Skip to Main Content
Multiple target detection in sensor networks is a challenging problem since the signal captured by individual sensor node is normally a linear/nonlinear weighted mixture of the source signals. Independent component analysis (ICA) has been widely used to solve the source estimation problem but most of the algorithms assume the number of sources is fixed and equals to the number of observations which generally is not the case in sensor networks. Even though several methods are put forward for the source number estimation, the centralized scheme hinders their application in sensor networks due to the extremely constrained resource and scalability issues. In this paper, a distributed source number estimation framework is developed, where the local estimation is generated within each cluster and a fusion algorithm is performed to combine the local results. We derive a posterior probability fusion method based on Bayes theorem and compare it with the Dempster rule of combination. Experimental results show that using the distributed framework, the confidence of source number estimation is improved over the centralized approach while at the same time, the network traffic can be significantly reduced and resources can be conserved.