Partially decentralized estimation in wireless sensor networks

We propose and analyse a partially decentralized estimation scheme (PDES), based on a one-hop neighbor clustered architecture. Under some realistic assumptions, we show that PDES, comparing with the decentralized estimation scheme (DES) in [J.J. Xiao and Z.Q. Luo, 2005; Z.Q. Luo, 2005], reduces the communication requirements and retains the statistic accuracy of the estimator by mean square error (MSE) criterion. Finally, we develop a simple algorithm of one-hop neighbor cluster which is a vital factor to consider PDES.