RETE-Based Context Reasoning In Smart Environment

With the continuously maturing and expanding of sensor technology, embedded technology, mobile computing, Context-aware reasoning, ubiquitous computing is ushering in a research boom. Applications of smart environmental in pervasive computing is gradually out of the lab and it got more and more attention. In order to provide users with accurate, personalized, intelligence service, how to use a variety of intelligent devices to obtain the original context data in living, working, public places, and to sense the current state of the current environment and to infer intent and motivation which is complex context information of current user even further. This is the key issue in the context-aware inference of smart environment.In this paper, the Rete-based context-aware inference engine is studied, And the distributed reasoning based on the Rete algorithm in smart environment is also studied. In this case, the main work of this paper includes the following aspects:(1) A simple and efficient key-value model is used to describe the context of smart environment in production system. And context is classified into atomic context and complex context according to the characteristics of production rule. Rete algorithm is proposed to context-aware reasoning. Rete reasoning network is compiled by rule decomposition and replacement according to compiling processes of Rete algorithm. In addition, a nested loop detection algorithm of rules is proposed to solve the conflict in rule set; a method that merge Single-mind node in Rete network is proposed to adapt to storage resources limit of intelligence environment.(2) A tree topology is used to describe sensor network of smart environment. The deployment of Rete network in smart environment both centralized and distributed Rete distribution is discussed, and a Rete distribution cost model based on transmission and computing costs is given. At last, based on this model, the Minimum Transmission Cost Rete Distribution Scheme algorithm is proposed. At the same time, two simulation comparison experiments between centralized and distributed Rete distribution is designed. The experimental results show that the Minimum Transmission Cost Rete Distribution Scheme has lower data transmission, especially in the case of large-scale rules and low frequency rule trigger.(3) A Rete-based inference engine for context is designed and implemented. It construct a simple set of rules, and use this set to simulated context reasoning, in order to verify the accuracy and effectiveness of this engine.