Distributed Sequential Estimation With Noisy, Correlated Observations

The problem of distributed sequential estimation of a nonrandom parameter over noisy communication links is considered, with observations that are correlated spatially across the sensor field. A recursive algorithm for updating the sequential estimator is derived assuming wide-sense stationary observations. It is shown that the performance of the sequential estimator is a trade-off between the quality of node observations and communication channels. A sufficient condition for the convergence of the estimator variance is derived, and asymptotic expressions for variance when this condition is met are obtained for both iid and correlated observations.