Task offloading to edge cloud balancing utility and cost for energy harvesting Internet of Things

Abstract

Task offloading from many Internet of Things (IoT) devices to a mobile edge computing (MEC) system, consisting of multiple collaborative edge servers (CES), results in reduced task execution delay as well as energy consumption. However, the exploitation of computation resources of CESs incurs additional costs. Existing works in the literature either focused on minimizing execution latency or energy consumption. In this paper, we have developed a task offloading policy that aims at making a trade-off between device utility and execution cost. The utility is defined as a function of task execution latency and energy consumption of energy-harvesting IoT devices. The task offloading problem is formulated as a subset selection problem that makes the desired trade-off. The offloading problem is proven to be NP-hard and thus we develop a meta-heuristic approach, namely SCOPE, inspired by Social Cognitive Optimization (SCO) to obtain the desired polynomial time execution. The results show its potency compared to the state-of-the-art methods in terms of task execution latency, energy consumption, utility per unit cost, and task drop rate.