Service Composition In Pervasive Environments

Pervasive Computing is the fusion of two major research areas: distributedcomputing and mobile computing. Distributed computing requires Remote communication,robustness, high reliability and safety requirements, while mobile computing stressed"whenever" and "wherever" calculation. Service Composition in pervasive environmentencountered some unseen problems in traditional Internet environment:(1) context-awareand adaptive capacity, service composition should be "people-centered", that is to takefull advantage of the context information provided, and automatically re-configure andadjust the execution plan when the context changes;(2) robustness, it’s often difficult topredict the behavior of devices in the ubiquitous environment such as mobility andfailure, so the portfolio of services need to be able to capture this change, and can stillwork in abnormal circumstances or rapid recovery;(3)heterogeneous and efficiency,universal service portfolio needs to be able to span a variety of heterogeneous devices andcommunication. The demand of the service portfolio of the device computing capabilityshould be low, which requires Composition algorithm has a high efficiency.This paper propose an improved abstract service composition algorithm based onHierarchical Task Network(HTN) and Description Logic(DL) as well as a network-awareservice selection algorithm based on network model and genetic algorithm. These twoalgorithms together to complete the portfolio of services in the ubiquitous environment.The research work of the paper can be included as:(1) the abstract service composition algorithm based on HTN and DLThis paper extends HTN planning domain with DL which is called HTN-DL. Thedetailed definition of the concepts of HTN-DL are given and an improved HTN planningalgorithm is proposed. The paper propose algorithms that map OWL-S to HTN-DLdomain based on existing methods that map OWL-S to HTN domain.(2) network-aware service selection algorithm based on network model and geneticalgorithmsThere is little attention to the impact of the network on QoS-based service selectionalgorithm, in accordance with the existing network coordinate system, combined with thecharacteristics of the service portfolio, we proposed an improved genetic algorithm-basedservice selection algorithm.