Multi-Index Cooperative Mixed Strategy for Service Selection Problem in Service-Oriented Architecture

Under the environment of Service-Oriented Architecture (SOA), service users have to decide how to choose the proper provider from the candidates to obtain optimal service level. The existing approaches based on the optimization theories usually ignore the decision conflicts among users, which result in periodic QoS fluctuation and poor system performance. This paper treated the service selection problem as an inconstant-sum non-cooperative nperson dynamic game with incomplete information. The mechanism that generated the decision conflicts is discussed at first, and then Multi-index Cooperative Mixed Strategy (MCMS) with corresponding algorithm for finding its optimal solution is developed. Simulation shows that MCMS offers service users an optimal strategy that reduces the fluctuation of QoS and improves performance of SOA. Moreover, the stability and equilibrium of MCMS make it acceptable to most of the users, thus MCMS is able to improve predictability and controllability of SOA significantly.