Decomposition and allocation of flat-structured problems

A classification is made of the applications of distributed problem solving (DPS) into hierarchically structured problems (HP) and flatly structured problems (FP). A formal description is given of an FP and its solving system. An improved problem decomposition (IPD) algorithm is presented by which a problem is decomposed in size into several tasks of identical property and each task is allocated to a proximate agent. Heuristic state space search is used to balance load. Experiments on a distributed transport dispatching system, DTD-1, indicate that this algorithm is effective and superior.<
>