A Brief Review of Population-based Methods for Task Offloading in Cloud-to-Edge Continuum.

Athanasios Chourlias,Theodoros Theodoropoulos,John Violos,Aris Leivadeas,Konstantinos Tserpes, Christos-Kyprianos Zalachoris

IEEE International Conference on Cloud Networking(2023)

引用 0|浏览3
Task offloading is one of the most important issues in the cloud-to-edge continuum in order to optimise quality of services metrics and maximize the revenue for users and providers. While the literature includes multiple mathematical, machine learning and control theory models each of them have its limitations such as the lack of a mathematical formulation or the requirement of historical training dataset. Population-based methods including swarm intelligence and evolutionary computation algorithms have many advantages that make them prominent meta-heuristic solutions into many combinatorial problems. Even if various population-based methods have been proposed for specific orchestration tasks in the distributed computing field, there is not any paper that presents and experimental compares them. In this paper we provide a comprehensive overview of the most popular population-based methods and discuss their pros and cons for their applicability for the task offloading challenge. We also make experiments with Google Cluster Dataset and conclude that particle swarm optimization significantly surpasses the other methods.
Task Offloading,Population-based Methods,Machine Learning,Service Quality,Machine Learning Models,Evolutionary Algorithms,Particle Swarm Optimization,Intelligence Algorithms,Evolutionary Computation,Swarm Intelligence,Swarm Intelligence Algorithms,Energy Consumption,Response Time,Performance Metrics,Resource Utilization,Cloud Computing,Comparative Experiments,Fitness Function,Local Optimum,Network State,Processing Nodes,Edge Server,Edge Computing,Edge Devices,Task Scheduling,Candidate Solutions,Prey Behavior,Swarming Behavior,Total Execution Time,Decision-making Dynamics
AI 理解论文
Chat Paper