A Comparative Analysis of Deep Learning Approaches to Mitigate DDoS Attacks

Distributed Denial of Service (DDoS) attacks represent a critical challenge to modern network security, flooding systems with excessive traffic to disrupt service availability. This paper investigates the efficacy of three deep learning techniques-Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Convolutional Neural Network (CNN)—in identifying DDoS attacks using the CICDDoS۲۰۱۹ dataset. Our comprehensive evaluation reveals that GRU achieves superior performance, with perfect scores (۱.۰۰) in accuracy, precision, recall, and F۱-score, surpassing both LSTM and CNN. These findings emphasize GRU's potential as a robust tool for enhancing DDoS detection systems, offering insights into its practical deployment in cybersecurity frameworks.

کلیدواژه ها:

DDoS attacks ، LSTM ، GRU ، CNN ، deep learning