The AI-enabled DDoS protection system faced multiple challenges. Attackers constantly evolved their methods, making static defenses ineffective. Continuous learning was required to maintain system efficiency.

Graph-based attack detection required significant computational resources. Processing large-scale graph structures without latency remained difficult. Reinforcement Learning (RL) struggled with real-time adaptation, as selecting mitigation strategies like traffic shaping needed instant execution without lag.

False positives reduced reliability. Rule-based systems misclassified legitimate traffic, and Explainable AI (XAI) methods like SHAP and LIME, though helpful, introduced performance trade-offs. Self-learning mechanisms demanded strict privacy controls, complicating data integration from multiple sources.

Traffic simulation required precise attack testing using tools like hping3 and Wireshark. However, maintaining system stability during testing remained a challenge. Collecting and labeling large datasets was labor-intensive and prone to errors, leading to inconsistent AI training.

AI bias and misclassification raised security concerns. Unbalanced datasets led to either blocking legitimate users or allowing malicious traffic. Ethical concerns regarding privacy and over-monitoring added complexity, requiring strong access controls.

Computational costs remained high. Running deep learning models like GNNs and RL frameworks required substantial processing power. Hardware constraints slowed real-time detection and response times. Scalability problems emerged as traffic volumes fluctuated, requiring efficient resource management.

These challenges were addressed with optimized