Analysis of Ring-Oscillator-based True Random Number Generator on FPGAs

In this paper, randomness of ring-oscillator-based True Random Number Generators (TRNGs) on Field Programmable Gate Arrays (FPGAs) are analyzed according to the number of constituent inverters and the number of ring oscillators. A typical ring oscillator is composed of an odd number of inverter chain, in which the output of the last inverter is fed back to the input of the first inverter. TRNG utilizes a different jitter caused by PVT variations as an entropy source. To evaluate ring-oscillator-based TRNG on FPGAs, we implement various TRNGs with different numbers of constituent inverters and ring oscillators on Xilinx Spartan-3 FPGA, and the randomness is evaluated by the NIST SP 800-22 test. According to experimental results, it shows that TRNG with the smaller number of constituent inverters and the larger number of ring oscillators provides the stronger randomness.