Résumé: The rapid development of internet of things (IoT) gadgets and the increase in the rate of sending requests from these devices to cloud data centers resulted in congestion and consequently service provisioning delays in the cloud data centers. Accordingly, fog computing emerged as a new computing model to address this challenge. In fogging, services are provisioned at the edge of the network using devices with computing and storage capabilities, which are located through the way to connect IoT devices to cloud data centers. Fog computing aims to alleviate the computing load in data centers and cut the delay of requests down, notably real-time and delay-sensitive requests. To achieve these goals, vitally important challenges such as scheduling requests, balancing the load, and reducing energy consumption, which affects performance and reliability in the edge-fog-cloud computing architecture, should be considered into account. In this paper, a reinforcement learning fog scheduling algorithm is proposed to address these challenges. The experimental results indicate that the proposed algorithm raises the load balance and diminishes the response time compared to the existing scheduling algorithms. Additionally, the proposed algorithm outperforms other approaches in terms of the number of used devices.

Langue: Anglais