Abstract: Side-channel analysis (SCA) extracts sensitive information from a device by analyzing information that is leaked through side channels. These measurements are correlated with specific operations executed on the device, e.g., encryption or decryption, allowing to extract useful information from the data. Countermeasures, however, disrupt the synchronization between the device’s operations and the corresponding side-channel data, yielding their alignment a prerequisite for successful SCA. This work describes parallels between side-channel analysis and molecular biology, and proposes a novel approach to align side channel traces using genomic sequence alignment methods. It was discovered that that Multiple Sequence Alignment techniques can align power traces with higher quality than elastic alignment (based on Dynamic Time Warping), thereby enabling downstream SCA methods, e.g., Correlation Power Analysis, to extract cryptography keys with up to 44% less traces.
This research considers the device as a species and each trace as the genome of an individual within that species’ population. Countermeasures are viewed as the evolutionary forces that facilitate the adaptation of the species to a particular environment, e.g., natural selection. This is a novel approach to align side-channel traces for improving SCA, drawing inspiration from the field of Bioinformatics, and more specifically the domain of multiple sequence alignment. A proof-of-concept implementation was developed, which used a commercially available SCA tool for the evaluation of the resulting trace alignments. Higher quality alignments were obtained, compared to the elastic alignment methods and authors were able to almost half the number of traces required by CPA to expose the secret key. While time performance was not the goal of this work, preliminary results showed that MSA-based alignment is slower than elastic alignment, thereby suggesting a trade-off between trace availability and alignment time.
