A Fully Programmable Nano-Watt Analogue CMOS Circuit for Gaussian Functions

摘要

A CMOS analogue circuit with Gaussian Function output signal is presented for neuro-fuzzy applications. Also, a circuit to generate the derivative of Gaussian function is proposed. We simply program the variance, the mean and the maximum values by three current sources, since current mode designing is considered. Having confirmed that the mathematics we discussed is matched with the results, simulation results have shown the deviation from the ideal Gaussian function is about 1%. Since all the blocks are designed such that all devices work in weak inversion region, ultra low power consumption obtained. Programmability, power consumption and layout of the proposed circuit are also presented.