Reliability Prediction and Web Service Selection Using Soft Computing Techniques for Service-Oriented Systems

Abstract

Building a wide variety of distributed systems is a complex task these days. Since, service oriented architecture (SOA) is a major framework for distributed systems, it’s reliability is the major concern while developing a related software. The assessment of reliability in service-oriented systems (SOS) mainly depends on the accessibility of web-services, which leans on different parameters i.e. unpredictable internet, communication links and the location of web services. Hence, reliability needs to be predicted for the better functioning of a system. Selection of an optimal web-service is also an important concern in SOS. Since, for an abstract task to perform in SOS, a large number of functionally equivalent web service candidates are available. The same web service candidate can perform differently with different users. So, a technique is required for building the personalized web service ranking framework for designers. Hence, for predicting the reliability of SOS and for selection of an optimal web service candidate from functionally equivalent set of web service candidates a most effective approach is desired. In this work, a novel methodology is proposed for predicting the reliability of Web Service candidate which basically uses the past failure experience of similar service users and a personalized framework for selection of an optimal Web Service candidate from functionally equivalent candidates' set which basically is associated with the past Web-Service usage experience of similar users. In this work, no additional invocation of Web service is required. The experimental results are compared with many other techniques proposed by other authors in literature which shows the effectiveness of proposed approach.