstructed signal(s) undergo sampling rate conversion 778 to make sure that the output sampling rate of the nuller-beamformer matches the sampling rate of the digital input(s) of GNSS receivers. The resulting signal(s) are sent to the digital output(s) 782 of the nuller-beamformer. Depending on the embodiment, the number of digital outputs can vary from one to the number of GNSS satellite channels. In some embodiments, one of the reconstructed signals can be also sent to the digital back end 786 and subsequently to the analog and mixed signal back end 790 for upsampling, digital-to-analog conversion, and conversion to RF. The RF output 794 can be connected to the antenna inputs of GNSS receivers, even if these receivers do not have digital inputs compatible with the nuller-beamformer’s digital output(s) 782.

[0041] In some embodiments, frequency-domain partitioning in block 758 and reconstruction in block 774 can use techniques based on fast Fourier transform (FFT) and inverse fast Fourier transform (IFFT), respectively. Such techniques can include weighted overlap & add (WOLA) FFT banks and polyphase filter banks.

[0042] In embodiments, LCMV beamforming algorithm described above can be used in the weight calculator module 762 for each frequency bin. The constraints can be different in each frequency bin.

[0043] A simplified LCMV beamforming algorithm can apply the constraints similar to (4) only to weights for selected temporal taps. Benign behavior of the weights for other temporal taps can be ensured by other means, e.g., soft constraints applied to the space-time covariance matrix. Other suitable types of constraints, as well as other algorithms besides LCMV may be used.

[0044] The embodiments contained herein should not be limited to disclosed embodiments but rather should be limited only by the spirit and scope of the appended claims. All publications and references cited herein are expressly incorporated herein by reference in their entirety.

[0045] Elements of different embodiments described herein may be combined to form other embodiments not specifically set forth above. Various elements, which are described in the context of a single embodiment, may also be provided separately or in any suitable subcombination. Other embodiments not specifically described herein are also within the scope of the following claims.