On the Complexity of Distributed Decision Problems
John N. Tsitsiklis, Michael Athans · Defense Technical Information Center (DTIC) · 1983
We study the computational complexity of finite versions of the simplest and fundamental problems of distributed decision making and we show that, apart from a few exceptions, such problems are hard (NP-complete, or worse). Some of the problems studied are the well-known team decision problem, the distributed hypothesis testing problem, as well as the problem of designing a communications protocol that guarantees the attainment of a prespecified goal with as little communications as possible. These results indicate the inherent difficulty of distributed decision making, even for very simple problems, with trivial centralized counterparts and suggest that optimally may be an elusive goal of distributed systems. (Author)