Dynamic Decentralized Sequential Detection Under Uniformly Distributed Observations

Decentralized sequential detection problems with communicating detectors are solved under the assumptions that the observations are independent and uniformly distributed and that decision errors are uniformly penalized. For the cases of one-step delayed, sequential, and nonsequential signaling, explicit decision rules are proposed and exact conditions for their member-by-member optimality are derived. These decision rules approximate high-performance threshold rules with very strict thresholds when decision errors are heavily penalized.