OptiGuard-GNN Cybersecurity Model: Leveraging Multi-Criteria Optimization and Graph Neural Networks for Enhanced Detection of Distributed Denial of Service Attacks

The escalating prevalence and sophistication of Distributed Denial of Service (DDoS) attacks necessitate advanced detection frameworks capable of delivering high accuracy while minimizing false alarms.This paper presents OptiGuard-GNN cybersecurity model, a novel cybersecurity model integrating a multi-objective, chaosenhanced particle swarm optimization (PSO) for dynamic feature selection with a graph-attention based ensemble classifier.By leveraging chaotic map initialization, multi-swarm cooperation, and adaptive parameter tuning, the proposed feature selection method optimizes detection accuracy, feature compactness, latency, and adversarial robustness concurrently.The classification model harnesses graph neural networks augmented with attention mechanisms, further combined in a stacked ensemble to capture complex feature interdependencies.Rigorous evaluation on the benchmark CSE-CIC-IDS2018 and CIC-IDS2017 datasets demonstrates near-ideal performance, achieving an outstanding accuracy of 99.986% alongside an exceptionally low false positive rate of 0.02%.These results affirm the model's effectiveness and resilience against evolving DDoS attack patterns, positioning OptiGuard-GNN cybersecurity model as a promising solution for real-time network security deployment.This research advances the frontier in intelligent intrusion detection by addressing critical challenges in scalability, adaptability, and robustness.