A Non-Dominant Solution for Bi-objective Web Service Composition Problem in Multi-Cloud Environment

Abstract

81 ABSTRACT: A single cloud cannot meet all types of complicated business process requirements for the sake of vender lock-in, lack of fault tolerance and security uncertainty. Therefore, multi-cloud environment (MCE) is a promising solution in competitive market. Nevertheless, uncertainty relevant to security principles such as availability, integrity and confidentiality is the main challenge in this ambit. Virtualization as a key technology is pervasively being exploited in datacenters; so, variety known/unknown cybersecurity attacks threat mutli-tenant cloud datacenters. On the other hand, each service provider publishes same web services with different pricing scheme and different security level. We propose a framework in which every cloud can register its pricing scheme and its cybersecurity threats on a data warehouse via its broker module. Later, each security level can be measurable by our novel approach. Then, the web service composition problem is modeled to multiobjective optimization by monetary cost and cybersecurity viewpoints. We design and implement a multi-objective genetic approach to combine web service composition based on user cost and security requirement. Our proposed method well constitutes Pareto curve by finding non-dominance points. Simulation on different scenarios proves our approach dominance against a single objective heuristic ones.