Intelligent task scheduling in edge-cloud computing: A metaheuristic rule-based framework
Abstract
Edge computing offers significant advantages such as lower latency, higher bandwidth and energy savings by reducing dependency on remote data centres. However, to provide these advantages effectively, task scheduling strategies in edge computing environments need to be designed efficiently. Task scheduling is a challenging optimization problem classified as NP-Hard with solution time varying depending on the problem size. Metaheuristic methods are widely used in solving such complex problems and produce successful results. However, the iterative nature of these methods leads to delays in the solution process, and given the limited resources of edge computing environments, this makes problem-solving even more difficult. In this work, we suggest a rule-based approach that combines metaheuristic algorithms and machine learning techniques for the task scheduling problem. In the proposed method, different scenarios are generated and rules are extracted from these scenarios. In the scenario generation phase, Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO) algorithms are used and the most appropriate solutions are evaluated in the rule-based model. In the implementation phase, task scheduling is performed using the extracted rules. The primary benefit of the suggested approach is that it significantly reduces the time and energy costs that metaheuristic algorithms spend in the solution generation process. Experiment results indicate that the suggested approach is 25 times faster than PSO and 27 times faster than ACO. Moreover, in terms of task scheduling performance, the proposed method achieves a 17% better makespan than PSO and a 20% better makespan than ACO. Experiments considering different scenarios show that the proposed method provides high efficiency and effectiveness.