Improving the performance of data center with real-time service image placement in mobile cloud environment

Abstract

Mobile Cloud Computing (MCC) is broadening the ubiquitous market for mobile devices. Because of the hardware limitation of mobile devices, the heavy computing tasks should be processed by service images (SIs) on the cloud. Due to the scalability and mobility of users and services, dynamic resource demands and time-varying network condition, SIs must be re-located to adapt the new circumstances. In this paper, we formulate the SI placement as an optimization problem which minimizes the communication cost subject to resource demand constraints. We then propose a real-time SI placement scheme which includes two sequent stages of clustering/filtering and condensed placement to solve the formulated problem. The former omits the infeasible slots prior to placement in order to improve computational complexity. The latter focuses on the SI placement through a novel condensed solution. The numerical results show that our solution converges to the global optimum with a negligible gap while performing much faster execution time compared with the exhaustive search method. This improvement leverages the real-time services especially in MCC environment.