An Intelligent Intrusion Detection System for DDoS Attacks Using Deep Neural Networks

Abstract The emergent dependence on internet-based facilities highlights the exigent need for strong network security, especially in alleviating Distributed Denial-of-Service (DDoS) outbreaks, which seriously disrupt service accessibility and cause significant financial losses. DDoS attacks devastate targeted systems with large volumes of traffic from numerous sources, resulting to downtime and performance dilapidation. Prompt recognition of such attacks remains a serious challenge in cybersecurity. Existing methods often suffer from high false positive rates and inadequate capability to detect the various and complex traffic patterns related with contemporary DDoS attacks, resulting in limited accuracy. This research work presents an enhanced intrusion detection framework leveraging deep learning techniques for effective identification of DDoS attacks. Three architectures Deep Neural Network (DNN), Convolutional Neural Network (CNN), and Gated Recurrent Unit (GRU) were used on the CICDDoS2019 dataset sourced from Kaggle. Comparative evaluation shows that the CNN model attained higher performance, exhibiting an accuracy of 99.73%, precision of 99.70%, recall of 99.85%, and F1-score of 99.77%. These outcomes reveal CNN’s ability to effectively distinguish between benign and harmful outbreaks while reducing false positives and false negatives. The outcomes validate the effectiveness of deep learning, especially CNN-based models, in exhibiting extremely accurate early exposure of DDoS outbreaks, thus improving network resilience against emerging cyber threats. Keywords Distributed Denial-of-Service (DDoS), Deep Learning, Intrusion Detection System (IDS), Convolutional Neural Network (CNN), Network Security, Cybersecurity