[book] We identify the research challenges in the new area of digital microfluidic biochips, targeting researchers in the fields of compiler technology, design and test, and electronic design automation, and thus offering a new application area for their methods. We propose a routing-based model of operation execution, along with several associated compilation approaches, which progressively relax the assumption that operations execute inside fixed rectangular modules. Since operations can experience transient faults during the execution of a bioassay, we show how to use both offline (design time) and online (runtime) recovery strategies.

==pdf==

[journal paper] OpenDrop consists of a do-it-yourself biochip, an automated software tool with visual interface and a detailed technique for at-home operations of microfluidics. We address the main challenges that users may encounter: accessibility, bio-protocol design and interaction with microfluidics. Our work provides a step towards a future in which everyone will be able to use microfluidic devices for their personal applications.

==pdf==

[journal paper] The current industrial trend is to produce biochips in the form of disposable cartridges, designed to execute specific bio-protocols. We propose the first methodology that automates the design of application-specific biochips, aiming at obtaining low-cost biochips, while maximizing fault-tolerance to fabrication defects.

==pdf==

[journal paper] Due to variability and randomness in biochemical reactions, droplet operations may finish earlier than expected. To exploit the resulted slack in the schedule, we implement a quasi-static compilation that takes advantage of both static and dynamic error recovery techniques. We determine offline a database of alternative implementations. During the execution of the bio-protocol, several implementations are selected based on the current execution scenario.

==pdf==

[journal paper] So far, to simply the compilation problem, researchers have assumed an abstract droplet size of one electrode. We develop the first routing algorithm for biochips that considers the real size of droplets. Our work targets a novel biochip architecture based on Active Matrix Electrowetting on Dielectric, which enables large arrays (i.e., of thousands electrodes).

==pdf==

[conference paper] During the execution of a bio-protocol, transient faults can occur and alter the droplet volumes. We propose two error recovery approaches: (1) create the copy droplet sequentially, after the error is detected (i.e., time redundancy), or (2) create the copy droplet in parallel, before the error is detected (i.e., space redundancy). Because neither time nor space redundancy is better, we implement a decision algorithm that optimizes their tradeoff for the actual context.

==pdf==

[conference paper] So far, droplet operations on biochips were scheduled considered an expected finish time. However, mixing and incubation of droplets depend on their chemical composition and may finish earlier than expected. We exploit the resulted slack in the schedule through an adaptive compilation strategy triggered at runtime, when operations experience variability in their execution time.

==pdf==

==pdf==

[conference paper] So far, biochips were considered to be of rectangular shape. However, in practice, non-regular architectures are common for application-specific biochips. We develop an algorithm that synthesizes application-specific biochips. Furthermore, our approach increases the fabrication yield by introducing redundant electrodes to tolerate fabrication defects.

==pdf==

[conference paper] During the execution of a bio-protocol, the droplet volumes can vary erroneously due to parametric faults, thus impacting negatively the correctness of the application. We develop an adaptive algorithm that determines the appropriate recovery actions at the moment when faults are detected.

==pdf==

[conference paper] We propose an abstract model for a bio-protocol, consisting of a sequencing graph, which can capture all the fault scenarios in the application. We demonstrate that, by taking into account fault-occurrence information, we can derive better schedule implementations, which leads to shorter application completion times, even in the case of faults.

==pdf==

So far, metamaterials are understood as materials - we want to think of them as mechanisms. We demonstrate metamaterial objects that perform a mechanical function. Such objects consists of a single block of material made of microscopic cells that coordinate macroscopic movement.

==pdf== ==video==