Minimizing the Impact of Unbalanced Splitting Errors on DMFBs Without Any Overhead

There has been a lot of attention to digital microfluidic biochips (DMFBs) in biochemical and medical industries. A major error source in operations on a DMFB is due to an unbalanced splitting error at a mixing operation. There has been a lot of existing research to tackle this problem; essentially two strategies have been studied so far. One strategy is a retry-based roll-back strategy; if an unacceptable error is detected, we roll back to a checkpoint to redo some part of the experiment again. The other strategy is to duplicate some concentration values, and mix droplets with the same concentration values to mitigate the impact of unbalanced splitting errors. Obviously, both techniques increase the number of mixing operations. This paper proposes a totally new technique compared to the above-mentioned existing two strategies. More precisely, our proposed method only selects the destinations of the output droplets from an intermediate node that has more than two outputs to be used later; our method does not need any additional overhead unlike the existing strategies. Although such a selection has not been studied in the community, this paper reveals that we can mitigate the impact of unbalanced splitting errors by an appropriate selection. To find the best selection, we introduce a measure called RRE (Ratio of Residual Error) which can tell how much each error affects the final target concentration very efficiently. We also show some simulation results by which we can confirm that our method can indeed reduce the impact of unbalanced splitting errors without any overhead.