Distributed resource scheduling in edge computing: Problems, solutions, and opportunities

Abstract

Edge computing has become popular in the last decade and will advance in future to support real-time actionable analytics at the devices. One of the fundamental problems for future edge computing is to make distributed resource scheduling (DRS) decisions both at the end devices and edge devices to support requirements including autonomous computation, scalability, low-latency, etc. Several surveys in the literature on edge computing have considered some aspects related to DRS, such as challenges and solution approaches, particularly for computation offloading and data management. However, to the best of our knowledge, there is no comprehensive survey on DRS in edge computing. This paper surveys the challenging issues, motivations, and existing works for enabling DRS in edge computing. We define and identify the unique issues for DRS in edge computing compared to traditional works on parallel and distributed systems. The motivations for DRS in edge computing have been described by pointing out the benefits and emerging application scenarios. This paper also provides a taxonomy to classify the existing works from three perspectives, i.e., systems, problems, and solution approaches. Finally, we have outlined several future directions that can help researchers to advance the state-of-the-art.