Sensitivity Analysis of Variational Quantum Classifiers for Identifying Dummy Power Traces in Side-Channel Analysis

The application of quantum machine learning (QML) to security-relevant problems has attracted growing attention, yet its practical behavior in realistic workloads remains insufficiently characterized. This paper investigates the feasibility and limitations of variational quantum classifiers (VQCs) in a representative side-channel analysis (SCA) setting focused on distinguishing real from dummy power traces. A controlled benchmark framework is developed to examine training stability, sensitivity to key design parameters, and resource–performance trade-offs under realistic constraints. Hardware-relevant factors, including finite measurement budgets and device noise, are incorporated to approximate execution beyond idealized simulation and to quantify inference-time robustness. Experimental results indicate that VQCs can extract meaningful discriminative patterns from structured side-channel inputs, while robustness and performance depend on encoding choices, circuit depth, and measurement conditions. These findings provide an empirically grounded perspective on the applicability and limitations of QML in side-channel security and offer practical guidance for future research in this area.