Security and Synthesis Solutions for Digital Microfluidic Biochips


Chakrabarty, Krishnendu

Liang, Tung-Che





Electrical and Computer Engineering


Considerable effort has been devoted in recent years to the design and implementation of digital microfluidic biochips (DMFBs) for biochemical analysis procedures, such as high-throughput DNA sequencing and point-of-care clinical diagnosis. As DMFBs make the transition to the marketplace for commercial exploitation, security and IP protection are emerging as important design considerations. Recent studies have shown that DMFBs are vulnerable to threats that can be classified into two main categories: (1) malicious attacks and (2) intellectual-property (IP) theft. Malicious attacks on the integrity of a DMFB system can occur before bioassay execution begins, e.g., attacks on sample integrity. Moreover, DMFBs are supported by computer-based controllers and sensor feedback. The signals transmitted from the controllers to the laboratory devices are also vulnerable to malicious attacks. Additionally, DMFBs are vulnerable to reverse engineering aimed at stealing biomolecular protocols (IP theft). Consequently, these issues need to be addressed using interdisciplinary expertise in microfluidics, microbiology, hardware design, and cybersecurity.

Prior work has also identified a number of failure mechanisms for digital microfluidic biochips. Defects such as microelectrode degradation can occur throughout the lifetime of the system. If an electrode is degraded during bioassay execution, fluidic operations associated with this degraded electrode will fail, resulting in bioassay failure. To ensure reliable bioassay execution in digital microfluidic systems, online and adaptive synthesis solutions are needed so that fluidic operations can be dynamically allocated to healthy regions.

Motivated by the above needs, this dissertation is focused on developing trustworthy and reliable methodologies for DMFB systems. The dissertation first identifies a set of security vulnerabilities in DMFBs, namely IP theft for bio-protocols, malicious attacks, and sample forgery. The dissertation presents a one-time-programmability daisychain structure that scrambles the scan-in and scan-out data from a microelectrode-dot-array (MEDA) biochip, which is the next-generation DMFB. Several attacks are developed to evaluate the security strength of this architecture, including a SAT-based attack. Simulations show that this proposed architecture can protect important bio-protocols from IP theft. The dissertation also presents a provably secure method that exploits the inherent sensing mechanism in MEDA biochips. This method validates assay execution by reconstructing the sequencing graph (i.e., the assay specification) from the droplet-location maps and comparing it against the golden sequencing graph.

In addition, this dissertation presents a robust molecular-barcoding technique that can be used to ensure the integrity of the samples and reagents in DMFBs. Related benchtop experimental studies are reported to show that the proposed method can thwart stealthy sample-forgery attacks. A small-in-size DMFB is fabricated and used for molecular barcoding without any user intervention.

Next, the dissertation presents solutions that address the reliability concerns in DMFBs. An online droplet routing method based on deep reinforcement learning is presented. The RL-based model is first trained in a DMFB simulator, and then it is used in fabricated DMFBs. The experimental results show that even though electrodes on a DMFB degrade over time, the RL droplet router can learn the degradation behavior and transport droplets using only healthy electrodes. The dissertation also presents a novel multi-agent reinforcement learning (MARL) framework for parallel droplet routing on MEDA biochips. The experimental results show that the MARL droplet router can reliably transport multiple droplets without unintended fluidic contamination. To prolong the lifetime of MEDA biochips, the dissertation finally presents a selective-sensing method such that only a small fraction of microelectrodes are utilized for droplet sensing during bioassay execution. The proposed selective-sensing method is implemented based on a new microelectrode cell design. This design is the first attempt to dynamically enable/disable droplet-sensing operations in MEDA biochips.

In summary, the dissertation tackles important problems related to key stages of DMFB design and usage. The results emerging from this dissertation provide the first set of secure and reliable methodologies for DMFBs. It is anticipated that the emerging DMFB industry will benefit from these solutions.



Computer engineering


Deep Reinforcement Learning








Multi-Agent Reinforcement Learning


Security Solutions


Security and Synthesis Solutions for Digital Microfluidic Biochips






Original bundle

Now showing 1 - 1 of 1
Thumbnail Image
25.11 MB
Adobe Portable Document Format