A Generalized Approach to Implement Efficient CMOS-Based Threshold Logic Functions
A Generalized Approach to Implement Efficient CMOS-Based Threshold Logic Functions
复制标题
实现基于 CMOS 的高效阈值逻辑功能的通用方法
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
T. Haniotakis
中科院分区：
文献类型：
--
作者：
S. Mozaffari;S. Tragoudas;T. Haniotakis
提出了一种基于整数线性规划的电流型阈值逻辑函数辨识框架。该方法最大限度地减少了晶体管数量，并受益于阈值逻辑函数的广义定义。结果表明，当输入权重不限于整数时，阈值逻辑功能可以在基于CMOS的电流模式逻辑中实现，减少了晶体管数量。提出了一种新的实现合理的权重。工艺变化，晶体管老化，电路寄生考虑。实验结果表明，与传统的基于CMOS的阈值逻辑实现相比，该方法可以在预定的硬件开销下实现更多的阈值功能，并且降低了大部分阈值功能的硬件需求。
An integer linear programming-based framework to identify the current-mode threshold logic functions is presented. The approach minimizes the transistor count and benefits from a generalized definition of threshold logic functions. It is shown that the threshold logic functions can be implemented in CMOS-based current mode logic with reduced transistor count when the input weights are not restricted to be integers. A novel implementation of rational weights is proposed. Process variations, transistor aging, and circuit parasitics are taken into consideration. Experimental results show that many more functions can be implemented with predetermined hardware overhead, and the hardware requirement of a large percentage of the existing threshold functions is reduced when comparing with the traditional CMOS-based threshold logic implementation.