An edge computational offloading architecture for ultra-low latency in smart mobile devices

Computation offloading is a notable paradigm under Mobile Edge Computing (MEC) that significantly improves the performance of Smart Mobile Devices (SMDs) through the provision of computation, energy and storage services with the aid of edge servers. Recent studies under computation offloading pay less attention to SMDs and their fast-changing context conditions. Hence, this paper proposes, first, a novel Context-aware Computation Offloading (CaCO) architecture, particularly considering the execution time and battery consumption of SMDs when running resource-intensive tasks before proposing offloads. Secondly, an Efficient Genetic Algorithm (EGA) is presented to obtain the optimized solution for the formulated task allocation NP-hard problem in accessible time complexity. Extensive experiments conducted with real Android SMDs and simulation results compared with other baseline algorithms show that the proposed algorithm is superior in performance and could effectively reduce energy consumption and task completion latency.