Selfish-Aware and Learning-Aided Computation Offloading for Edge-Cloud Collaboration Network

Mobile-edge computing (MEC) raises the problem of selfish user devices that utilize less computing resources than expected to execute offloading tasks or maliciously discard computation tasks. However, most of the existing work either focused on the task offloading or concentrated on the trust mechanism in MEC systems. By jointly considering the two challenges, in this article, we propose a selfish-aware and learning-aided computation offloading scheme for edge-cloud collaboration network. Specifically, we first design a selfishness evaluation mechanism to evaluate the selfishness of the user devices based on the historical interaction records of the edge-cloud collaboration network. Then, we construct the task offloading model which introduces the selfishness evaluation mechanism to suppress the selfish user devices. On this basis, we further formalize the selfish-aware task offloading as an optimization problem of the weighted sum of time latency and energy consumption. Thereafter, we take one step further formalizing the optimization problem as a Markov decision process (MDP) and design a task offloading algorithm based on deep reinforcement learning (DRL) to find the optimized task offloading decision. The simulation results demonstrate that our work can decrease the time latency and energy consumption as well as suppress the selfish user devices.