Fusion and Diversity Trade-Offs in Cooperative Estimation over Fading Channels

In this paper we consider a network of distributed sensors that are trying to measure a parameter of interest cooperatively, by exchanging their acquired information repeatedly over fading channels. We consider different ways of using the available bandwidth, in terms of what each node can send to its neighbors. More specifically, we characterize the impact of local fusion and show how it is a suitable policy when graph connectivity is low. When poor link qualities are the main bottleneck, on the other hand, we show how a diversity approach can be more beneficial. The proposed framework highlights the underlying tradeoffs between fusion and diversity approaches in cooperative networks. It furthermore explores the impact of multiple sensing on the overall performance.