Joint Computation and Traffic Loads Balancing Task Offloading in Multi-Access Edge Computing Systems Interconnected by Elastic Optical Networks

Multi-access edge computing systems powered by elastic optical networks can allocate bandwidth flexibly and meet end users' low-latency requirements. However, edge servers and optical links have limited resources. Sending tasks to local edge servers cannot always provide satisfactory computation service. A task may be blocked due to insufficient computing resources if the local edge server is heavy-loaded. A task may also be blocked due to inadequate spectrum resources even if the task is transmitted to another light-loaded edge server. In brief, imbalanced load distribution will affect the service availability and degrade the system performance. This letter proposes a joint computation and traffic loads balancing task offloading (JCTLB-TO) scheme to improve the acceptance ratio (AR), aiming to minimize the completion delay while balancing computation and traffic loads. A heuristic JCTLB-TO algorithm is proposed, where the link weight is introduced to optimize routing-path establishment and spectrum allocation. Simulation results show JCTLB-TO significantly outperforms the existing schemes in terms of AR.