Routing Planning for Video Transmission in Cloud Content Delivery Networks Based on Q-Learning

In the cloud content delivery networks, when an edge CDN node lacks the video requested by a user, it needs to send video requests to the origin server or other edge CDN nodes. To enhance user experience quality, the target node that receives the video request needs to establish a low-latency video transmission path to send videos to the requesting node. However, existing video transmission strategies have not fully considered the dynamic network congestion status within the cloud content delivery network. Therefore, this paper proposes a Q-learning based Adaptive Video Routing algorithm (Q-AVR) specifically for video transmission issues within cloud content delivery networks. It aims to reduce end-to-end video transmission latency and improve network bandwidth utilization. In the video transmission path construction process, edge CDN nodes exchange information through sending data packets. Each node stores the information in a Q-table and makes routing decisions based on the Q values. This algorithm optimizes end-to-end video transmission by learning network status in real-time. After simulation validation, the results show that the Q-AVR algorithm can effectively reduce end-to-end video transmission latency and improve network bandwidth utilization.