The location-estimating, null-steering (LENS) algorithm for adaptive microphone-array processing

Abstract

This document develops and evaluates the Location-Estimating, Null-Steering (LENS) algorithm for adaptive array beamforming in the case of a. known target location. Such beamformers are useful in spatial-filtering applications that enhance a desired target source while attenuating non-target, jammer sources in a given environment. Although LENS is designed for general beamforming purposes, this document em­phasizes the use of LENS to create a background-noise-reducing hearing aid. LENS processing is innovative in that it uses a novel robustness-control mechanism to yield a beamformer that avoids target cancellation under adverse conditions. Most traditional beamforming systems realize robustness control through the use of con­straints in the beamforming optimization, which is an approach that is both indirect and difficult to understand. LENS, on the other hand, achieves direct and obvious ro­bustness control by separating robustness control from the beamforming optimization in the following two-step procedure: first it solves a minima.By-constrained beamform­ing optimization in terms of the LENS parameter set [beta]_oct , and then it evaluates and ... The advantages of LENS processing are not limited to improved system robust­ness, however. Its design allows implementation using a relaxation-based approx­imation to direct-solution LENS procGssing (the LENS equiwdent of direct covari­ance matrix inversion processing for traditional systems). Simulations demonstrate that this relaxation-based implementation can combine efficient implementation, fast beamformer adaptation, and good beamforming performance, which is difficult to achieve with traditional systems.