Quantum intrusion detection system using outlier analysis.

Area of Science:

• Cybersecurity
• Quantum Computing
• Machine Learning

Background:

• Current cybersecurity measures struggle to identify intruders amidst large network traffic volumes.
• Distinguishing legitimate traffic from Distributed Denial of Service (DDoS) attacks remains a significant challenge.

Purpose of the Study:

• To introduce a novel Quantum Machine Learning (QML) technique for enhancing security protocols in secure communication.
• To improve the accuracy and speed of detecting malicious network traffic, specifically DDoS attacks.

Main Methods:

• Utilizing quantum neural networks for improved detection accuracy and speed.
• Preprocessing network traffic data and encoding it into quantum bits via angle embedding.
• Employing outlier analysis, min-entropy, and quantum state fidelity to differentiate normal and abnormal network patterns.

Main Results:

• The proposed QML methodology demonstrated superior performance compared to conventional methods like AMM-CNN and ANN.
• Achieved a remarkable detection accuracy of 99.87% for DDoS attacks.
• Entropy measurement of network header data effectively identified security concerns.

Conclusions:

• The QML-based approach offers a more effective and secure solution for modern communication networks.
• This advancement significantly enhances the ability to detect and mitigate sophisticated cyber threats like DDoS attacks.
• Quantum Machine Learning holds substantial promise for the future of robust cybersecurity.