ABSTRACT

In the Beyond Moore era, QCA (Quantum-dot Cellular Automata) has been widely studied because of its low power consumption and high device concentration. As the core of chip design, EDA technology has become an indispensable step in QCA research. It refers to the use of computer-aided design software to complete large-scale circuit function design, logic synthesis, placement and routing, design rule check and other processes. In the present study, there are still many defects in QCA placement and routing. The main problems it faces are low success rate, too large area and too many cross lines. To address these problems, our team had proposed a hybrid strategy based on the genetic algorithm and the improved A-star algorithm. But the running time of the previously proposed algorithm is not satisfying, besides it lacks self-adaption feature which is important as the problem size varies among different circuits. This paper replaces the genetic algorithm with a simulated annealing algorithm, and introduces a self-adaption strategy to better accommodate different circuit sizes. Experimental results show that the new algorithm combination possesses a higher success rate, lower running time and more intelligent adaption ability.

摘要

在超越摩尔时代，QCA（量子点元胞自动机）因其低功耗和高器件集中度而得到了广泛的研究。EDA技术作为芯片设计的核心，已成为QCA研究中不可或缺的一步。是指利用计算机辅助设计软件完成大规模电路功能设计、逻辑综合、布局布线、设计规则检查等过程。目前的研究中，QCA布局布线仍然存在很多缺陷。其面临的主要问题是成功率低、面积过大、交叉线过多。针对这些问题，我们团队提出了一种基于遗传算法和改进的A-star算法的混合策略。但之前提出的算法的运行时间并不令人满意，此外，它缺乏自适应功能，这一点很重要，因为不同电路的问题大小各不相同。本文用模拟退火算法代替遗传算法，并引入自适应策略以更好地适应不同的电路尺寸。实验结果表明，新的算法组合具有更高的成功率、更低的运行时间和更强的智能适应能力。