面向普适计算的扩展的证据理论方法

Extended Method of Evidence Theory for Pervasive Computing

普适计算作为一种新型计算模式 ,从根本上改变人们对什么是计算的思考 .由于它需对多源信息进行融合 ,因此该文作者认为它是一种包含融合计算的模式 ,能通过多层次、多视角的融合 ,为人们提供更方便的信任度高的访问信息和计算服务 .基于普适计算应用的需要 ,该文讨论了扩展的证据理论方法 ,该方法采用可靠性因子评估多源证据觉察上下文信息 ;引入时效函数衡量多源证据的有效性与时间的关系 ,并将其组合到信任函数中 ,描述信任mass的时变规律 ;利用功率来度量多源证据觉察上下文信息间的相关程度 ,并通过去相关将其转化为相互独立的证据 ,扩展和完善了经典证据理论提供的方法 ,弥补了其不足之处 ,提高了不同应用场合下服务的质量 (QoS) ,确保了普适计算的服务宗旨 .利用支持普适计算模式的智能空间中的场景 ,验证了扩展的有效性 .

As a new kind of computing paradigm, pervasive computing changes the idea of what is computing radically. Because it fuses relative multi-source information among different computing nodes, a kind of paradigm with fusion computing can be regarded, which supply information access and computing service for people by information fusion from multiple levels and multiple visions confidently, credibly and conveniently. Owing to the need of pervasive computing, extended method of evidence theory is studied, which can assess reliability of multi-source evidence context-aware by credibility factor, add mass belief function by time-difference-calibration and power function by correlative degree among evidences to classic D-S method of evidence theory under considering the associated relationships between validity and time-efficiency and independency of evidences. The mass belief function has timely tracked dynamic process of evidences. The power function has measured correlative degree of evidences, based on this correlative degrees, de-correlation work can be done by transforming for conflict evidences. The method extends and improves the classic D-S method, overcomes its shortcoming, updates and improves QoS of different application fields, ensures and implements the target of pervasive computing paradigm. By application examples of Smart Space, such as Smart Meeting Room, as a test bed of pervasive computing paradigm, the validity of its extension and improvement has been tested successfully.