Designing ITC selection algorithms for wireless sources enumeration

A common approach for estimating the number of wireless sources is to adopt model order selection based on information theoretic criteria (ITC). In this paper we study the generalized information criterion (GIC) and propose a design method for setting the penalty in practical situations, where the sample size is finite. The design is based on the maximum probability of correct model selection, that can be approximated using the statistic of the ratio between the largest eigenvalue and the trace of a white central Wishart matrix. For this metric we provide the exact distribution and a new approximation.