HERCULE: High-Efficiency Resource Coordination Using Kubernetes and Machine Learning in Edge Computing for Improved QoS and QoE

Edge computing (EC) environments are increasingly essential in ensuring low latency and high throughput for modern applications and in smart cities. Scheduling applications in EC environments should be designed to address challenges such as uneven workload distribution, high latency, and frequent request retransmissions, which are not well addressed by current state-of-the-art solutions. In this work, we propose a smart Kubernetes scheduling solution that embeds a machine learning model into the Kube-scheduler for more effective application deployment. By analyzing application keywords and edge server metrics, such as geographic location, workload, and keyword-based request frequency, the proposed solution dynamically selects the optimal edge server for application deployment. The simulation results depict substantial improvements, including reduced latency, better workload equilibrium, increased achievable throughput, and minimized retransmissions between servers. Compared to existing approaches, the proposed model provides a more robust solution to the complex and dynamic requirements of edge computing environments.