Skip to Main Content
A multi-model formulation is obtained for the Hierarchical Dynamic Resource Allocation Problems. which is a class of two-level, multi-decision maker control problems with non-deterministic dynamics and feasibility objective. Each decision maker in the lower hierarchical level (known as a Local Manager) is assumed to possess a different incomplete model of the underlying dynamical system. The fesibilty of such problems can be determined in a distributed fashion without first centralizing the model knowledge possesed by each Local Manager. Instead, the Local Managers are required to incrementally pass their respective model knowledge to the coordinating decision maker (known as a Resource Manager) during the computation proces. It is demonstrated that such a distributed approach to feasibility test may result in considerable reduction of computation over a centralized approach which begins with a full model knowledge of the underlying dynamical system. The saving in computation that can be accomplished is found to be problem dependent.
American Control Conference, 1987
Date of Conference: 10-12 June 1987