We investigate state estimation via wireless sensor networks over fading channels causing random packet loss. Packet loss probabilities depend upon the time-varying channel gains and transmission power levels used by the sensors. We develop a predictive controller which trades off sensor energy expenditure versus state estimation accuracy. The latter is measured by the expected value of the future covariance matrices provided by the associated time-varying Kalman Filter. To further conserve energy at the sensors, the controller is located at the gateway and sends coarsely quantized power increment commands, only whenever necessary. Simulations based on real channel measurements show that the proposed approach gives excellent results.
47th IEEE Conference on Decision and Control (CDC 2008). Proceedings of the 47th IEEE Conference on Decision and Control (Cancun, Mexico 9-11 December, 2008) p. 1103-1108