Prediction Accuracy Improvement of Side-channel Information Leakage by Using EM-Circuit Co-simulation of PDN with Filters

Ever since the side-channel attack (SCA) was proposed, there is a danger that the secret key used for encryption can be recovered by analyzing the noise in the power supply system. Therefore, it is important to predict the extent to which electromagnetic interference (EMI) measures can suppress such attack. To predict power noise and SCA, models of IC and a power distribution network (PDN) are required. Previous studies have targeted the case of implementing one decoupling capacitor on the PDN to evaluate the validity of the IC model, as well as the application of an equivalent circuit model of the PDN. In the current work, to improve the prediction accuracy when multiple decoupling capacitors are mounted, we examined the application of an electromagnetic (EM) field and circuit co-simulation. In this co-simulation, ports are placed at the mounting points of the electronic components on the EM field model of PDN. We then obtain a PDN multi-port network, and connect it to the electronic components in a circuit simulator. Finally, the transfer impedance from the IC to the measurement port is predicted by circuit simulation. This improves the prediction accuracy because the couplings between components are included in the analysis. In addition, multiple decoupling conditions can be evaluated at high speed because the capacitor mounting conditions on the circuit model can be changed. We implemented an AES circuit on the FPGA and predicted power noise and SCA. The results show that the prediction accuracy of noise and SCA is improved.