Speculative constraint processing for hierarchical agents

Hiroshi Hosobe, Ken Satoh, Jiefei Ma, Alessandra Russo, Krysia Broda

Abstract: Speculative computation is an effective means for solving problems with incomplete information in multi-agent systems. It allows such a system to compute tentative solutions by using default knowledge about agents even if communications between agents are delayed or fail. Previously we have proposed a logical framework for speculative constraint processing for master–slave multi-agent systems. In this paper, we extend the framework to support more general multi-agent systems that are hierarchically structured. We provide an operational model for the framework and present a prototype implementation of the model.