SDN-OptiNet: An Artificial Intelligence-Driven Framework for Real-Time Network Optimization Using Software-Defined Networking

The method of enhancing performance, reliability, and system efficacy in communication networks by effectively handling resources, including traffic flows and routing policies, is known as network optimization. Several machine learning approaches often failed to provide real-time adaptability; hence, to overcome all these limitations, this study introduces an artificial intelligence-driven framework for real-time network optimization using software-defined networking. An ensemble model is designed by integrating isolation forest, one-class support vector machine, and an autoencoder to enhance robustness against abnormal traffic patterns that lead to performance degradation. The proposed mechanism designs a graph neural network-based digital twin with reinforcement learning to attain fine-grained real-time optimization under varying network loads, topologies, and attack scenarios. The integration of the software-defined networking control plane with optimized policies ensures scalability, automated enforcement, and reduced operational complexity. Python-powered backend with a Django-based web interface is utilized for real-time monitoring and visualization, gradually increasing performance metrics, such as latency, jitter, packet loss, and throughput, ensuring practical usability. Extensive validation on DDoS Detection and Mitigation, SDN ID, and CSE-CIC-IDS2018 datasets with selected baselines proved that the proposed mechanism offers superior outcomes in throughput (1100 Mbps), latency (40 ms), packet loss (2.5%), and jitter (6.2 ms), highlighting that the proposed technique offers a scalable and context-aware solution for next-generation high-speed networks.