Efficient and Optimized Multimedia Routing using a Fully Connected Neural Network

Efficient multimedia routing is critical for meeting the Quality of Service (QoS) requirements of modern networks. Traditional routing algorithms often fail to dynamically adapt to the varying requirements of multimedia streams, leading to increased latency, packet loss, and suboptimal bandwidth utilization. This paper presents an optimized routing algorithm that leverages machine learning for dynamic multimedia stream classification and prioritization. A Fully Connected Neural Network (FCNN) model, trained on features extracted from packet content, classifies traffic into video, audio, combined, and subtitle categories. The routing algorithm dynamically updates paths based on classification results, ensuring optimal resource allocation by prioritizing higher-bandwidth routes for critical multimedia streams. The proposed approach is validated using a network emulator, where extensive experiments demonstrate significant improvements in throughput, latency, and packet delivery ratio compared to conventional routing protocols. The results highlight the potential of machine learning-based classification to enhance adaptive, real-time multimedia streaming, ensuring a more reliable and efficient network performance.