Side-channel attack on COSO-based TRNG to estimate output bits
Kotaro Hayashi, Ryuichi Minagawa, Naoya Torii · 2022
As the number of IoT devices increases, security is one of the critical factors. Since random numbers are used to generate nonce and encryption keys, degradation of the randomness generated by TRNGs (True Random Number Generators) causes damage to system security. Several attacks have been proposed to reduce the randomness of the TRNG by changing the operating conditions, such as frequency injection, glitch impact, temperature change, and power supply voltage manipulation. In addition, the side-channel attack against TRNG is proposed to predict the random number output. This paper evaluates temperature variation and proposes side-channel attacks against COSO-based TRNG (coherent sampling ring oscillator-based TRNG). The results show that the temperature variation does not affect the randomness of the TRNG under the temperature range of 0°C to 80°C. On the other hand, the side-channel attack correctly read the random output numbers in 97.9 % of the cases. Note that the target circuit was connected to the output pin and side channel information was obtained from it. We propose two countermeasures against the side-channel attacks and demonstrate their effectiveness.