State estimation & self-localization using distributed Kalman filter & recursive expectation maximization algorithm in sensor networks

Knowing the fact that online expectation maximization is a well-known methodology for static parameters estimation in a general state-space model, this paper describes fully how a decentralized version of online EM algorithm can be implemented in a sensor network for the self-localization problem. This is done through the propagation of messages that are exchanged between neighboring nodes of network. The algorithms used for state/parameter estimation are performed in a fully collaborative manner. Comparing parameter estimation formulas of On-line EM algorithm with RML method easily shows the simplicity of the former, while the results are approximately the same for both.