A Q-learning-based Approach for Optimizing Workflow Migration in Fog Environments
Nour El Houda Boubaker, Karim Zarour, Nawal Guermouche, Djamel Benmerzoug · 2023
The Fog Computing architecture is designed to meet the increasing demands of future Internet services and applications by providing computing resources in close proximity to users, with a primary focus on achieving low latency. However, as users move constantly between different access points, the need for service migration arises to maintain consistent service levels and ensure a high-quality user experience. The dynamic nature of the Fog environment presents challenges in managing service migration effectively, as services may need to be relocated to guarantee a consistent Quality of Service. However, frequent service migration can introduce delays and overhead, which can potentially have a negative impact on the overall system performance. In this study, we concentrate on addressing the issue of frequent service migration in Fog computing to minimize service delay and energy consumption. To effectively tackle this challenge, we propose a novel algorithm based on reinforcement learning. Specifically, we employ the Q-learning technique to ensure optimal decision-making for workflow migration. Through extensive evaluations, our proposed algorithm consistently outperforms other strategies, showcasing its superiority in handling the identified problem.