FRCNC - An Enhancing Model for Classifying Packet Traffic at Internet Routers

In the modern networked world, it is a major challenge to control and categorise the traffic of the network, as per various applications. The networks are still evolving and are now offering various services like video streaming, VoIP, and enterprise data and so to manage the traffic effectively and therefore have optimization bandwidth, minimization of delay and Quality of Service (QoS). Traffic classification allows identifying and assigning network resources to per application type and enhances performance and reduced congestion. The paper suggests that one of the possible solutions is Federated Reinforcement Learning Convolutional Neural Networks Classification model (FRCNC). In model, Convolutional Neural Networks (CNN) extract specific features from network packet data, supporting the identification of traffic patterns. At the same time, Federated Reinforcement Learning (FRL) enhances the classification of network traffic by application. The method enables network routers to train and refresh a common model without sharing data and ensures privacy and does not overload central servers. Also, it adapts dynamically to network conditions on thresholds. FRCNC enhances efficiency of traffic classification and management, maximises QoS and distributes resources based on the application types and minimises network congestion. The model has a great potential in the management of future network systems, where the performance and security requirements are constantly becoming more and more complicated.