Fast Distributed Consensus-based Estimation (Fast-DiCE) for Cooperative Networks

In distributed networks several nodes aim to estimate the signals broadcasted by the sources in a cooperative fashion by exchanging information among neighboring nodes. Consensus based estimation is a specific class of In-Network-Processing (INP) techniques and the Distributed Consensusbased Estimation (DiCE) algorithm is an efficient realization for the Least Squares (LS) estimation problem. In this paper, we modify the update functions of the DiCE algorithm by applying Nesterov‚s optimal gradient descend method leading to the novel fast-DiCE algorithm. Furthermore, we extend the distributed estimation algorithms with respect to the MMSE-criterion. The performance of the discussed schemes will be evaluated considering two different applications: a cooperative sensor network and the uplink of a small-cell mobile communication network. The results indicate the improved convergence of the fast-DiCE algorithm resulting also in a reduced communication overhead among the cooperating nodes.