Model order identification and parameters estimation using array processings for small sample support
Yu Rong, Alphonso A. Samuel, Daniel W. Bliss · 2017
We propose a sequential algorithm for determining the number of narrow band source signals, and estimation of the parameters associated with the signal (angles and powers) and of the receiver noise power using the noisy sample observed at the receiver sensor arrays. Previous parameter estimates (or initial estimates) are refined multiple times in our sequential approach and also the Newton-based refinement gives continuous-valued estimates so the estimation performance is not limited to the grid resolution. By benchmarking against the Cramer Rao Lower Bound (CRLB), the estimation performance for all parameters of the proposed algorithm achieves near optimal performance even in the low SNR and small sample support region, in which, the sample size can be smaller than the number of sensors in the array. At the same time, the detection (or model order identification) performance outperforms other relevant algorithms.