Enhanced Minimum Variance Beamforming

Abstract

The MVDR criterion for achieving maximum wavefront Signal-to-Noise Ratio (SNR) gain with a spatial sensor array beamformer is well established. The MVDR beamformer has its technical origins in the early 1960’s (see [1] for bibliography). Eigenstructure based, signal subspace wavenumber spectrum analysis procedures appeared in the early 1970’s [2, 3] and have evolved to the present day high resolution procedures typified by the MUSIC algorithm [4]. The difference between a beamformer, which can produce a directional waveform estimate, and a wavenumber analyzer which provides a source direction estimate only is emphasized. Accordingly, the evaluation of MVDR and eigenstructure techniques has been along two parallel paths. This dual perspective has precluded the establishment of common theoretical and practical approaches which could result in the exploitation of the advantages of both techniques. The enhanced minimum variance (EMV) beamformer represents an attempt to find such a common viewpoint and to subsequently address current space-time processing technical issues from this unified perspective. Such issues as detection of threshold signals, the determination of the number of sources present, reducing computational burden, coherent multipath arrivals and wideband signals are of particular concern.