Realization of a pseudo-random number generator utilizing two coupled Izhikevich neurons on an FPGA platform

Abstract

Paired neurons exhibit diverse dynamic behaviors, including chaotic patterns. This paper presents an FPGA-based implementation of a high-speed pseudo-random number generator using two coupled Izhikevich oscillators. The dynamical characteristics of the neuronal model are investigated via MATLAB-based simulations, while the proposed generator is effectively modeled and simulated utilizing the Xilinx system generator framework. The model is then synthesized using the Xilinx Synthesis Tool followed by its implementation on the evaluation board of the Xilinx Spartan-6 XC6SLX9 FPGA. A post-processing procedure incorporating the exclusive OR operation has been employed to enhance the randomness of the output bits. The proposed pseudo-random number generator has a lower implementation cost compared to similar works, while achieving a maximum frequency of 49.6 MHz and a bit generation rate of 28.4 Mbit/s. The quality of the generated bit sequences is evaluated through various statistical analyses, including the scale index method, autocorrelation test, information entropy analysis, and the NIST test suite. The tests result demonstrate that the numbers generated through the proposed method exhibit a high entropy value, non-periodic behavior, and a lack of correlation. The proposed random number generator has potential applications in security and encryption systems.