Performance analysis of the SVD-based detection criteria for the case of two coherent signals

A novel set of signal detection criteria that are based on the singular values of the forward-backward data matrix is presented. These criteria are Akaike's information-theoretic criterion and the minimum description length criterion. After formulating the problem, the two detection criteria are presented, and then their performance is analyzed. Finally, simulation results that demonstrate the performance of these criteria are presented. The performance of these criteria in the presence of coherent multipath signals is analyzed. It is shown that the probability of detection is a function of the signal-to-noise ratio, the angle separation, the number of snapshots, and the intersignal correlation, specifically the constant phase angle between the direct path and its specular multipath component. Simulation results are also presented.<
>