Result: Hamming Weight-Based Simulation of Correlation Power Analysis for AES Key Extraction

This study investigates the effectiveness of Correlation Power Analysis (CPA) using the Hamming Weight model to extract AES encryption keys in a fully software-simulated environment. By leveraging Python programming, we emulate power traces not from hardware devices but through Hamming Weight calculations derived from byte-level operations during AES encryption. Simulated plaintexts are randomly generated, and key hypotheses are evaluated using Pearson correlation between expected bit-switching activity and simulated traces. The method achieved approximately 50% accuracy with just 10 plaintexts and up to 85% accuracy when using over 1,000 simulated inputs. Correlation coefficients above 0.90 were consistently observed for most key bytes. While the simulation avoids the complexity of real-world noise and hardware interference, it also lacks authentic electrical characteristics. This highlights both the novelty and the limitation of a software-only CPA framework. The findings underline the vulnerability of AES to side-channel attacks and suggest countermeasures like masking to reduce risk.