A NOVEL FRAMEWORK FOR THE NETWORK TRAFFIC ANALYSIS USING A CONTROLLER IN SOFTWARE-DEFINED NETWORKING

Abstract

The rapid growth of modern networks and diverse traffic patterns has highlighted traffic management as a core challenge in network administration. Traditional networks, with their rigid architectures and limited programmability, fail to meet the dynamic requirements of today’s applications. Software-defined networking (SDN) has emerged as a novel paradigm that decouples the control and data planes, enabling centralized control and intelligent network programmability. This thesis outlines a topology-aware intelligent network traffic analysis framework using the Ryu SDN controller for enhanced network performance and decision-making efficiency. A topology-aware SDN environment is designed using Mininet as the emulator and OpenFlow as the communication protocol. The proposed framework leverages the Ryu controller’s Python-based modular architecture to implement dynamic traffic analysis and adaptive flow management. Various network topologies are constructed to simulate diverse operational environments and evaluate the framework’s adaptability. The described SDN environment enables real-time monitoring of network parameters and flow optimization, ensuring effective data transfer under various traffic loads. Performance evaluation is conducted using key parameters, including latency, throughput, jitter, packet loss, and controller response time, across different network conditions. The obtained results indeed present a significant enhancement in network performance, as they generate up to a 22% gain in throughput and a 25% reduction in latency, along with decreased packet loss. Importantly, the comparative benchmarking confirms the performance robustness and scalability of the proposed SDN model, especially for more dynamic and larger topologies. As a result, this research contributes to the advancement of SDN-based network intelligence by combining topology awareness alongside traffic analysis and performance monitoring. The implications of this work lay the foundation for deploying efficient, scalable, and adaptable network management solutions applicable to real-world domains, such as cloud computing, and IoT-driven system.