Research On The Key Technologies For Service Compositon Based On Trust-ware And Evolution

Service-oriented computing has become a mainstream computing model for complex distributed applications in an open heterogeneous environment. When single service cannot meet user needs, according to the shared context, service composition will compose a few function-limited Web services in accordance with the service description, constraints, available resources and services to form new, which meet the user functionality and produces value-added services. Thus, dynamic Web service composition has become the core technology for service-oriented computing, as the research focus in recent years.The quality of services (QoS) is a key factor for the service provider (SP) to win in the market. However, in an open network environment, how to ensure a high QoS service composition is facing many challenges. Therefore, this paper focuses on the theory and method of how to improve the QoS service composition for a more in-depth study, especially on the QoS service evaluation and selection strategies, trust-based QoS reasoning and evolution methods, as well as the strategy and method for related servies based on QoS and trust. Research work has been done as following.(1)A QoS service evaluation and selection strategy based on environment-ware is proposed. In previous studies, the service entity QoS evaluation result is often the weighted QoS value based on user-ware, rather than the actual QoS can be provided by current service entities. By using the QoS under different load conditions, the QoS which can be provided by the SP service entities is characterized, and a set of trust evaluation, trusty evolution services mechanisms is proposed to derive creditability service entities and QoS feature vector, combined with the current service load, the QoS provide d by current service entities can be obtained. A new service selection algorithm based on the above service QoS evaluation result is proposed, to enhance the service composition QoS better.(2)A web service composition strategy based on trust reasoning and evolution is proposed. To change the situation of trust missing and trust generalization in traditional trust reasoning, we propose new trust reasoning methods constrained by the trusty entities, the entity set, successive approximation of evaluation entities, etc. The new trust reasoning system could greatly enrich trust relationship among entities and distinguish conspiracy through deduction, inversion and recursion. It is capable of overcoming the problem aroused by the lack of direct trust and early stage trust in the traditional trust evolution. Therefore, it changes traditional methodology on trust relationship modeling. Based on the new trust reasoning and evolution method, a novel web service composition strategy is proposed, improving the success rate of service composition greatly.(3)A quick track service composition strategy based on credible link evolution is presented. Notes that in actual service composition, only by choosing a high QoS service will not be able to mix high QoS services composition. In fact, the service composition QoS is not only related to service QoS, but also has relationship between services' depending(matching) levels, and the service consumer (SC). In this paper, based on the interaction between the behaviors of main services, the service trust reasoning strategies is expanded in order to access and reveal the interdependence between the service entities, the quality and credible relationship of combined link and the service compositions. Thus, a quick track service composition strategy based on credible link is proposed, providing new ideas for credible quick track service composition.(4)An environment-aware particle swarm optimization algorithm (EAPSO) is presented. Services composition that can not adapt complicated variety in general PSO algorithm has slow convergence speed and strict speed demand. This paper presents an environment-aware particle swarm optimization algorithm (EAPSO) by simulating the bird flocking environment-aware behavior in seeking food process. EAPSO decreases unsteadiness resulting from random search and increases algorithm speed by method recollecting optimized population and increasing visual field. The simulation which is in the typical services composition stage shows that EAPSO is faster in convergence and more effective in global search.