Robust STAP algorithms using prior knowledge for airborne radar applications

Abstract Space–time adaptive processing (STAP) schemes have shown promise for airborne radar applications. However, the majority of schemes develop an estimate of the covariance matrix for the test cell by averaging over surrounding range cells, called reference data. This method is only guaranteed to ensure good performance when a large set of homogeneous reference data is available with the same statistics as the test cell. In this paper, a new methodology is proposed for obtaining a covariance matrix that uses a priori knowledge. The approach is useful for detecting weak signals in cases with discretes in some range cells which do not appear in other range cells. The focus is on the use of a simple model for ground clutter that incorporates our prior knowledge on the structure of the ground clutter. The new methodology can be applied to most existing STAP schemes. This is illustrated by applying the methodology to three specific existing schemes. The modified schemes are generally shown to outperform the existing schemes in non-stationary measured data cases.