Efficient Implementation of Distributed Maximum Likelihood Estimation Method in Clustered Wireless Sensor Networks
X. F. Zhang, H. Y. Liu · 2013
The problem of efficient implementation of maximum-likelihood (ML) estimation of an unknown deterministic vector parameter in a clustered Wireless Sensor Network (WSN) is considered in this paper. In our previous work, we developed a distributed estimation manner through the combination of Lagrangian multiplier and block coordinate descent methods. In this paper, we give rigorous proof that the distributed estimation approach converges to the ML estimation result assuming ideal communication links. We also perform simulations to demonstrate the performance of the approach under additive-white Gaussian noise channel. Then we propose the efficient implementation of the most complicated step of the approach based on the preconditioned conjugate gradient method. Moreover, we present a hierarchical architecture for the implementation of the approach, which suggest our distributed estimation approach is favorable for practical uses. Copyright © 2013 IFSA.