Cluster based Routing Scheme for Distributed Regression in Wireless Sensor Networks: Gaussian Eliminations

This paper addresses the issue of the in-network implementation of data regression in wireless sensor networks (WSNs). In particular, a cluster based routing scheme is proposed for achieving the distributed Gaussian elimination in solving the regression algebraic equation. In doing this, the tree structure is incorporated to coordinate the clusters so as to decouple the intersections with each other via a new message passing protocol. The cluster based routing scheme proposed here contributes not only to ease network topology management, but also to speed the convergence of distributed computation. Experimental results are reported to illustrate the validation of the proposed approach.