Deep Learning-powered DDoS Attack Mitigation for Cloud Infrastructure

Distributed Denial-of-Service (DDoS) attacks severely threaten cloud infrastructures by compromising availability and reliability. This paper presents an optimized, ensemble deep learning model (CNN-LSTM hybrid) for DDoS detection and mitigation, evaluated on CICDDoS2019 and NSL-KDD datasets with in-depth validation, ablation, and case analysis. Real-world attack trends, advanced feature engineering, interpretability, and Python-based implementation are discussed. The framework demonstrates high accuracy, low false positive rates, and sub-second reaction times, making it highly suitable for operational cloud environments.