Efficient Multisite Computation Offloading for Mobile Cloud Computing

Abstract

Mobile cloud computing is a promising approach to overcome limitations of mobile devices in terms of battery consumption and capability to execute computation-intensive applications. It consists in outsourcing some parts of computation-intensive applications to cloud servers, so that the overall energy consumption and execution time of application's running are optimized. In addition to classical public/private clouds, computations can also be offloaded to nearby clouds, called cloudlets, which may have different access delays, proximity to end users, computation capability and/or monetary cost. In such a heterogeneous context, the offloading decision becomes very complicated and time-consuming, especially when the number of surrogate sites and application units increase. In this paper, we propose a multisite offloading solution which uses the genetic algorithm to timely achieve the best-possible offloading outcome in a heterogeneous multisite context. Simulation results demonstrated that our proposal can outperform other existing multisite alternatives for different number of surrogate sites and various application sizes.