Distributed problem solving in time-constrained domains

Abstract

Distributed problem solving (DPS) techniques assume a priority partitioning of the problem and also problem solving knowledge. In most time-constrained domains (TeDs), the problem and consequent partitioning of knowledge is not trivial. A graph-based partitioning technique for rule-based systems using individual agent capabilities was developed earlier. In TeDs, the application's time constraints are an additional factor to be considered while partitioning the knowledge. In this paper. we extend the knowledge partitioning technique by proposing  heuristics to be applied in TCDs. Further, a highilevel Petri net (HLPN)-based knowledge model is used instead of the graph proposed in the earlier approach. HLPN is a formal modelling technique which integrates various aspects of intelligent problem-solving like knowledge representation, verification and reasoning.