Edge Server Placement and Task Allocation for Maximum Delay Reduction

When edge computing is deployed for delay-sensitive applications such as autonomous driving systems and online gaming, it is important to reduce the maximum delay because real-time performance for all users must be ensured from Quality-of-Service (QoS) perspective. The primary delays in edge computing include network delay during data transmission and waiting time at the edge server. Since the waiting time at edge servers depends on server utilization, an increase in utilization bias leads to an increase in maximum delay. If a user is extremely far from the edge server, the network delay for that user will also increase. Conventional edge computing methods focus on reducing the average propagation delay of user-processing requests (tasks). However, these methods increase the utilization variance of each edge server, thus increasing the maximum delay. In this paper, we propose a method for determining both edge server placement and task allocation to reduce the maximum delay. Our method uses a genetic algorithm to optimize server utilization and the distance between users and servers. The maximum delay has been successfully reduced compared with that using conventional methods by simultaneously optimizing the server utilization and distance between users and servers.