RI-SAR: Randomized Input SAR ADC Resilient to Power Side Channel Attacks
Sumanth N. Karanth, Sirish Oruganti, Meizhi Wang, Jaydeep P. Kulkarni · 2023
Analog-to-Digital Converters (ADCs) are vital components in sensor systems, responsible for converting analog signals into digital representations. However, studies have highlighted the susceptibility of ADCs to side-channel attacks, such as Power Side Channel Attacks (PSCAs) and Electromagnetic Side Channel Attacks (EM-SCAs), which compromise the confidentiality and integrity of the system. In this paper, we comprehensively investigate the security concerns in Analog Mixed Signal Circuits, with a specific focus on Successive Approximation Register (SAR) ADCs. Our research involves the design and evaluation of an 8-bit SAR ADC in 65nm commercial technology. The analysis of power side channel attacks using Convolutional Neural Network (CNN) techniques reveals that the LSB bits are more secure than the MSB bits, achieving root mean square (RMS) error of 26/256 on average. Based on these findings, we propose a novel circuit-level protection technique called Randomized Input SAR (RI-SAR) that demonstrates the effectiveness of increasing the RMS error of PSCA attacks to 102/256, giving a 392% increase. This technique maintains comparable performance to the unprotected ADC. The proposed technique introduces minimal overheads and can be applied to other ADC architectures. Comparative analysis with prior works highlights the advantages of RI-SAR in terms of simplicity and robust security.