This chapter presents a formal approach to synthesizing ADDM algorithms, the underlying motivations being the need to coordinate the decisions of the entities to achieve global optimal behavior for the overall system and to establish a scientific basis to rate the quality of their decisions. The literature on synthesizing decentralized decision-making algorithms and evaluating the quality of distributed decision making, is sparse. Rotithor [51] proposes distributing the overall tasks system state estimation and decision making of a decentralized decision-making system among independent decision-making processing elements and attempt to improve the overall performance of the system by optimizing the performance of the individual decision makers. Rotithor models the problem of load balancing in distributed computing and notes substantial performance improvements. Tsitsiklis and Athans [52] consider the distributed team decision problem wherein different agents obtain different stochastic measures related to an uncertain random vector from the environment and attempt to converge, asymptotically, on a common decision, and they derive the conditions for convergence. In [53], Tsitsiklis and Athans study the complexity of basic decentralized decision problems that are variations of team decision problems and conclude that optimality may be an elusive goal. It is noted that in contrast to the problems studied in [52] [53], here, each entity derives its own decisions possibly different from those of other entities while attempting to achieve global optimal behavior.
This chapter is based on the distributed hybrid control paradigm of Kohn and Nerode [199] [200] in that every...
TABLE OF CONTENTS 
