Task Offloading under Hard Deadlines for Mobile Edge Computing Systems

Abstract

Considering a mobile edge computing system with randomly arriving tasks; Each of them is given a hard deadline constraint in a non-preemptive system. While previous works have proposed various solutions to maximize service ratios, they often overlook the delay in message exchange between users and servers to acquire information. In this context, a hybrid offloading policy is proposed, combining optimal scheduling techniques on a single server with a greedy algorithm. The policy aims to tackle these issues and achieve a high service ratio in the considered scenarios by accounting for the delay in message exchange between users and servers. Through performance evaluation, the proposed hybrid method is shown to be efficient compared to previous works, which may have significant disadvantages such as high computational complexity or lengthy training times.