Adaptive Sidelobe
Threshold Setting and
Performance in
Inhomogeneous Clutter



Daniel E. Kreithen, Carl F. Pearson, and Christ D. Richmond
MIT Lincoln Laboratory
244 Wood Street, Room J-118D
Lexington, MA 02173-9108
tel: (781) 981-0144

Abstract The homogeneous clutter assumption is often made when analyzing and designing post-adaptive nulling detection algorithms. In real-world scenarios, however, this assumption is invalid, and often leads to markedly elevated false alarm rates and generally poor target detection performance. A number of recent theoretical papers address the problem of target detection in post-adaptively nulled nonhomogeneous clutter [1,2], and demonstrate significant progress in this area. The present paper develops and analyzes criteria for setting thresholds in the two-dimensional adaptive sidelobe blanker (ASB) detection algorithm, which consists of a first stage adaptive matched filter (AMF) detector [3] followed by an adaptive coherence estimator (ACE) detector [4].The ASB detector is theoretically analyzed in [2], where it is shown that there are infinite combinations of thresholds for the AMF and ACE which achieve the same false alarm rate. At one extreme it is possible to run the ASB as a strict AMF detector, which provides near optimum detection performance in homogeneous clutter with very little suppression of clutter discretes and sidelobe targets. At the other extreme, we can run the ASB as an ACE detector, which provides very good clutter discrete and sidelobe target suppression, but significantly degraded target detection performance. Clearly, it would be helpful to devise criteria that allow a sensible choice of thresholds somewhere between these two extremes. We propose two criteria for threshold setting for the ASB: (i) the "allowable loss" criterion and (ii) the "minimal loss" criterion. The allowable loss criterion, as the terminology implies, allows a given amount of target loss in exchange for clutter-discrete-induced false alarm and sidelobe target mitigation. The minimal loss criterion imposes a tolerable upper bound on false alarm rate given a minimum amplitude clutter discrete to arrive at a threshold combination. This paper formulates both criteria, and examines the behavior of the ASB in the presence of clutter discretes using an idealized model and simulation.

1] D. E. Kreithen and A. O. Steinhardt, "Target Detection in Post-STAP Undernulled Clutter," 29th Annual Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, CA, October 29­November 1, 1995.

[2] C. D. Richmond, "Statisical Performance Analysis of the Adaptive Sidelobe Blanker Detection Algorithm," to appear at 31st Annual Asilomar Conference on Signals, Systems, and Computers, Pacific Grove CA, November 2­5, 1997.

[3] F. C. Robey, D. R. Furhman, E. J. Kelly, R. A. Nitzberg, "A CFAR Adaptive Matched Filter Detector," IEEE Transactions on Aerospace and Electronic Systems, vol. 28, no. 1, 208­216 (1992).

[4] E. Conte, M. Longo, M. Lops, G. Ricci, "Asymptotically Optimum Radar Detection in Compound-Gaussian Clutter," IEEE Transactions on AES, vol. 31, no. 2, 617­625 (1995).



Direct comments and questions to:

MIT Lincoln Laboratory. All rights reserved.