Simulation on fast random bit generation using laser chaos

Abstract

Random numbers are of central importance nowadays in applications such as cryptography and statistical simulations. Conventional random numbers generators are of two kinds: true random number generator and pseudo-random number generator. Recently, a random number generator using the chaotic dynamics of semiconductor lasers has been proposed and demonstrated at rates exceeding gigabits per second. However, the approaches to invoke the chaotic dynamics were mainly based on laser subject to optical feedback which leads to the unwanted features that may cause residual correlation. In this project, we numerically demonstrate the generation of random bits by using an optically injected semiconductor laser, which avoids the problems caused by optical feedback. In order to extract the random numbers from the chaotic signal, several post-processing procedures are tested with the Standard Randomness Test Suit provided by NIST (National Institute of Standards and Technology). In the first stage of the project, the power intensity of the signal is used as the chaotic signal source and we successfully generate random numbers at 0.2GHz which pass the tests. After this, the real part of the signal is chosen as the chaotic signal source. Keeping other parameters fixed, relations between the number of tests passed and the sampling rates, differentiation order and number of bits selected out in each 8-bit ADC result is researched into. It is shown that at a sampling frequency of 2.5GHz, taking 2 bits out of each 8-bit result with a differentiation order of 10th can give the best result. And it is verified by the Standard Randomness Test Suit that this scheme passes all test with a generation speed of 5GHz. In the final stage of the project, the post-processing procedure which mixes the random numbers generated from real part and imaginary of the signal (both sampling at 2.5GHz, taking 2bits each 8-bit ADC) and differentiates the series at 15th order is tested. And the random numbers generated successfully passes the tests.