Architecture of RETE network hardware accelerator for real-time context-aware system

Abstract

Context-aware systems, such as intelligent home-care systems or mobile communication devices that are aware of the channel environment, need reasoning ability with numerous rules to manage the current context. Reasoning techniques based on rule-based systems can be used for the efficient reasoning method of these numerous rules. The RETE algorithm has been used for the matching of reasoning rules in rule-based systems. However, the characteristics of the RETE algorithm cause it to have poor performance in Von Neumann computer systems. In this paper, we propose a novel architecture for the RETE network hardware accelerator, which provides efficient reasoning processing performance. Using the parallel RETE network hardware architecture, this accelerator can overcome the architectural constraints imposed by Von Neumann computer systems. © Springer-Verlag Berlin Heidelberg 2006.