An adaptive estimation algorithm for distributed networks with noisy links

In this paper we deal with online estimation of an parameter, using measurements collected by an ad hoc wireless sensor network (WSN), where the links between nodes (sensors) are noisy. Although the recently introduced distributed adaptive estimation algorithms (adaptive networks) are promising solutions for the mentioned problem, however, when the links are noisy, the performance of these algorithms drastically decreases. In this paper we propose a new adaptive estimation algorithm based on incremental cooperation between nodes and least mean-square (LMS) adaptive filter that mitigates the effect of noisy link and offers an acceptable performance in networks with noisy channels. In the proposed algorithm, each nodes updates its local estimate by means of a combination of both time and spatial update. More precisely, every node uses its previous-time local estimate and only a part of the prior node estimate to update its current local estimate. As our simulation results show, the proposed algorithm offers a way to use adaptive estimation in noisy links.