Edge Servers Deployment Based on 5G Micro Base Stations

With the intelligent development of augmented reality, smart home and other applications, more and more applications pursue high bandwidth, low latency and wide connectivity, so mobile edge computing technology has attracted more and more attention. Mobile edge computing will deploy the edge server closer to the user's network edge, and the user will unload the task to the server for execution, so as to reduce the task execution time and the energy consumption of the local terminal. However, people often only pay attention to the core problem of mobile edge computing: computing offload. As the first step of mobile edge computing, server deployment has received little attention. Therefore, this paper studies the deployment of edge server, and proposes an algorithm to deploy edge server in 5G micro base station. The algorithm considers the computing time of tasks and the load balance of servers, reduces the access delay of tasks and balances the load between edge servers. Simulation results show that the deployment strategy is superior to other algorithms in terms of server load and task access latency.

I. Introduction

With the rapid development of computer and Internet of things, smart city gradually appears in our life. More and more mobile application tasks become intensive and complex, which puts forward higher requirements for terminal computing power and storage. Mobile cloud computing (MCC) [1] technology is proposed to solve this problem. MCC has powerful computing power and rich computing resources to solve the terminal intensive tasks. However, the distance between the cloud server and the terminal will cause large communication delay, and MCC adopts a centralized way to process tasks, and unloading all tasks to the cloud will lead to excessive cloud load [2]. To solve these problems, mobile edge computing (MEC) [3]–[4] technology came into being. MEC refers to the deployment of edge servers (ES) with computing and storage capabilities closer to the user's network edge, which can not only share the pressure of the cloud server, but also provide users with low latency and high reliability computing services. As one of the core technologies of 5G [5], MEC has been widely concerned by scholars. Because the coverage of 5G micro base station is far less than 4G, the reasonable deployment of ES brings lower delay and higher efficiency for the network intensive cellular network. However, at present, people's research focus is more on another important technology of MEC: computing offload. Although the deployment of edge server has great advantages in task execution, it has been ignored [6]. The current research on the deployment technology of edge server mainly focuses on the deployment of cloudlet [7]. Xu et al. [8] regarded the deployment of cloudlets as a formulaic integer linear programming (ILP). They proposed an effective heuristic algorithm to dynamically allocate user requests to different cloudlets. Chen et al. [9] designed efficient heuristic algorithm and clustering algorithm to minimize the access delay of tasks, and proved the effectiveness of the algorithm. In order to minimize the cost of deployment, a clustering algorithm is designed, and an effective heuristic algorithm is proposed to minimize the number of cloudlets. Liang et al. [10] proposed a k-means based location aware service deployment algorithm for cloudlets. They divide the mobile devices into multiple device clusters according to their geographical location, and then deploy the service instances to the edge cloud server nearest to the center of the device cluster.