Abstract
The chip-level design problem is critical in pin-constrained electrowetting-on-dielectric (EWOD) biochips, which not only affects the number of control pins and PCB routing layers from the manufacturing cost point of view, but also determines the functional reliability induced by excessive applied voltage. Existing works either greedily minimize the number of control pins with degraded routability, or disregard the differences in driving voltages on the electrodes, where the trapped charge due to excessive applied voltage causes significant reliability issue. This paper presents the first SVM-based classifier for electrode addressing in chip-level design stage, which simultaneously optimizes the number of control pins, routability, as well as reliability. Experimental results on both real-life chips and synthesized benchmarks show that, compared with the state-of-the-art method, the SVM-based electrode addressing method obtains significant improvements in both routability and reliability.