A method of blind separation for coherent source based on single vector sensor

Abstract

A method of blind signal separation based on second order statistic analytic equations is provided for a single sensor. This method takes second order statistics equations as restrictive conditions to construct blind separation criterion, then traditional blind separation methods are used to separate source signals through iterative optimization of genetic algorithm. In theory, not only can the coherent signals with overlapping spectrumbe separated, but the parameters of the sources such as DOA and intensity can be also estimated. Simulation demonstrates effectiveness of algorithm and separation performance.