QoS Broker based Architecture for Dynamic Web Service Discovery and Composition

With the widespread proliferation of web services, the demand on highly reliable and available services increases to satisfy business and personal needs. As the number of functionally similar web service increases on the Internet, Quality of Service (QoS) based web service selection and composition has gained much attention. However, the existing architectures for runtime service composition requires discovery of all functionally similar services, which consumes too much time. Instead, it can be possible to discover only those services which are highly reliable and thus included in composition. This issue can be solved by introducing a broker based architecture for automated dynamic web service discovery and composition in which the composition operation accepts as input a set optimized services along with their QoS specifications. The proposed architecture is able to reduce the composition time by performing optimized selection of services using Local Selection and Local Optimization (LSLO) approach during service composition. The architecture is fault tolerant and ensures improved reputation prediction accuracy of selected services using Mixed Opinion based Reputation Prediction (MORP) approach. In this paper, a comparative analysis is also performed on the basis of set of criterion to analyze the existing dynamic web service discovery and composition architectures for their strengths and weaknesses.