The effectiveness of profiling deep learning side-channel attacks relies on the assumption that training and attack data follow the same distribution. However, when the profiling device differs from the target device, process-voltage-temperature (PVT) variations and clock jitter countermeasures cause distribution shifts in power traces, rendering models trained on the source device ineffective on the target. Existing domain adaptation methods typically rely on a single distributional constraint without jointly constraining kernel mean embeddings and covariance structure, thus limiting their effectiveness against strong defenses such as clock jitter. We propose Robust Feature Alignment for Side-Channel Analysis (RFA-SCA), an unsupervised domain adaptation framework that aligns source and target feature distributions without requiring the target-device key during training. RFA-SCA combines three loss components. MMD loss reduces distributional discrepancy via kernel mean embeddings in a reproducing kernel Hilbert space (RKHS), CORAL loss regularizes covariance-level feature discrepancies, and conditional entropy loss optimizes decision boundaries in the target domain. In randomized attack trials using a finite number of target-domain traces on four benchmark datasets (ASCAD, CHES CTF 2018, SAKURA-G, XMEGA), RFA-SCA achieves 100% key recovery across all six cross-device and countermeasure scenarios. Furthermore, RFA-SCA reduces the number of traces required for successful attacks by 14%–33% compared to the best baseline in the ASCAD Desync, ASCAD Gaussian Noise, and SAKURA-G scenarios. A systematic ablation study over seven loss-component variants across all six scenarios further supports the role and complementarity of the three components.

The radial contraction-expansion motion paradigm is a novel steady-state visual evoked experimental paradigm, and the electroencephalography (EEG) evoked potential is different from the traditional luminance modulation paradigm. The signal energy is concentrated chiefly in the fundamental frequency, while the higher harmonic power is lower. Therefore, the conventional steady-state visual evoked potential recognition algorithms optimizing multiple harmonic response components, such as the extended canonical correlation analysis (eCCA) and task-related component analysis (TRCA) algorithm, have poor recognition performance under the radial contraction-expansion motion paradigm. This paper proposes an extended binary subband canonical correlation analysis (eBSCCA) algorithm for the radial contraction-expansion motion paradigm. For the radial contraction-expansion motion paradigm, binary subband filtering was used to optimize the weighting coefficients of different frequency response signals, thereby improving the recognition performance of EEG signals. The results of offline experiments involving 13 subjects showed that the eBSCCA algorithm exhibits a better performance than the eCCA and TRCA algorithms under the stimulation of the radial contraction-expansion motion paradigm. In the online experiment, the average recognition accuracy of 13 subjects was 88.68% ± 6.33%, and the average information transmission rate (ITR) was 158.77 ± 43.67 bits/min, which proved that the algorithm had good recognition effect signals evoked by the radial contraction-expansion motion paradigm.