Decentralized Bayesian Detection Theory
This chapter presents an overview of the available results on Bayesian hypothesis testing for decentralized systems. For system optimization, a person-by-person optimization methodology for this team decision problem is adopted. The fusion center is one team member, whereas the aggregation of the local detectors is the other team member. If, however, the observations at the local detectors are assumed to be conditionally independent, the local decision rules reduce to threshold tests. The fusion center determines a global decision and feeds it back to the local detectors. Local decision rules are adaptive in that they are based on the incoming observations and the global decision received. Detailed analysis and performance results are available in Al-Hakeem and Varshney. The on-line signal processing effort is comparable in centralized and distributed detection systems. But the computational effort in the design of decentralized detection systems, which is carried out off-line, is quite intensive.