Performance Comparison of Machine Learning-driven Hybrid Scheduling Algorithm in CIW Network

Introduction: This work aims to provide efficient traffic scheduling while considering different traffic priorities in a cloud-integrated wireless optical broadband access network (CIW). Method: The proposed approach utilizes a machine learning algorithm, K-Nearest Neighbor, to assign three levels of task prioritization. These prioritized tasks are then scheduled to cloudlets using a hybrid algorithm, enhancing overall network performance. Results: Simulation results compare the proposed approach with existing methods in terms of priority packet drops and packet delivery ratio, while maintaining consistent makespan (maximum completion time) and execution times. Discussion: In the CIW network, cloudlets play a critical role in providing cloud services with low latency and high energy efficiency. Efficient task scheduling is therefore essential to optimize processor utilization, maximize packet delivery ratio, reduce waiting time, and ensure that tasks are completed before their deadlines. Conclusion: The proposed machine learning–based hybrid traffic scheduling algorithm outperforms existing algorithms, such as Earliest Deadline First and First Come First Serve, in terms of network performance metrics.