Distributed detection with decision feedback

A distributed detection scheme using feedback is proposed and studied. On each observation, peripheral detector decisions are combined in a central processor. Feedback links from the centre to each peripheral exist. Using a simple traffic reduction protocol, the central decision is communicated to all peripheral detectors and serves to update peripheral decisions on a new observation. The sequence of hypothesis tests representing the decision process is shown to be consistent. It is argued that the average overall data rate can be appreciably reduced for this scheme. Performance results for a Rayleigh channel are presented and suggestions for further investigations made.