Fourth-Order MUSIC Algorithm Based on Minimum Redundant Virtual Extended Array

A novel forth-order multiple signal classification (MUSIC) algorithm was proposed based on the virtual array technique. Since the effective array aperture is extended according to the higher-order steering vector expression, a virtual spatial covariance matrix is derived using the forth-order cumulants. In this way, the redundant information contained in the cumulants are removed, and the computational complexity is reduced for the dimension of the virtual covariance matrix is decreased, which is equal to the number of the virtual array elements. Numerical results show that for Gaussian signals, the performance of the second-order methods is optimal. However, for non-Gaussian signals, the direction estimation variance of the proposed method is decreased and the resolution probability of two close signals is improved, compared to those of the second-order MUSIC algorithm