Evaluation of Knowledge-Aided STAP Using Experimental Data

Recent advances in knowledge-aided space-time adaptive processing (KA-STAP) have resulted in significant performance improvements for ground moving target indication (GMTI) radar systems. In particular, the use of prior knowledge including terrain, clutter discretes, and previously detected targets has been shown to be effective for mitigating the poor performance often encountered when operating in heterogeneous clutter environments. This paper provides an evaluation of KA-STAP techniques based on extensive processing of experimental data. Two major performance issues are addressed: high false alarm rates due to under-nulled clutter discretes and target cancellation due to corruption of the STAP training data by other targets in the scene. Each of these problems is demonstrated using experimental multi-channel X-band radar data. Methods for using prior knowledge to improve performance are presented and processing results using the experimental data are provided that show how KA-STAP can lead to significantly improved detection performance relative to conventional STAP processing.