A robust loaded reiterative median cascaded canceller

Abstract

A robust, fast-converging, reduced-rank adaptive processor is introduced, based on diagonally loading the reiterative median cascaded canceller (RMCC). The new loaded reiterative median cascaded canceller (LRMCC) exhibits the highly desirable combination of: (1) convergence-robustness to outliers/targets/nonstationary data in adaptive weight training data, like the RMCC; (2) convergence performance that is approximately independent of the interference-plus-noise covariance matrix, like the RMCC; and (3) fast convergence at a rate commensurate with reduced-rank algorithms, unlike the RMCC. Measured airborne radar data from the MCARM space-time adaptive processing (STAP) database is used to show performance enhancements. It is concluded that the LRMCC is a practical and highly robust replacement for existing reduced-rank adaptive processors, exhibiting superior performance in nonideal measured data environments.