MUSIC-based direction of arrival estimation methods with virtual source elimination for single vector sensor in isotropic ambient noise

摘要

In the scenario of isotropic ambient noise,the inequality between the noise powers of the pressure and velocity channels of a single acoustic vector sensor causes a virtual source for subspace-based methods,which seriously degrades the performance of the conventional multiple signal classification(MUSIC) method.In order to solve this problem,an increased-rank MUSIC method and a weighted-amplitude MUSIC method are presented.The increased-rank MUSIC method augments the dimension of the signal subspace to eliminate the effect of the virtual source.Among the increased-rank,weighted-amplitude,and conventional MUSIC methods,the increased-rank MUSIC method gains the highest estimation accuracy in the cases of one source.Moreover,its performance approaches to the the Cramer-Rao bound(CRB).However,it fails in the case of two sources.The weighted-amplitude MUSIC method weights the amplitude of the received signal to eliminate the virtual source and it can identify two sources.