Energy Efficient and QoS Aware Multi-Level Mobile Cloud Offloading

Abstract

Mobile devices are constrained in battery capacity and processing power to execute heavy computation tasks. So mobile cloud offloading is the process in which compute-intensive tasks are migrated from mobile devices to the remote cloud servers to reduce the execution time and the energy consumption on mobile devices. In this work, we consider the combination of direct task upload and task upload via cloudlet to minimize the overall energy consumption on mobile devices for executing the latency-sensitive tasks in the mobile cloud computing (MCC) environment. We propose integer linear programming (ILP) based formulation, machine learning based formulation, and standard simulated annealing based formulation for task uploading decisions. Based on our experiment, ILP and machine learning-based approaches outperform other approaches like the naive random approach and the greedy-based uploading approach. Also, when the tasks with tighter deadlines, random and greedy approaches fail to produce solutions to the uploading decision, whereas the ILP based approach, simulated annealing approach, ML-based uploading approaches are comparatively successful in producing solutions.