Adaptive processing in a nonstationary spaceborne environment

This paper describes the impact of clutter nonstationarity on space-time adaptive processing (STAP) performance in a sparse aperture, spaceborne, moving target indication radar scenario and then presents mitigating techniques. Clutter nonstationarity introduces large estimation losses when applying traditional STAP techniques, since STAP requires knowledge of clutter statistics for effective cancellation. Results have previously been shown indicating significant improvements achieved by using reduced-dimension STAP techniques minimizing nonstationarity via localized training; in one instance, the estimation loss was reduced from an average of about 14 dB using joint domain optimum (JDO) STAP to about 6.5 dB using the extended factored algorithm (EFA). This work introduces a steering vector correction method and also applies the extended sample matrix inversion (ESMI) technique to further mitigate clutter nonstationarity. Both of these methods produce good results, with steering vector correction outperforming ESMI in the scenarios tested. Using the combination of EFA and steering vector correction, the average estimation loss was reduced from about 14 dB to 4 dB for the scenario mentioned above, similar to the 3 dE3 loss expected in stationary clutter environments with the same amount of data used to compute the covariance estimate. TABLE OF CONTENTS ........................................................................