Energy-Efficient Offloading for Mission-Critical IoT Services Using EVT-Embedded Intelligent Learning

Abstract

Mobile edge computing (MEC) is a promising technique to alleviate the energy limitation of Internet of Things (IoT) devices, as it can offload local computing tasks to the edge server through a cellular network. By leveraging extreme value theory (EVT), this work proposes a priority-differentiated offloading strategy that takes into account the stringent quality of service (QoS) requirements of mission-critical services and green resource allocation. Particularly, Lyapunov optimization is first introduced to derive an upper-bound queue minimization problem with the consideration of energy consumption and task priority. The peaks-over-thresholds (POT) model is then applied to evaluate the stationery status and cooperate with Wolf-PHC learning to optimize resource allocation. Finally, simulation results verify that the proposed offloading policy performs well in terms of its energy-saving capability while satisfying different demands of mission-critical IoT services.