【作者】 叶剑； 李锦涛； 史红周； 朱珍民； 顾晓光；

【Author】 Jian Ye1, 2 +, Jintao Li1, Zhenmin Zhu1, Hongzhou Shi1, Xiaoguang Gu1,2 1Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China 2 Graduate School of Chinese Academy of Sciences, Beijing 100190, China

【机构】 中国科学院计算技术研究所；

【摘要】 普适计算环境通常采用集中式上下文推理机制,不但加重了推理服务器的负担,而且没有充分利用终端计算能力。而基于规则的模糊推理易于分解为独立的命题判定,这就为分布式推理提供了条件基础。本文提出一种面向分布式推理的模糊Petri网模型——dFRPN。dFRPN采用层次模型,通过两层划分,表征推理服务器和终端的推理任务执行;dFRPN在传统的FRPN定义中引入负载度和负载增长的概念,在Petri网控制中监控节点资源占用,在推理、迁移和提交之间做出决策。通过改进模糊推理Petri网(FRPN)的定义、机制和算法,dFRPN可以有效规约普适计算环境中的分布式推理机制,反映了分布式推理的本质特征。更多还原

【Abstract】 In a pervasive computing environment, contexts are initially got and stored on those nodes scattered over the environment. However, the traditional reasoning about contexts applied a centralized approach which aggravated the load of reasoning server and cost for communication of context. As an important approach for context reasoning, the rule-based reasoning can be easily decomposed of independent propositions. Therefore, the rule-based context reasoning can be decomposed and distributed over those nodes of the pervasive environment to decrease the computing load of reasoning server. In this paper, we propose a distributed fuzzy reasoning Petri net model with load detection (dFRPN) towards the decomposition of distributed fuzzy reasoning. The dFRPN model is able to formalize the distribution of fuzzy reasoning over every node of the whole environment. The dFRPN model is characterized as a hierarchical structure to make a description of reasoning on nodes more clear. Considering the limited capabilities of some mobile nodes such as PDA and smart phone, we add the special migration transition to define the load detection and relative reasoning actions. At the end of the paper, the feasibility of the dFRPN model is validated through a case of context-awareness based personalized recommendation system.更多还原