Dynamic Energy Management with Improved Particle Filter Prediction in Wireless Sensor Networks

Abstract

Energy efficiency is a primary problem in wireless sensor networks which employ a large number of intelligent sensor nodes to accomplish complicated tasks. Focused on the energy consumption problem in target tracking applications, this paper proposes a dynamic energy management mechanism with an improved particle filter prediction in wireless sensor networks. The standard particle filter is improved by combining the radial-basis function network to construct the process model and the novel algorithm is adopted to predict the prior position of target. For dynamic awakening, the idle interval of each sensor node is estimated according to its sensing tasks. A cluster head rotating approach is introduced from low-energy adaptive clustering hierarchy for collecting data through the large sensing field. A group of sensor nodes which are located in the vicinity of target will wake up and have the opportunity to report their data. Distributed genetic algorithm is performed on cluster heads to optimize the sensor node selection. In target tracking simulations, we verify that the improved particle filter has more robustness than standard particle filter against the sensing error and dynamic energy management enhances energy efficiency of wireless sensor networks.