Balancing Loads among MEC Servers by Task Redirection to Enhance the Resource Efficiency of MEC Systems

• Main Authors: Jaesung Park, Yujin Lim
• Format: Article
• Language: English
• Published: MDPI AG 2021-08-01
• Series: Applied Sciences
• Subjects: task redirection; load balancing; lexicographically minimum; resource efficiency; distributed consensus
• Online Access: https://www.mdpi.com/2076-3417/11/16/7589

To improve the resource efficiency of multi-access edge computing (MEC) systems, it is important to distribute the imposed workload evenly among MEC servers (MECSs). To address this issue, we propose a task redirection method to balance loads among MECSs in a distributed manner. In conventional methods, a congested MECS selects only one MECS to which it redirects tasks. By contrast, the proposed method enables a congested MECS to distribute its tasks to a set of MECSs, the loads of which are lower than that of the congested MECS by determining the number of tasks that it redirects to each selected MECS. We prove that our task redirection method drives a MEC system to a state where the resulting MECS load vector is lexicographically minimal. Through extensive simulation studies, we show that compared with the conventional methods, the proposed method can achieve the smallest load difference between the load of the MECS, the load of which is the highest, and that of the MECS, the load of which is the smallest. By lexicographically minimizing the MECS load vector, the proposed method decreases the average task blocking rate when the task offload rate is high. In addition, we show that the proposed method outperforms the conventional methods in terms of the number of tasks, the delay requirements of which are not satisfied.