Airborne radar STAP using long-memory clutter models

Traditional analysis of airborne adaptive radars usually assumes the ideal condition of statistically independent and identically distributed (IID) secondary data. To the contrary, measured clutter data are spatially correlated and exhibit long-memory characteristics. In addition, clutter can appear heterogeneous and so the secondary data are no longer IID. The aim of this paper is to define an efficient simulator of long-range dependence (LRD) homogeneous or non-homogeneous clutter space-time steering vectors. To this end, multi-channel airborne radar measurement (MCARM) data have been analyzed to estimate the power spectral density (PSD) of space-time snapshots in real clutter environments. In this paper we will refer to the considered long-memory clutter models as to the LRDSTAP clutter models. LRDSTAP models extend the simplified general clutter model (GCM) by including the LRD characteristics of clutter PSDs. Simulated data from LRDSTAP are used to evaluate the impact of long-memory data on the convergence measure of effectiveness (MOE) of the sample matrix inversion (SMI) STAP processor. The presented method demonstrates reliable results when applied to MCARM data files including either spatially homogeneous or nonhomogeneous clutter.