Source number estimator using the peak-to-average power ratio modified by Gerschgorin radii

In this paper, a source number estimator using the peak-to-average power ratio modified by Gerschgorin radii is proposed. The eigenvectors of the sample covariance matrix are first employed to calculate the peak-to-average power ratio, and then a Gerschgorin transform is taken to the sample covariance matrix, next the new peak-to-average power ratio values modified by Gerschgorin radii are introduced to minimum description length criterion (PGMDL). Simulation results indicate that the proposed method shows better performance than conventional detection methods such as AIC and MDL especially at lower SNR, and has no affectedness by the intensity difference between the multiple sources.