Cooperative mixed strategy for service selection in service oriented architecture

In service oriented architecture (SOA), service brokers could find many service providers which offer same function with different quality of service (QoS). Under this condition, users may encounter difficulty to decide how to choose from the candidates to obtain optimal service quality. This paper tackles the service selection problem (SSP) of time-sensitive services using the theory of games creatively. Pure strategies proposed by current studies are proved to be improper to this problem because the decision conflicts among the users result in poor performance. A novel cooperative mixed strategy (CMS) with good computability is developed in this paper to solve such inconstant-sum non-cooperative n-person dynamic game. Unlike related researches, CMS offers users an optimized probability mass function instead of a deterministic decision to select a proper provider from the candidates. Therefore it is able to eliminate the fluctuation of queue length, and raise the overall performance of SOA significantly. Furthermore, the stability and equilibrium of CMS are proved by simulations.