Statistical Signal Processing in Engineering

22Cooperative Estimation

Cooperative estimation refers to system where there is a set of nodes accessing local measurements related to a common phenomenon. Each node can estimate locally the set of parameters in the conventional way, but cooperation among the nodes gives a benefit to all as the accuracy of the estimates refined from cooperation is largely improved wrt the individuals. Nodes are complex entities as they have a sensing system for measuring, processing for estimating, and communication capability to enable the cooperation with other nodes. In cooperative estimation, each node cooperates with others to gain a benefit without exchanging the local measurements as this would infringe some confidentiality issues related to the measurements, or the availability of a local model. One example is when every node has a partial coverage of the parameters and cannot estimate them all as the number of measurements is insufficient for a single node, but cooperation with others lets every node act as if all measurements were put in common, thus gaining measurements and improving the accuracy.

The reference model is illustrated by Figure 22.1, where there are a set of K measuring points for the same set of parameters θ. In the additive noise model, the data at the i th node is

and the relationship is in general non‐linear. Each node can carry out the local estimate by using any ...