Fully distributed acoustic source localization in wireless sensor network

Abstract

Each node in wireless sensor network has limited energy for communication and computation, so how to reduce the energy consumption efficiently is an important research problem. In this paper, we propose a fully distributed expectation maximization algorithm (FDEM) for acoustic source localization to reduce communication cost significantly. It just requires local data transmission and simple computation. Each sensor node transmits energy observations to its nearest neighbors (nodes within one-hop communication range) and also receives energy observations from its nearest neighbors. Long range wireless communication is avoided and the energy is saved. Simulation results for a static source localization and a moving source tracking demonstrate that the FDEM algorithm can provide a good tradeoff between localization accuracy and communication cost.