For the decentralized sequential detection problem studied by Hashemi and Rhodes, it is pointed out that likelihood-ratio tests are not necessarily optimal at the sensors when the decisions made by the sensors are allowed to depend on all their past observations. It is also argued that likelihood-ratio tests are indeed optimal if one restricts the decisions made by each sensor to depend on its present observation and its past decisions.
- Decentralized sequential detection
- likelihood-ratio tests
ASJC Scopus subject areas
- Information Systems
- Computer Science Applications
- Library and Information Sciences