A gigabit TRNG with novel lightweight post-processing method for cryptographic applications

Abstract

In cryptography, there is no simple methodology to obtain a True Random Number Generator (TRNG) that meets statistical requirements such as unpredictability, uniform distribution (bias) and independence (correlation) for random numbers. The statistical weakness occurring especially as a result of physical randomness and tried to be overcome by post-processing techniques is an important deficiency of TRNGs. In this study, the use of a novel lightweight post-processing technique based on chaotic substitution box (s-box) is proposed that can successfully solve this problem of TRNGs. The proposed post-processing is applied in real time to the ring oscillator-based TRNG in two different scenarios on the Altera Cyclone IV GX Field-Programmable Gate Array chip. Bias, correlation, entropy, Chi-square, Berlekamp-Massey and National Institute of Standards and Technology Special Publication 800-22 test techniques are used for statistical verification of real-time results. In addition, for the proposed method, the area-energy consumption parameters of TRNG are examined and a detailed literature (performance) comparison has been made. The acquired results indicate that the proposed method can overcome the statistical weakness problem of TRNGs by providing better trade-off than the methods in the literature. In addition, it has been observed that the performance of TRNG is high due to the advantages of the method such as simplicity, low area-energy requirement and high output bit rate. Performance and statistical analysis results confirm the cryptographic suitability of the proposed method and that TRNG can be used successfully in lightweight architecture and applications.