DOA estimation of correlated sources using SMT

This paper uses a recently developed technique that relies on the Sparse Matrix Transform (SMT) to estimate the covariance matrix of D signals received by M-elements linear antenna array, each signal is of length N (the number of snapshots is N where N <; M). SMT based covariance estimation is particularly suited for singular covariance matrices and those with small eigenvalues. Direction of arrival (DOA) estimation using the MUSIC algorithm relies on estimating the eigenvectors associated with the noise eigenvalues which are usually minimal. Also, when the sources impinging on an array are correlated, the covariance matrix is singular, and the performance of the MUSIC algorithm degrades significantly depending on the degree of correlation among sources. This makes SMT particularly suited for DOA estimation using MUSIC for partially or fully correlated sources, and especially scenarios where it is not practical to take a large number of snapshots (such as radar applications). This paper employs SMT in the MUSIC algorithm using real radar backscatter data as the sources. Limitations and benefits of SMT based DOA estimation are discussed.