Robust cascaded canceller using projection statistics for adaptive radar

Abstract

Adaptive radar requires independent and identically distributed (i.i.d.) training data, or snapshots, in order to obtain fast SINR convergence performance in the presence of correlated interference such as jamming and/or clutter returns. Targets, clutter discretes, and impulsive jamming are examples of non i.i.d., real-world data components that corrupt interference training data. Such data are considered to be statistical outliers. Recent outlier detection work for space time adaptive processing (STAP) training data selection has involved use of the generalized inner product (GIP) test statistic. In this paper, we use a prewhitening method followed by a robust projection statistics (PS) algorithm for 2D outlier removal prior to each building block in a reiterative adaptive cascaded canceller. SINR performance is shown to be superior using 2D PS compared to 2D GIP to excise multiple outliers