Localization of acoustic sources utilizing a decentralized particle filter

This paper presents recent results of a decentralized localization scheme. Several sensors are embedded in an acoustic wave field. We assume that the field variables of interest are governed by a discrete-time spatial-distributed state-space equation. In particular, acoustic waves are emitted by sources and observed by sensors. For decentralized localization, a further decomposition of the state-space model gives a decentralized model. A decentralized estimation algorithm infers jointly the field and source states. The messages between sensors are sent over an imperfect channel. For qualitative statements of the estimation, its performance is explored by an analytic sequential Weiss-Weinstein bound on the estimation error.