An extended information PF for wideband acoustic source tracking using a distributed AVS array

Consider the problem of tracking multiple acoustic sources in three dimensional (3-D) space using a distributed acoustic vector sensor (AVS) array. Traditionally, indirect approaches are widely used, by which the direction of arrival (DOA) of the source at each sensor is estimated first, and the DOA estimates are then employed to intersect a 3-D position. The performance of position estimation can be seriously degraded by inaccurate DOA estimates at the first stage, and multiple source localization is impossible unless the DOA estimates can be associated to each source accurately. In this paper, a particle filtering (PF) approach is developed to directly fuse the signals collected from distributed sensors and track the 3D positions. To enhance the tracking performance and keep the computational complexity affordable, an extended information filter is developed to achieve the optimal resampling. The simulations show that the proposed tracking approach significantly outperforms the indirect localization approaches and is able to track multiple active sources accurately.