Efficient Implementation of Memristor Cellular Nonlinear Networks using Stochastic Computing

Abstract

Cellular Nonlinear Networks (CNN) were intro-duced by Leon Chua and Lin Yang in 1988, and are shown to be a very powerful parallel computing architecture. Later on, CNN have been designed using the processing and memory capabilities of memristors. On the other hand, Stochastic Computing (SC) has been proposed as a way to reduce the number of processing elements in a circuits. In this work, we propose using SC to implement a CNN. Specifically, we choose a memristor-based CNN, where all the operations are done using SC. As an example of application, we have used Matlab to create a CNN that performs edge detection on 512x512 grey-scale images. Results show excellent capability, while at the same time the low number of needed elements will allow to implement it in a low cost FPGA-based system.