A predictive localization algorithm based on RBF neural network for wireless sensor networks

The motion trajectory of nodes in indoor environment is relatively fixed because of the spatial constraint. In addition, mobile node usually moves according to some rules of its own. The localization error would increase when mobile nodes in indoor wireless sensor networks cannot receive the location information sent from anchor nodes due to some unknown transient disturbance. To minimize the localization error, we propose a predictive localization algorithm based on RBF neural network (PLRNN). The algorithm extracts and learns the intrinsic moving rules of mobile nodes. Through the extracted moving features, the location of mobile nodes can be predicted. Simulation results confirm that this algorithm can realize predictive localization with higher accuracy in blind period.