Input driven consensus algorithm for distributed estimation and classification in sensor networks

This paper deals with the problem of simultaneously classifying sensors and estimating hidden parameters in a network with communication constraints. In particular, we consider a network where sensors measure a common parameter with different precision rank. The goal of each unit is to estimate the unknown parameter and its own specific type through local communication and computation. Here, we present a decentralized version of the centralized maximum likelihood (ML) estimator. Each sensor computes local sufficient statistics by using its own observations and transmits its local information to its neighborhood. By using an Input Driven Consensus Algorithm (IDCA), the local information can be gradually propagated through the entire network, allowing to estimate the global parameter. We prove the convergence of the proposed algorithm and we show that the relative classification error converges to that of the centralized ML as the network dimension goes to infinity. We also compare this strategy with implementation of expectation-maximization (EM) algorithm via numerical simulations.