A FRAMEWORK FOR QoS AWARE WEB SERVICE SELECTION

Abstract

Service oriented computing is a paradigm for distributed computing that uses services as fundamental elements for developing applications. Some popular examples of services are Grid services and Web services. According to some researchers, the service paradigm can be extended to include hardware devices, network resources, a piece of code, and even a human being as a service. A service is associated with one or more input/output types. Service composition refers to the process of combining different simple services to obtain a relatively complex service that did not exist on its own. Web services are application software which can be remotely accessed through the Internet. A user’s request, in general, cannot be satisfied by a single (basic) web service, and so a composed web service is needed. QoS aware service composition takes into consideration QoS parameters (e.g., execution time, availability, reliability). Finding an optimal QoS aware composed service is a computationally intractable problem. Thus heuristic and meta-heuristic based algorithms have been suggested in the literature. We have found that the performance of the existing metaheuristic based algorithms for the problems considered in the thesis is not quite satisfactory. This thesis suggests some meta-heuristic based algorithms for web service selection that are modified versions of the existing algorithms. These include modified grey wolf optimizer (MGWO) algorithm, modified artificial bee colony (mABC) algorithm, and a new approach based on orthogonal array and fruit fly optimization algorithm (OL-FOA). To test the efficiency of the proposed approaches, a test suite of 2500 web services is chosen from the public repository. To establish the competitiveness of these approaches, we compared its performance against four other meta-heuristic approaches, namely, fruit fly optimization (FOA), differential evolution (DE), modified artificial bee colony algorithm (mABC), and global-best ABC (GABC). The empirical results show that the proposed approach outperforms its counterparts yielding a set of web services with better QoS values.