Preprocessing of the Physical Leakage Information to Combine Side-Channel Distinguishers

The security and privacy of modern computing devices have become an important design metric with the unprecedented increase in the amount of personal information stored in the digital domain. Side-channel attacks have been demonstrated to be one of the primary threats for the security and privacy of these devices. Understanding the working principles of side-channel attacks has, therefore, become an important research problem. An efficient preprocessing technique is proposed in this work for an attack scenario where the amount of time to collect physical leakage (PL) and access to the device is limited. The proposed preprocessing technique utilizes different side-channel distinguishers to decrease the required number of PL measurements for a given success rate by enhancing the quality of the leakage signal. Two commonly used distinguishers, Pearson correlation and mutual information, are combined in this work. For the first time, combined distinguishers are used to improve the performance of both the preprocessing and the attack steps. The success rate of the proposed attack framework outperforms the conventional single distinguisher side-channel attacks by 33% and 30% for unmasked advanced encryption standard (AES) and masked AES, respectively.