A Self Referencing Technique for the RC-pLMS Adaptive Beamformer and Its Hardware Implementation

Abstract

In this paper, we propose a self referencing scheme for the reduced complexity parallel least mean square (RC-pLMS) adaptive beamforming algorithm as means of robustness against possible interruptions in the reference signal and its hardware implementation. The RC-pLMS is a single stage, non-blind, least mean square (LMS) algorithm with modified input vectors formed as a linear combination of the current and the previous input sample. In this context, its convergence and its stability are critically dependent on the availability of its reference signal and are known to severally degrade when discontinued. Thus, for robustness against the pre-mentioned and with respect to the RC-pLMS accelerated convergence and low residual error profile, we propose the use of it’s filtered output, as an alternative learning sequence, whenever the original reference signal is discontinued, i.e. self-referencing. The proposed self referencing approach is evaluated in infinite and finite precision modes on software and on hardware, i.e. Field Programmable Gate Array (FPGA), respectively. Hardware and software simulation validates the RC-pLMS robustness against different reference signal obstruction scenarios, through the use of the proposed self-referencing approach, while maintaining an accelerated convergence behavior, a low complexity architecture and a high precision beam pointing accuracy.