An implementation of 2D locally coupled relaxation oscillators on an FPGA for real-time autowave generation

Abstract

This paper introduces a digital implementation of a new simple programmable autowave generator network on a field programmable gate array (FPGA). The network is a two dimensional reaction-diffusion Cellular Neural Network which consists of relaxation oscillators. The introduced implementation successfully simulates 25,600 neurons in real-time with novel Cellular Neural Processing Network architecture which uses floating-point number format. The implementation allows simulating the network with high numeric resolution and real-time monitoring of the evolution of autowaves. This FPGA implementation of the network provides a suitable platform to explore spatiotemporal behavior and to implement wave computing algorithms.