Hybrid Deep Learning Neural Network for Optimizing Video Streaming Traffic Prediction over a Content Delivery Network

The Internet has transformed significantly in recent years, giving rise to innovative applications focused on digital content consumption. Advanced frameworks that integrate Content Delivery Networks (CDNs) and Artificial Intelligence (AI) have emerged to optimize content delivery and enhance user experiences. These frameworks use deep learning models to improve traffic prediction, bandwidth management, and network optimization for video streaming. The proposed system preprocesses video data and analyzes traffic patterns during peak hours, identifying underutilized channels and redirecting traffic to less congested routes. This ensures faster content delivery, reduces latency, and prevents bottlenecks during high-demand periods. A Hybrid CNN–LSTM deep learning model is designed to enhance traffic prediction for video streaming over CDNs. By combining the spatial learning capabilities of Convolutional Neural Networks (CNNs) with the temporal modeling strength of Long Short-Term Memory (LSTM) networks, the architecture captures complex spatiotemporal traffic patterns. The model accurately classifies network states, such as depletion, filling, and stalling, achieving a classification accuracy of 99.99%. Its strong generalization ability and low latency make it a scalable and intelligent solution for optimizing video content delivery and improving user experience in dynamic network conditions.