On the use of order statistics for improved detection of signals by the MDL criterion

Abstract

The minimum description length (MDL) criterion for model selection has been successfully applied to the problem of estimating the number of sources in noise. We show that by employing an order statistics approach in the existing solution, a new processor with uniformly improved performance is constructed at the cost of greater complexity. We prove that the resulting MDL-based processor is consistent. Results of simulations of practical scenarios of interest demonstrate how significant the performance improvement can be.