Smooth Order Statistics for Estimating the Number of Coherent Signal Sources

This document explores how to determine the number of coherent signal sources through smooth order statistics estimation. The methodology involves establishing signal source models and implementing simulations based on the Minimum Description Length (MDL) criterion. By employing this approach, we can achieve more accurate source number estimation with enhanced reliability. During signal model construction, key factors such as signal strength, frequency, and phase relationships are systematically incorporated. The MDL criterion implementation involves evaluating model complexity through eigenvalue decomposition of covariance matrices - typically calculated using MATLAB's eig() or svd() functions on spatially smoothed data matrices. This enables optimal model selection by balancing estimation accuracy against parametric complexity through information-theoretic penalties. Consequently, this framework yields detailed and precise results that facilitate deeper understanding and analysis of coherent signal source characteristics and behaviors, with practical implementation suggesting iterative thresholding on ordered eigenvalues to identify significant signal components.