A Dynamic Edge Caching Framework for Mobile 5G Networks

Mobile edge caching has emerged as a new paradigm to provide computing, networking resources, and storage for a variety of mobile applications. That helps achieve low latency, high reliability, and improve efficiency in handling a very large number of smart devices and emerging services (e.g., IoT, industry automation, virtual reality) in mobile 5G networks. Nonetheless, the development of mobile edge caching is challenged by the decentralized nature of edge nodes, their small coverage, limited computing, and storage resources. In this article, we first give an overview of mobile edge caching in 5G networks. After that, its key challenges and current approaches are discussed. We then propose a novel caching framework. Our framework allows an edge node to authorize the legitimate users and dynamically predicts and updates their content demands using the matrix factorization technique. Based on the prediction, the edge node can adopt advanced optimization methods to determine optimal content to store so as to maximize its revenue and minimize the average delay of its mobile users. Through numerical results, we demonstrate that our proposed framework provides not only an effective caching approach, but also an efficient economic solution for the mobile service provider.

Introduction

Mobile Edge Caching (MEC) is an emerging architecture concept that provides storage and computing capabilities at the “edge” of 5G networks. The basic idea of the MEC network is to “move” the edge servers closer to the mobile users, thereby reducing service delay/latency for customers while mitigating network congestion. With over five million cell towers currently on earth, it is anticipated that MEC can save mobile network operators up to 35 percent on backhaul usage, and the latency in mobile services can be reduced by 50 percent [1]. However, the development of MEC networks faces many challenges, especially in managing cache resources. Due to the special characteristics of the MEC network, for example, small coverage and limited computing and storage resources, conventional cache management approaches may become ineffective. For example, typical cache management schemes rely on global network information about content demand, which is rarely available in the distributed environment of the MEC network. Thus, there is an urgent need for innovative methods to deal with this problem.