Covariance Matrix Estimator Performance in Non-Gaussian Spherically InvariantRandom Processes

摘要

This report describes the performance of the covariance matrix estimator in non-Gaussian spherically invariant random processes (SIRP). Analytic expressions are derived for the variance of the estimator. Specific consideration is given to the special cases of Weibull and K-distributed processes as a function of the shape parameter. Validation is achieved via Monte Carlo simulation. The expressions reveal the increase in the estimator variance for non-Gaussian SIRP's as well as the sample support size required to reduce the variance to that of the Gaussian case.