Pyramid Entropy Source for True Random Number Generator on FPGA

Abstract

True random number generator (TRNG) is a crucial part of any modern cryptographic primitives. Designing high entropy TRNG is still a challenging task to compete. Usually, TRNGs use physical mean as random source to generate random numbers which are unpredictable. TRNG has different application but not limited to symmetric key cryptography, random simulation, nonce, user authentication protocols, and gaming. In this work, ring oscillator (RO)-based pyramid entropy source is proposed for generating device-independent true random numbers. The de-synchronization technique has been utilized for extracting the randomness from entropy source wherein linear-feedback shift register (LFSR) is used as post-processor to improve the statistical characteristics of proposed TRNG. The design of TRNG has been implemented on Altera Cyclone II EP2C20F484C6 FPGA which consumed 892 logic elements (5%), 67.75 mW power to generate 131,072 true random bits. It yields maximum entropy of 0.999987 through entropy analysis. Statistical properties of the proposed TRNG have been evaluated using NIST tests. Restart experiment is also conducted to evidence the true randomness of the proposed TRNG.