Effect of Temperature Variation in Experimental DPA and DEMA Attacks

Side-Channels attacks are usually performed to measure the vulnerability of cryptocircuits against malicious attacks. The conditions in which the attacks are carried out have influence in their effectivity. In this sense, temperature variations should be considered to assess the complete vulnerability of a system, but they have not been deeply considered in the literature. For this purpose, experimental DPA and DEMA attacks are carried out over one of the widest used and studied block cipher, namely AES algorithm, implemented in a Spartan-6 FPGA. The effectivity of DPA and DEMA attacks under different temperatures: 10, 25, 50 and 70°C have been studied experimentally. The attacks have been made over the 128 bits of two randomly chosen keys. The security achieved for each attack is measured using the Measurements to Disclose (MTD) the key, which determines the minimum number of patterns needed to retrieve the secret key. From the results we can obtain interesting conclusions: DPA attack is more effective than the DEMA attack over the AES implementation on FPGA. On the other hand, we conclude that the key has influence on the MTD value, but the variability between keys is of the same magnitude as the variability between temperatures, meaning that temperature variation is not a decisive factor in the effectiveness of an attack.