DDoS Attack Mitigation in Cloud Networks Using Hybrid Metaheuristic and Machine Learning Framework

Abstract

Cloud computing has become the backbone of modern enterprise infrastructure due to its scalability and cost-efficiency. However, this centralized nature makes it a prime target for Distributed Denial of Service (DDoS) attacks, which aim to exhaust network resources and render services unavailable. Traditional detection mechanisms, such as static firewalls and standalone Machine Learning (ML) algorithms, often struggle with the high dimensionality of network traffic data, leading to high False Positive Rates (FPR) and substantial detection latency. To address these challenges, this paper proposes a novel hybrid framework that integrates Grey Wolf Optimization (GWO) with a Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) deep learning model. The GWO algorithm is utilized as a wrapper-based feature selection technique to eliminate redundant features, thereby solving the curse of dimensionality. Subsequently, the CNN-LSTM architecture captures both spatial and temporal features of the traffic flows for accurate classification. Experimental evaluation was conducted using the benchmark CIC-DDoS2019 dataset. The results demonstrate that the proposed hybrid model achieves an accuracy of 99.2% and reduces detection latency by 14% compared to standard Random Forest (RF) and standalone Convolutional Neural Networks (CNNs) models. These findings suggest that bio-inspired optimization combined with deep temporal learning provides a robust defense mechanism for securing cloud environments against evolving DDoS threats.