Median cascaded canceller using reiterative processing

Abstract

A novel, robust adaptive processor is introduced, based on reiterative application of the Median Cascaded Canceller (MCC). The MCC, though a highly robust adaptive processor, has a convergence rate that generally is dependent on the effective rank of the interference-plus-noise covariance matrix.The Reiterative Median Cascaded Canceller (RMCC) introduced here exhibits the highly desirable combination of 1) convergence-robustness to outlierdtargets in adaptive weight training data, like the MCC, and 2) fast convergence performance independent of the interference-plus-noise covariance matrix and at a rate commensurate with the Sample Matrix Inversion (SMI) algorithm, unlike the MCC. Both simulated data as well as measured airborne radar data from the MCARM Space-Time Adaptive Processing (STAP) database are used to show performance enhancements. It is concluded that the RMCC adaptive processor is a highly robust replacement for the SMI adaptive processor and all its equivalent implementations.

Robust Auto-Regressive Spectrum using a Reiterative Median Cascaded Canceller

“…The RMCC algorithm [1][2][3] is an adaptive signal processing algorithm that may be used to form a robust estimate of an N t × 1 equivalent linear prediction weight vector having the general form w = [w 1 w 2 … w Nt ] T , from which a robust AR spectrum can be generated [4]. The quantity N t denotes the number of degrees of freedom (DOF) or the number of input channels used, and superscript T denotes transpose.…”

“…Such reiterative processing is used to improve the convergence of the resulting robust linear prediction weight vector. It was shown in [2,3] that reiterative processing is required for achieving good performance in several cases.…”

“…The N t × 1 steering vector for linear prediction is [1 0 0 … 0] T [1][2][3][4]. Often, array processing input channels are not in the correct form to use linear prediction techniques like the GSCC because their steering vector does not match the required form:…”

“…2 IEEEAC paper #1351, Version 3, Updated Nov. 21,2006 which bear no correlation with true frequencies used in the formation of the input data. Choosing model orders greater than the number of true frequency sources often leads to the presence of spurious peaks.…”

“…Subsequent AR spectra using contaminated data may be badly distorted or have missing / additional spectral peaks in many real-world applications. This paper presents a novel method to form a robust AR spectrum using the Reiterative Median Cascaded Canceller (RMCC) algorithm [2,3]. In Section 2 we discuss an RMCC processor.…”