Task offloading in MEC systems interconnected by metro optical networks: A computing load balancing solution

Abstract

Multi-access edge computing (MEC) is a promising technology to satisfy end users’ ever-increasing demand for low latency. By building small-scale cloud infrastructures at the network edge, tasks from end users can be offloaded to geographically nearby edge servers to speed up processing. However, since the computing resources of each edge server are limited, “the nearest is the best” may not apply in some cases. The task may be blocked due to insufficient computing resources if the nearest edge server is heavy-loaded or overloaded, especially in peak times. Imbalanced computing load may affect user experience and degrade system performance. Hence, this paper integrates computing load balancing with task offloading and studies how to reasonably select the target edge server for each task and find the routing path to the target edge server by considering the load distribution of edge servers. The aim is to minimize the completion delay while balancing the global computing load. Two heuristic algorithms are designed for the computing load balancing task offloading (CLB-TO) scheme, termed the transmission delay-based CLB-TO (TD-based CLB-TO) and the genetic algorithm-based CLB-TO (GA-based CLB-TO). Simulation results show that the proposed algorithms achieve higher acceptance ratio and shorter completion delay than the other four alternative algorithms.