QoS-Aware Cloud Service Composition via an Enhanced Multi-Agent Bird Swarm Algorithm

Abstract

The disruptive spread of IT based delivery across the internet, with profound implications in terms of Big Data has made it very difficult to make a good and smart choice on the most suitable service when lots of alternative services provide services offerings that are similar but have different Quality of Services (QoS) settings. Service composition combining a set of individual services in order to create new more complex value-added services has thus arisen as a major approach. But the major issue is how to deal with uncertainties of QoS due to dynamic environment, which results in changes in salient characteristics such as response time, availability and cost. Following on the work of developing a multi-Agent swarm based algorithm, this paper presents an improved and adaptive solution to web service Composition: Multi-Agent Bird Swarm Algorithm (MA-BSA). Adopting agent-based principles, each candidate service is abstracted as an independent agent collaborating to search the solution space to find optimal compositions according to user specified QoS preferences. A composite function, i.e. Weighted Aggregation function resolves fears QoS attributes into a common fitness value score towards a balanced and versatile service selection. We demonstrate our approach on synthetic and real-world datasets which consist of over 52,000 web services. Experimental results show that MA-BSA not only enhances compositional quality but is also more computationally efficient than leading existing methods. This extension of our earlier work represents a much smarter and more scalable way to do service composition in dynamic and uncertain domains.