Structured Covariance Estimation: Theory, Application, and Recent Results

The maximum-likelihood approach to structured covariance estimation and spectrum estimation has wide applicability in time series analysis, spectroscopy, adaptive beamforming and detection, remote sensing, radio astronomy, and radar imaging. Standard structured covariance EM algorithm with full model matrices is computationally demanding. Computational requirements drastically reduced when model matrices are sparse. Sparse structure may be achieved through appropriately chosen data preprocessing. We are investigating application in problem of airborne radar imaging from multiple viewpoints, previously computationally unrealistic