An energy-efficient task migration scheme based on genetic algorithms for mobile applications in CloneCloud

The limitations of mobile devices have attracted researchers to work out an energy-efficient mechanism to enhance user experience. The emerging mobile cloud computing (MCC) provides a new approach to solve this problem. Some parts of mobile applications, i.e., heavy computational tasks, are migrated to remote servers with powerful computational resources, which can improve the performance of mobile devices. This paper focuses on a popular MCC architecture, CloneCloud, and constructs a scheduling problem of task migration as a constrained stochastic shortest path problem in a directed acyclic graph. And then it designs a scheduling algorithm based on genetic algorithm to obtain the optimal task migrations. A user flexibly makes migration decisions through its own mobile device and migrates some tasks to the clone in CloneCloud without any change of application codes. Furthermore, this scheme facilitates mobile devices to collaboratively process computational applications. Real testbed experiments in Android smartphone demonstrate that the smartphone is able to save at most 59.42% energy within a time constraint by using the proposed task migration scheme.