A machine learning approach for detecting WPA3 downgrade attacks in next-generation Wi-Fi systems.

Area of Science:

• Cybersecurity
• Wireless Network Security
• Machine Learning Applications

Background:

• Wi-Fi Protected Access 3 (WPA3) is critical for securing modern Wi-Fi and 5G&B Radio Access Networks (RAN).
• Wireless channels are vulnerable to downgrade attacks, forcing networks from WPA3 to WPA2 to exploit security flaws.
• Traditional Intrusion Detection Systems (IDS) lack adaptability to evolving network environments and sophisticated attacks.

Purpose of the Study:

• To develop and evaluate a hybrid adaptive machine learning approach for detecting downgrade attacks in WPA3 networks.
• To enhance the security of WPA3 networks against adversaries exploiting downgrade vulnerabilities.
• To introduce a novel ML-based Feature Selection and Thresholding for Downgrade Attacks Detection (MFST-DAD) approach.

Main Methods:

• A three-stage approach: traffic data preprocessing, adaptive feature selection, and real-time detection/prevention using ML algorithms.
• Utilizing machine learning algorithms for classifying traffic, selecting features, and setting adaptive thresholds.
• Experimental validation on a custom dataset to assess performance in WPA3 personal and enterprise transition modes.

Main Results:

• The proposed MFST-DAD approach achieved 99.8% accuracy in detecting downgrade attacks.
• A Naive Bayes classifier demonstrated high effectiveness within the MFST-DAD framework.
• Successful detection was confirmed in both WPA3 personal and enterprise transition modes.

Conclusions:

• The developed ML-based approach effectively detects downgrade attacks in WPA3 networks.
• The MFST-DAD method offers a significant improvement over traditional IDS for WPA3 security.
• This research confirms the viability of adaptive ML techniques for securing next-generation Wi-Fi systems.