Multiple broad-band source location using steered covariance matrices

The authors present an approach for reducing the threshold observation time required to achieve high-resolution localization of multiple broadband sources. The proposed techniques are based on a space-time statistic called the steered covariance matrix (STCM). The STCM, like the well-known cross-spectral density matrix (CSDM), has asymptotic properties which facilitate high-resolution source localization. In broadband settings, however, the STCM has the advantage that it can be estimated with much greater statistical stability than the CSDM. The STCM is used in conjunction with minimum variance and linear predictive spectral estimation to obtain the steered minimum variance (STMV) and steered linear prediction (STLP) methods. Analytical and simulation results are presented that indicate that the STMV and STLP methods exhibit lower threshold observation times than their CSDM-based counterparts