Bias and variance analysis of MUSIC location estimates

Abstract

Presents a statistical performance analysis of the MUSIC location estimator. Explicit and concise asymptotical expressions for both the bias and variance of MUSIC location estimates are derived for multiple source cases. These expressions are valid over a wide range of SNR extending down into the resolution threshold region of MUSIC, where analysis is of particular interest. These expressions, though asymptotical, can also be accurately applied to limited number of observation cases. In comparison with previous analysis, this new statistical analysis results in significantly less complicated statistics and derivations and is more general in that it accurately examines both the bias and variance of MUSIC location estimates. Further, this analysis approach can be applied to other eigenspace spatial-spectrum based location estimators.<>