Article Published In Vol.8 (Jan-Feb-2020)

AI Solutions for SDN Routing Optimization using Graph Neural Networks in Traffic Engineering

Author : Sri Harsha Grandhi and G. Arulkumaran

Network routing optimization is crucial for improving the efficiency and performance of communication systems in dynamic environments. The primary objective of this paper is to address the challenges of network congestion, fluctuating traffic, and security threats, including DDoS attacks, by leveraging advanced machine learning techniques. The proposed work introduces a hybrid approach that integrates machine learning methods to dynamically optimize routing decisions in real-time, adapting to changing network conditions. This method combines various techniques to handle traffic behavior, congestion points, and anomalies, ensuring efficient routing and reducing delays. The results show that the proposed approach achieves 92% classification accuracy, 88% precision, 85% recall, and 89% F1-Score, indicating strong performance in traffic classification and decision-making. However, challenges such as high computational overhead, increased latency, and vulnerabilities to adversarial attacks remain, pointing to the need for further improvements. The study also explores federated learning to reduce computational burdens, enhance system responsiveness, and improve scalability, suggesting a pathway for more efficient and secure network routing solutions in real-world applications.

Keywords: Software-Defined Networking, Routing, Random Forest. Ant Colony Optimization, Machine Learning, DDoS Attack Mitigation, Congestion Control