Adaptive Deep Learning Techniques for Real Time Detection of Hyper Volumetric DDoS Attacks in Cloud and High Speed Networks

Hyper-volumetric Distributed Denial of Service (DDoS) attacks have grown significantly in frequency and intensity, especially in high-speed network and cloud environments, overwhelming conventional intrusion detection systems. Deep learning (DL) provides adaptive and high-capacity models capable of recognizing complex traffic patterns, offering superior performance for DDoS detection compared with traditional methods. This paper proposes an adaptive deep learning-based framework for real-time detection of hyper-volumetric DDoS attacks in cloud and high-speed networks. The architecture integrates multivariate time-series modeling with GPU-accelerated deep neural networks to achieve high accuracy and low latency, validated through experiments using realistic high-volume data streams and benchmark traffic datasets. Results demonstrate that temporal models such as Long Short-Term Memory (LSTM) networks and Temporal Convolutional Networks (TCNs) can reliably detect and classify attacks under dynamic conditions. The study also examines feature importance using SHAP values to optimize performance and interpretability