Survey of resource allocation methods for distributed manufacturing systems

Abstract

Distributed manufacturing systems are more flexible in their arrangement and use of their resources. This permits a more dynamic and opportunistic allocation of manufacturing resources to work elements than is possible with today's centralized allocation systems. However, unlike solving the problem from a single solver view, there is no absolute control or tight coordination over how and when to decompose the problem, when to undertake a solution or the procedure for verifying consistency. Such activities will largely be left to the distributed entities to decide either individually or jointly. As a consequence the resulting allocations may suffer and induce myopic behaviour lacking in stability and regularity in the performance of the total system. Thus, the issues connected with solving the problem often move beyond the pure algorithmic focus of computation to include how the problem is decomposed, how or to whom it is assigned, and the organization for solving the subproblems, strategies used for resolving conflicts, the manner in which communication is conducted, etc. These issues, in fact, highlight the holistic nature of the solution and are used here as parameters to compare and contrast the approaches reported in the literature.