Cramer-Rao bound-based adaptive quantization for target tracking in wireless sensor networks

This work deals with the problem of target tracking in wireless sensor networks where the observed system is assumed to evolve according to a probabilistic state space model. We propose to improve the use of the variational filtering (VF) by quantizing the data collected by the sensors to higher levels respecting the tradeoff between the information relevance of sensor measurements and the energy costs. In fact, VF has been shown to be suitable to the communication constraints of sensor networks. Its efficiency relies on the fact that the online update of the filtering distribution and its compression are simultaneously performed. But till now, it has been used only for binary sensor networks. In this paper, we propose an adaptive quantization algorithm taking benefit from the VF properties. At each sampling instant, by minimizing the Crameacuter-Rao bound, the adaptive quantization technique provides the optimal number of quantization bits per observation. The computation of this criteria is based on the target position predictive distribution provided by the VF algorithm. The simulation results show that the adaptive quantization algorithm, for the same sensor transmitting power, outperforms both the VF algorithm using a fixed optimal quantization level (minimizing the MSE) and the VF algorithm based on binary sensors.