Michael Zatman

A computationally efficient two-step implementation of the GLRT

In this paper, the performance of a new two-step adaptive detection algorithm is analyzed. The two-step GLRT consists of an initial adaptive matched filter (AMF) test followed by a generalized likelihood ratio test (GLRT). Analytical expressions are provided for the probability of false alarm (P/sub FA/) and the probability of detection (P/sub D/) in unknown complex Gaussian interference. The analysis shows that the two-step GLRT significantly reduces the computational load over the GLRT while maintaining detection and sidelobe rejection performance commensurate with the GLRT. The two-step GLRT detection algorithm is also compared with another two-step detection algorithm: the adaptive sidelobe blanker (ASB). Both the two-step GLRT and the ASB are characterized in terms of the mainbeam detection performance and the rejection of sidelobe targets. We demonstrate that for a given P/sub FA/, the two-step GLRT has a broad range of threshold pairs (one threshold for the AMF test and one for the GLRT) that provide performance identical to the GLRT. This is in contrast with the ASB, where the threshold pairs that maximize the P/sub D/ are a function of the targets signal-to-interference-plus-noise ratio (SINR). Hence, for a fixed pair of thresholds, the two-step GLRT can provide slightly better mainbeam detection performance than the ASB in the transition region from low to high detection probabilities.

Properties of Hung-Turner projections and their relationship to the eigencanceller

Hung-Turner (1983) projections have been proposed as a computationally efficient method of computing adaptive array weights. An exact relationship between Hung-Turner projections and the eigencanceller is noted, and used to analyze their performance. A simple and intuitive derivation is given for the expected SINR loss of Hung-Turner projections. This leads to a formula for the optimal number of snapshots (training samples). An expression for the expected sidelobe levels of weights computed using Hung-Turner projections is also derived.

Underwater acoustic MIMO channel capacity

The underwater acoustic channel is rich in multipath scattering, making it a promising environment for the application of MIMO communications. However, for typical underwater communications channels, there are difficulties associated with the large fractional bandwidths, significant Doppler dispersion and latencies measured in seconds. We use simulated and experimental data to estimate the channel capacity for typical shallow water MIMO channels, and motivate particular beamforming algorithms which cope with the difficult acoustic environment.

Forward-backward averaging in the presence of array manifold errors

We investigate the use of forward-backward (f/b) averaging for estimating the covariance matrix used for adaptive beamforming and space-time adaptive processing (STAP). We demonstrate that the estimation loss is reduced by the use of f/b averaging and, for some STAP cases, f/b averaging can even quadruple the available sample support. We also show that unknown array manifold errors have little effect on the effectiveness of f/b averaging. The gain from f/b averaging is demonstrated on data from the mountaintop database.

Radar resource management for UESA

This paper describes how a circular UHF electronically scanned array like that being considered by the US Navy could be used to simultaneously perform both airborne early warning (AEW) searches and high update rate tracking of a few targets. Taxonomies of the various search and track modes are reported, and an example radar resource configuration is described in detail.

Degree of freedom architectures for large adaptive arrays

Large phased array radars possess thousands of elements, but practically it is only possible to process a few tens of adaptive degrees of freedom (DOFs). However, with the use of element-level digitization and digital beamforming the next generation of phased array radars will have great flexibility in their choice of DOF architecture. This paper shows that to achieve the best performance the radars DOF architecture must adapt to both the radars operating mode (e.g. volume search, horizon search, track and wideband discrimination) and the interference environment (e.g. direct path jamming, multi-path jamming and clutter). The best performing DOF architectures for different combinations of radar mode and interference environment are described.

Forwards-backwards averaging for adaptive beamforming and STAP

We investigate the use of forwards backwards (f/b) averaging for estimating the interference-only covariance matrix used for adaptive beamforming and space-time adaptive processing (STAP). We demonstrate that the estimation loss is reduced by the use of the extra samples available through f/b averaging, and in STAP a new type of f/b averaging can quadruple the number of samples available. Importantly, we present a new result which shows that unknown array manifold errors have little effect on the performance of f/b averaging. The gain from f/b averaging is demonstrated on data from the Mountaintop database.

ABF limitations when using either Toeplitz covariance matrix estimators or the parametric vector AR technique

Two methods which have been proposed for estimating adaptive beamformer weights from a small number of samples are the parametric vector AR technique and the use of Toeplitz covariance matrix estimation. In this paper the relationship between these two techniques is derived. This relationship is used to show that both methods have the same performance limitations when used with antenna arrays which are not precisely uniform linear (or uniform rectangular for two dimensional arrays). In the presence of non-ideal arrays it is shown that both methods degrade to performance levels consistent with deterministic (i.e. non-adaptive) nulling.

On the performance of the parametric filter adaptive beamformer

The parametric vector auto regressive technique has been proposed for adaptive beamforming and STAP. In this paper the expected signal to interference plus noise ratio performance of two variants of the algorithm, with and without forwards-backwards averaging, is derived.

Subspace domain forwards-backwards averaging

In this paper a procedure which filters out roughly half of the array manifold errors for approximately centro-symmetric arrays is described. The procedure-subspace domain forwards-backwards (f/b) averaging-improves the performance of subspace based direction finding (DF) algorithms such as MUSIC and ESPRIT. Experimental data from the Mountaintop radar system are used to confirm the theoretical results.