Backhaul Bandwidth Consideration for Workload Placement in Hierarchical Edge Cloud Architecture

Abstract

Edge computing is expected to be one of the key technologies for fifth-generation mobile communication. It brings computing and storage resources to the edge of the network, enabling cloud computing capabilities and an information technology service environment with low-latency communication. However, when an edge server occupies the same facility as a base station, its available computation resources may be limited for physical reasons, such as restricted space and inadequate temperature control. To provide high-load services for mobile users, reference [1] proposed an edge cloud with servers distributed in a hierarchical topology. The proposed architecture can provide high-load services by using a hierarchical collection of edge servers, but the communication delay increases when a requested service is processed by a server at a significant functional distance from the user. The increased delay affects the service quality because almost all services handled by edge servers will be delay-sensitive. To decrease the completion delay of such a task, we propose a method of workload placement that considers both computation and communication delays. The workload placement determines which edge server will process a requested task in the hierarchy architecture for a mobile user. Our method consists of a procedure to determine the workload placement and to exchange key information for that determination. In this paper, we present our proposed method and its evaluation by computer simulation.