基于RTD可编程逻辑门的数字电路3层网络综合算法

随着集成电路的不断发展，CMOS器件的工艺逐渐达到其物理设计极限，研究新器件和新设计方法成为集成电路继续发展的必经之路. 阈值逻辑门因具有强大的逻辑功能而备受关注，共振隧穿二极管（RTD）因其负阻特性在设计阈值逻辑门时更具优势. 由于阈值逻辑门与二进制神经元模型有相似之处，因此可用神经网络模型实现逻辑函数，从而为电路设计提供新的思路. 对基于RTD可编程逻辑门的3层网络算法中的隐层综合算法进行了改进，提出采用汉明距离最大优先覆盖的方法对真向量进行覆盖，从而提高了真向量的覆盖效率，减少了隐层函数个数，并采用真假向量标记的方法简化了隐层综合算法.提出的算法比原隐层综合算法简单，进一步简化了基于RTD可编程逻辑门实现n变量函数的电路.

Three-layers network synthesis algorithm of digital circuits based on RTD programmable logic gates

With the development of integrated circuit, the CMOS technology will face some fundamental physical limittations. The new devices and new design methods thus become very important to the further development of integrated circuit industry. Threshold logic gate has been paid too much attention because of its powerful logic function, and the resonant tunneling diode (RTD) appears to be more suitable for designing the threshold logic gate because of its negative resistance characteristics. As the threshold logic gate and binary neuron model have similar structure, neural network model can be used to implement the logic functions, leading to a new idea for circuit design. In this paper, we improve the hidden layer algorithm of three layers network algorithm based on the RTD programmable logic gate, and propose a method which adopts the biggest distance of Hamming distance in preference to cover the true vector. Our method can improve the coverage efficiency of the true vector, and reduce the number of the hidden layer function. Meanwhile, we use the mark of true vector and false vector to simplify the algorithm. Compared to the original algorithm, the proposed algorithm is concise, and can simplify the circuit of the RTD programmable logic gate to implement the n variable function.