Distributed consensus in noisy wireless sensor networks

In this paper, the average consensus problem in wireless sensor network (WSN) is considered. The sensor nodes communicate over imperfect radio channels. The information is distorted by measurement noise, quantization noise and channel noise. While previous studies only model it as zero mean additive noise, this piece of work utilizes a noise model that makes no assumption on the noise probability density function and proposes an average consensus algorithm in wireless sensor networks. The proposed method is topology unaware, and can be applied to general networks. Theoretical analysis shows that it provides unbiased global average estimation. Numerical results show that the proposed algorithm has better mean square error than three other conventional algorithms.