Enhancing Electromagnetic Side-Channel Analysis in an Operational Environment

Side-channel attacks exploit the unintentional emissions from cryptographic devices to determine the secret encryption key. This research identifies methods to make attacks demonstrated in an academic environment more operationally relevant. Algebraic cryptanalysis is used to reconcile redundant information extracted from side-channel attacks on the AES key schedule. A novel thresholding technique is used to select key byte guesses for a satisfiability solver resulting in a 97.5% success rate despite failing for 100% of attacks using standard methods. Two techniques are developed to compensate for differences in emissions from training and test devices dramatically improving the effectiveness of cross device template attacks. Mean and variance normalization improves same part number attack success rates from 65.1% to 100%, and increases the number of locations an attack can be performed by 226%. When normalization is combined with a novel technique to identify and filter signals in collected traces not related to the encryption operation, the number of traces required to perform a successful attack is reduced by 85.8% on average. Finally, software-defined radios are shown to be an effective low-cost method for collecting side-channel emissions in real-time, eliminating the need to modify or profile the target encryption device to gain precise timing information