Sensor array processor evaluation with a standardized test case (STC)

A standard test case (STC) for the comparative evaluation of the plethora of modern space-time signal processing techniques is summarized. The STC is thought to be realistic in the context of superimposed multiple discrete and continuous ambient noise fields for a uniform linear array of sensors. Attention is given to the MUSIC direction finder (DF), the ESPRIT DF, the linear prediction based DF and the enhanced minimum variance distortionless response beamformer. The asymptotic estimate of the source subspace dimensionality is used, and the performance of several algorithms for the estimation of the dominant source subspace spanning eigenvectors is evaluated with respect to convergence rate and variance. The issue of false alarms induced by overestimating the source subspace dimensionality relative to the actual number of discrete sources is discussed. No clear performance advantage can be found between gradient search and orthogonal iteration, which are the best of the subspace estimators considered