Hardware Reference
In-Depth Information
Chapter 8
Conclusions
This topic has described a set of design techniques for cyberphysical microfluidic
biochips. Contributions include design methods that consider timing uncertainties in
operations, optimization to handle droplets with various volumes, layout optimiza-
tion of cyberphysical PCR biochips, and a design flow for cyberphysical pin-limited
microfluidic biochips. Compared to prior design methods, this research work has
led to a computer-aided design flow that considers physical constraints and enables
the development of cyberphysical biochip systems.
We have presented the first step towards physical-aware optimization and control
software design. We have developed an algorithm for droplet tracking based on
pictures captured by CCD cameras, and a reliability-driven error recovery strategy.
Using feedback from sensors, the control software can determine whether there are
errors on the biochip. The control software discards the region where an error has
occurred to ensure reliable chip operation. We have also developed a re-synthesis
method for cyberphysical biochip, which can dynamically alter the schedule of
fluid-handling operations based on sensors' feedback.
An important contribution of this topic includes a dictionary-based error recovery
method for low-cost portable cyberphysical microfluidic systems. In order to derive
error dictionaries, a technique has been developed for realistic fault simulation.
By utilizing error dictionaries, we can achieve a cost-effective implementation of
cyberphysical systems to recover from errors that occur during chip usage.
In order to mimic the behavior of faulty fluid-handling operations, we have
developed a fault simulation framework that can integrate data collected from
real experiments. In prior work, error recovery and synthesis were based on the
assumption that all operations have the same failure probability, but this assumption
may not reflect reality. In order to address this problem, we have presented failure
models of biochips based on real data, and used these models to guide fault
simulation.
