Decentralized acoustic source localization with unknown source energy in a wireless sensor network

摘要

Acoustic source localization is one typical application for monitoring tasks in a wireless sensor network (WSN). The centralized energy ratios based method is often used when the source energy is unknown. However, the centralized approach requires that there are some super nodes served as the processing centers which have greater communication bandwidth, higher computation capability and more energy. In this paper, we propose a decentralized energy ratios based acoustic source localization method using the incremental gradient (IG) algorithm and its modification, the normalized incremental gradient (NIG) algorithm. During localization, the localization message cycles through the network from neighbor node to neighbor node, till the location estimation converges to a limit point or the itineracy exceeds a predefined cycle number. On each visited sensor, the estimation is updated based on the energy readings and locations of this sensor and the prior one. Results of simulation and real world experiments show that the NIG algorithm can obtain the approximate accuracy of the exhaustive search (ES) method with much less computation cost.