Distributed Denial of Service Attack Detection in Network Traffic Using Deep Learning Algorithm.

Area of Science:

• Computer Science
• Cybersecurity
• Artificial Intelligence

Background:

• Internet security is a growing concern with increased IT and cloud computing adoption.
• Traditional security measures struggle against sophisticated Denial of Service (DoS) and Distributed Denial of Service (DDoS) attacks.
• Deep learning offers advanced capabilities for detecting complex network traffic attacks.

Purpose of the Study:

• To evaluate deep learning models for detecting DDoS attacks.
• To compare the performance of Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), and Gradient Recurrent Unit (GRU) models.
• To analyze detection accuracy and execution time on recent datasets.

Main Methods:

• Utilized deep learning models: RNN, LSTM, and GRU.
• Tested models on the CICDDoS2019 and CICIDS2017 datasets.
• Conducted a comparative analysis of model performance, focusing on accuracy and execution time.

Main Results:

• All evaluated deep learning models achieved high accuracy (0.99) on the CICDDoS2019 dataset.
• Gradient Recurrent Unit (GRU) demonstrated significantly reduced execution time compared to RNN and LSTM.
• Comparative analysis confirmed the efficacy of deep learning for DDoS detection.

Conclusions:

• Deep learning models, particularly GRU, provide accurate and efficient solutions for DDoS attack detection.
• The study contributes a competent method for enhancing network security against sophisticated attacks.
• Findings suggest GRU as a preferred model for DDoS detection due to its speed and accuracy.