Abstract
The dominant mode rejection (DMR) beamformer [Abraham and Owsley, OCEANS (1990)] modifies the Capon's minimum variance distortionless response to operate with low rank sample covariance matrices (SCM). The DMR imposes an eigenstructure on the SCM by replacing the eigenvalues of the noise subspace with an estimated noise power. Standard DMR estimates the noise power by averaging the noise subspace eigenvalues. This estimated noise power is negatively biased when the dominant subspace dimension is overestimated. The proposed median DMR estimates the noise power from the median of the noise subspace eigenvalues, based on the Marchenko-Pastur probability distribution. The median estimator is not dependent on the estimated interferer subspace dimension. Simulations demonstrated that the median DMR improves the white noise gain (WNG) when compared to the standard DMR in snapshot deficient scenarios with overestimated interferer subspace dimension. Increased WNG also implies increased robustness to array perturbations [Gilbert and Morgan, BSTJ (1955)]. This work compared the median DMR to standard DMR in simulations with perturbed array element phase responses. The scenario included two interferers and background white noise. The median DMR preserved deeper notches than standard DMR in this scenario, increasing the output signal-to-noise ratio by roughly 1.1 dB. [Work supported by ONR code 321US.]