Context-Aware Self-Adaptive Composition Of Pervasive Services

With the development of computing technology and network technology, pervasive computing is becoming a novel computing paradigm and pervasive application systems have been developed rapidly. The idea of pervasive computing is human centric, i.e. user can acquire service at any time any place. At the mean time, service oriented architecture (SOA) is considered as the ideal software architecture for pervasive applications. Traditional Web services have been developed as pervasive services and service composition techonology also has been widely used in dynamic constructing pervasive applications. In pervasive space, however, the devices have different computing and storage ability and the network is in dynamic. Thus, traditional service composition technology cannot be applicated directly. In both acadimica and industry, researchers have widely investigated this problem, and gotten several good results. But none of them really achieved dynamic pervasive service composition which provids suitable service for user.For this,we did research on following subjects.(i) Ontology based context modeling and management systemAccording to the characteristic of context, several ontologies are builded to represent context. Ontology based context model is able denote the general characteristics of entities which is suitable to most of domain entities. It also can describe the personal characteristic through extending ontology. Furthermore, for the heterogeneous context source, pre-process mechanism to handle source information and converts the source information into the standard attributes in context ontology are proposed.(ii) Petri net based pervasive service process model and automatic service compositionBase on Petri net, pervasive process model is builded to describe the function of pervasive services. This model is able to describe pervasive service and such loosely coupled pervasive applications. This paper proposes automatic service composition mechanism based on this model, which can automatically generate service specification according to user's input, expected output and actions.(iii) Context based composite service path constructionThe heregeneous of pervasive service leads to the different performance of service with same function. For the composite service, system gets the best service policy in accordance with the context of both pervasive space and user, based on which the service instances are selected to construct composite service path. This service path can offer best performance for user.(iv) Mobile agent based composite service executionMobile agent based composite service execution mechanism which is a distributed service execution mechanism is presented. It can greatly reduce the data transformation and overcome the centrilized disadvantages. This paper further discusses the agent residing policy and the data transformation mechanism between agents, which can reduce the data transformation and prevent the multi-migration of agent, so that improve the system performance.Based on the study of context modeling and magament, this paper investigates the pervasive service composition and related technology and proposes context aware pervasive service composition mechanism which can dynamically provide services. This service composition mechanism provides composite services based on users' requirement and available services, and then it generates composite service path based on context, so that offers best composite service. Furthermore, traditional centrilized composite service execution engines are not effective and robost, this paper proposes mobile agent based composite service execution machenism to achieve composite services.