Computation offloading in mobile edge computing networks: A survey

Computation offloading is one of the key technologies in Mobile Edge Computing (MEC), which makes up for the deficiencies of mobile devices in terms of storage resource, computing capacity, and energy efficiency. On one hand, computation offloading of task requests not only relieves the communication pressure on the core networks but also reduces the delay caused by long-distance data transmission. On the other hand, emerging applications in 5/6G also rely on the computation offloading technology for efficient service provisioning to users. At present, the industry and academia have conducted a lot of researches on the computation offloading methods in MEC networks with a diversity of meaningful techniques and approaches. In this paper, we present a comprehensive survey of the computation offloading in MEC networks including applications, offloading objectives, and offloading approaches. Particularly, we discuss key issues on various offloading objectives, including delay minimization, energy consumption minimization, revenue maximization, and system utility maximization. The approaches to achieve these objectives mainly include mathematical solver, heuristic algorithms, Lyapunov optimization, game theory, and Markov Decision Process (MDP) and Reinforcement Learning (RL). We compare the approaches by characterizing their pros and cons as well as targeting applications. Finally, from the four aspects of subtasks dependency and online task requests, server selection, real-time environment perception, and security, we analyze the current challenges and future directions of computation offloading in MEC networks.

Automated summary

This paper presents a comprehensive survey of computation offloading in Mobile Edge Computing (MEC) networks, covering applications, offloading objectives, and offloading approaches. The key issues in offloading objectives, such as delay minimization, energy consumption minimization, revenue maximization, and system utility maximization, are discussed. The methods to achieve these objectives, including mathematical solver, heuristic algorithms, Lyapunov optimization, game theory, and Markov Decision Process (MDP) and Reinforcement Learning (RL), are compared. Finally, the current challenges and future directions of computation offloading in MEC networks are analyzed from various aspects.