Detecting Malicious False Frame Injection Attacks on Surveillance Systems at the Edge Using Electrical Network...

Area of Science:

• Cybersecurity and Digital Forensics
• Electrical Engineering and Signal Processing

Background:

• National critical infrastructure relies heavily on video surveillance for security and anomaly detection.
• False Frame Injection (FFI) attacks pose a significant threat by replaying old frames to mask live feeds.
• Existing video analysis methods for FFI detection are computationally intensive, hindering real-time, on-site deployment.

Purpose of the Study:

• To investigate the feasibility of FFI attacks on edge-compromised surveillance systems.
• To propose and evaluate an effective technique for detecting injected false video and audio frames using Electrical Network Frequency (ENF) signals.

Main Methods:

• Monitoring surveillance feeds for embedded Electrical Network Frequency (ENF) signals, which are stable across power grids.
• Utilizing the time-varying nature of ENF as a forensic tool for authenticating surveillance feeds.
• Collecting ENF signals from power grids to create a reference database and extracting ENF from recordings using Short-Time Fourier Transform and spectrum detection.

Main Results:

• Demonstrated the effectiveness of ENF signal analysis in detecting abnormalities indicative of FFI attacks.
• Showcased the robustness of ENF extraction methods in the presence of noise and harmonic interference.

Conclusions:

• ENF signal monitoring provides a viable and computationally efficient method for detecting FFI attacks on surveillance systems.
• This approach enables real-time, on-site detection of false video and audio frames, enhancing critical infrastructure security.