Trace Alignment Preprocessing in Side-Channel Analysis Using the Adaptive Filter

Trace alignment can improve the subsequent side-channel analysis against the trace. Most current trace alignment schemes are, however, typically operated under a high signal-to-noise ratio (SNR), which demands them to be noise reduced before alignment when practical applications in the complex environment. In this paper, we propose a novel strategy for applying adaptive filtering in trace alignment preprocessing under low SNR conditions. The approach selects a trace as the reference signal of the adaptive filter, and the impulse response describing the trace offset is calculated iteratively for each trace. Different from conventional trace alignment methods, the error between the two traces in iteration determines how to eliminate the offset between trace, which eliminate most of the noise effects in the iteration process. In parallel, the filter after iterating will also function as a low-pass filter in the alignment process. Experimental studies based on three side-channel datasets demonstrate the efficacy of the proposed approach. Compared with other alignment methods, with the reasonable computational resource cost and complexity, the average number of traces required has reduced the average number of traces required by 75%, the average confidence has improved by 60%, and the success rate has increased by 72%. Our approach provides great potential for applications in trace alignment preprocessing of side-channel analysis.