DL-Based Physical Tamper Attack Detection in OFDM Systems with Multiple Receiver Antennas: A Performance-Complexity...

Area of Science:

• Electrical Engineering
• Computer Science
• Signal Processing

Background:

• Orthogonal frequency division multiplexing (OFDM) systems are vulnerable to physical tamper attacks, such as unwanted antenna orientation changes.
• Detecting these attacks is crucial for maintaining system integrity and security.
• Existing methods may struggle with environmental variations and complexity trade-offs.

Purpose of the Study:

• To propose novel deep learning (DL)-based approaches for physical tamper attack detection in multi-antenna OFDM systems.
• To address the challenges of environmental changes and detection performance versus complexity.
• To validate the effectiveness of the proposed methods in diverse real-world scenarios.

Main Methods:

• Developed two semi-supervised anomaly detection algorithms utilizing channel state information (CSI) estimates.
• Trained algorithms exclusively on tamper-attack-free data for general applicability.
• Evaluated performance considering environmental factors and computational complexity.

Main Results:

• Experimental validation in office and hall environments demonstrated effective tamper attack detection.
• The optimal proposed method achieved a 93.32% true positive rate and a 10% false positive rate.
• The methods showed proper detection performance across different complexity levels.

Conclusions:

• The proposed deep learning approaches offer a robust solution for physical tamper attack detection in OFDM systems.
• The methods balance detection accuracy with manageable complexity, outperforming traditional approaches.
• These findings have significant implications for enhancing the security of wireless communication systems.