Processing Three-Channel SAR-ATI for GMTI

This paper investigates two methods for GMTI using three-channel SAR-ATI data. Motivated by previous results in the two-channel case, a deterministic method for finding an Eigendecomposition of the SAR-ATI sample covariance matrix is presented. The CFAR detection capability of each of the eight parameters of the Eigendecomposition is evaluated using measured three-channel airborne data and compared to the wellknown DPCA-ATI metric. Results suggest that all moving target energy is contained in a single parameter, the largest Eigenvalue, and that for a given theoretical CFAR a lower real-world false alarm rate is achieved in comparison to DPCA-ATI. Motivated by this result ICA is then applied to the same measured threechannel data and found to both detect and distinguish two different types of moving target simultaneously present in a maritime environment.