Improving STAP Performance in Bistatic Space-Based Radar Systems using an Efficient Expectation-Maximization Technique

In this paper, we describe a space-time adaptive processing (STAP) approach for bistatic space-based radar (SBR) systems. A candidate SBR system employing a transmitter at medium earth orbit (MEO) and an airborne receiver is defined. A STAP configuration for performing wide area ground moving target indication (GMTI) is also specified. STAP performance is analyzed using simulated data in two different bistatic SBR geometries. Standard STAP performance degradation is shown to be a function of the relative severity of clutter non-stationarity in the two scenarios. To provide enhanced estimation of the clutter statistics, we apply an approach based on an efficient form of an expectation-maximization (EM) algorithm for estimating covariance matrices in non-stationary interference. The STAP performance of the EM technique is compared to that of standard STAP processing and higher order Doppler warping. The EM approach is seen to produce a usable Doppler space fraction (USDF) that is significantly improved over the other techniques. It has several powerful features, including the ability to align non-linear angle-Doppler contours, adaptivity to interference without applying a fixed range-dependent transformation, and the ability to simultaneously compensate for different range dependencies in higher range and Doppler ambiguities. The EM STAP approach also naturally accommodates knowledge-aided STAP techniques that improve understanding of clutter reflectivity variations.
See also ADM002017. Presented at the 2005 IEEE International Radar Conference Record held in Arlington, Virginia on May 9-12, 2005. Published in the Proceedings of the 2005 IEEE International Radar Conference Record. The original document contains color images.