Structural Attacks and Defenses for Flow-Based Microfluidic Biochips.

Area of Science:

• Biotechnology and Biomedical Engineering
• Microfluidics
• Diagnostics

Background:

• Flow-based microfluidic biochips (FMBs) are increasingly commercialized for diagnostics.
• Outsourced design and manufacturing expose FMBs to physical and intellectual property (IP) theft attacks.
• Structure-based (SB) attacks pose a significant threat to FMB integrity and diagnostic accuracy.

Purpose of the Study:

• To demonstrate the first structure-based (SB) attack on commercial FMBs.
• To develop methods for detecting SB attacks and safeguarding FMBs against IP theft.
• To propose novel countermeasures for FMB security.

Main Methods:

• Structure-based attacks were performed by maliciously decreasing FMB reaction chamber heights.
• Experimental validation used fluorescence microscopy to correlate chamber height with DNA amplification intensity (R² = 0.987).
• Deep learning-based anomaly detection algorithms were employed to identify microstructural anomalies, achieving ~96% accuracy.

Main Results:

• SB attacks successfully produced false-negative results in FMBs.
• A high correlation was observed between chamber height reduction and decreased fluorescence intensity.
• Deep learning models demonstrated high accuracy in detecting deliberately introduced microstructural anomalies.

Conclusions:

• FMBs are vulnerable to SB attacks that compromise diagnostic accuracy.
• Deep learning offers effective detection of microstructural anomalies in FMBs.
• A novel device-level watermarking scheme using intensity-height correlation can protect FMBs from IP theft and SB attacks.