The W-disjoint Orthogonality of Underwater Acoustic Signals and Underdetermined Source Counting for Acoustic Vector Sensor

摘要

In this paper, we survey the multi-sources distinguish ability of the direction of arrival (DOA) histogram based on acoustic vector sensor (AVS) by verifying the W-disjoint orthogonality (WDO) of the underwater acoustic signals. The relationships between the WDO property and the source number, frequency and window length were studied. We also conduct the underdetermined source counting by modeling the DOA histogram from AVS as a Gaussian mixture model (GMM) with a Dirichlet prior. The parameters of the GMM are estimated by a maximum a posteriori (MAP) approach employing the expectation-maximization (EM) algorithm. The underdetermined sources are counted by picking out the Gaussian components fitting to the source clusters according to the estimated parameters.