Authors: J. Ding, K. Chakrabarty and R.B. Fair
Affilation: Duke University, United States
Pages: 181 - 185
Keywords: architectural optimization, integer linear programming, microelectrofluidics, partition map, reconfigurable architecture, scheduling, electrowetting
We present an architectural design and optimization methodology for performing biochemical reactions using two-dimensional electrowetting arrays. We define a set of basic microfluidic operations and leverage electronic design automation principles for system partitioning, resource allocation, and operation scheduling. Fluidic operations are carried out by properly configuring a set of grid points. While concurrency is desirable to minimize processing time, the size of the two-dimensional array limits the number of concurrent operations of any type. Furthermore, functional dependencies between the operations also limit concurrency. We use integer linear programming to minimize the processing time by automatically extracting parallelism from a biochemical assay. As a case study, we apply our optimization method to the polymerase chain reaction.