Non-Identical Inverter Rings as an Entropy Source: NIST-90B-Verified TRNG Architecture on FPGAs for IoT Device Integrity

Abstract

True random key generator (TRNG) architectures play a notable role in strengthening information security infrastructure. The development of new entropy sources based on reconfigurable hardware is always in demand, especially for the integrity of devices in IoT applications. TRNGs can be adopted for generating unique device IDs that form the data network in the IoT. A ring oscillator (RO) is an efficient entropy source which can be implemented on FPGAs or realised as ASIC hardware. This work proposes a non-identical RO array as an entropy source. The TRNG architecture, based on an increasing odd number of inverters per ring, was extensively studied. The various statistical and hardware analyses provided encouraging results for this reliable entropy unit. The suggested device-independent non-identical RO structure was implemented on five different types of FPGA hardware belonging to the Xilinx and Intel families, consuming 13 registers and nearly 15 combinational functions. This TRNG achieved a throughput of 3.5 Mbps. While the emergence of the Gaussian response evaluated true randomness, the NIST 800-90B and NIST 800-22 tests yielded good results in terms of the justification of randomness evolving from the proposed TRNG architecture.

Critical Analysis of Beta Random Variable Generation Methods

“…It is at this point that we will take a closer look at some of the most important concepts related to random and pseudo-random sequences and the tests that must be carried out to check the quality of the sequences. In general, it is possible to speak of true random numbers and pseudo-random numbers (see, for example, [6][7][8][9][10][11] among others). The former are based on physical sources and do not need an initial input to be generated.…”