A physical unclonable function based true random number generator
Date of Issue2017-05-19
School of Electrical and Electronic Engineering
Random number is used in various fields: Monte Carlo simulation, cryptography, factorizing large integers, games and gambling etc. Currently, many true random number generators are based on quantum mechanics (photon arrival time) and chaos systems (atmospheric noise). However, the present quantum solutions are too bulky to implement on chip-level and harvesting atmospheric noise is too slow for some applications. On the other hand, Physical Unclonable Function (PUF) is defined in the last decade, and now is often used in cryptography as authentication and secure key storage. A silicon based PUF utilizes the uncontrollable random variations in the IC manufacturing process to produce unreproducible chip-unique “fingerprints”. This allows PUF to serve the purpose of device authentication and storage of secure keys. Since the variations are random, PUF can be used as an entropy source of true random number generators. In this report, a novel architecture of PUF-based true random number generator is proposed. The sequences produced pass all NIST tests and yield a uniqueness of around 0.5 which clearly indicates its unclonablility. Moreover, tests have shown that the sequences produced from the same hardware instance for multiple runs show little Pearson correlation between each other. Therefore, predicting future sequences based on the current sequence is hard for possible adversary. Due to its simple structure, hardware cost is only 131 LUT and 69 flip-flops. Limited by the UART interface, the generation rate is only 90kbits/s, however, the propagational delay of the generator block is only 56ns and therefore a faster rate can be expected.
Final Year Project (FYP)
Nanyang Technological University