Study On Web Service Composition On The Basis Of QoS Multi-attributes Decision Making

In recent years, with the development of service-oriented computing, such as SOA, grid computing and cloud computing, the fundamental unit: Web service is becoming more and more importantly and also being a hot research focus. However, a single service can provide only a limited function, if users mean to build a complicated upper application or they want a one-station service experience, the composition of services would be a must. As the service with same function but different in quality proliferated, the problem of selecting a set of services which has the optimal quality from numerous candidate service composition plans becomes the most critical issue in the study of Web service composition.The quality of a Web Service(QoS) often presents a multi-dimensional and contradictory characteristic. Multi-attribute decision making(MADM) theory addresses the problem of making an optimal choice that has the highest degree of satisfaction from a finite set of discrete alternatives characterized by multiple and conflicting attributes. Therefore, in this paper, the QoS based Web service composition(QWSC) problem is modeled as a MADM problem and the compromise ratio method(CRM) is chosen for its superiority over other MADM methods to solve the MADM model. However, traditional MADM methods couldn't handle a problem with numerous alternatives. Therefore, a heuristic algorithm(Genetic Algorithm based Compromise Ratio Method, GACRM) is proposed to solve the above issue. In combination with the advantage of CRM(Compromise Ratio Method, CRM) in terms of ranking alternatives, together with the superiority of GA(Genetic Algorithm, GA) in terms of global search, GACRM is capable of finding an approximate optimal solution from a massive search space. Experimental results demonstrate that the proposed algorithm resembles the basic CRM in terms of ranking order of best alternatives, whereas outperform the basic CRM in terms of time cost. Moreover, it shows a scalable performance for large-scale QWSC problem.The proposed innovative algorithm can not only guarantee the quality of the composite service, but also can meet users' preferences and respond to a large-scale service composition request in a timely manner. Thus its broad applicability can be seen.