The energy-based multisource number estimation in a wireless sensor network

Many existing methods about multisource localization need to know the number of sources. However, the number of sources may not be known beforehand in real applications. In this paper, the estimation of the number of acoustic sources with signal strength measurements in a wireless sensor network is considered. And two methods are proposed. In the first one, we use Voronoi diagram to filter out the sensors with relatively strong measurements, and then cluster the filtered sensors geographically to determine the number of sources. In the second one, we use the information theoretic criterion, minimum description length (MDL), to estimate the number. We assume that the measurement of a sensor only depends on its closest source, and thus reduce the computation burden of parameter estimation in the MDL scheme. Extensive simulations have been conducted and demonstrate that both of the two methods are effective and perform well.