Scalable distributed Kalman filtering through consensus

Kalman filtering is a classical technique with a number of potential distributed applications in sensor networks. In this paper we consider a specific algorithm for distributed Kalman filtering proposed recently by Olfati-Saber [1]. We design a communication access protocol for wireless sensor networks that is tailored to converge rapidly to the desired estimate and provides scalable error performance as number of sensors increases. By exploiting the structure of the distributed filtering computations, we derive an optimal communication resource allocation policy for minimizing the component-wise state estimation error. We provide simulation results demonstrating the performance of our architecture.