Correlation Power Analysis of Modes of Encryption

Abstract

Over the past few years Development in the field of cryptanalysis revealed that compromising implementation of cryptographic algorithm is much easier than targeting algorithm separately. Side channel analysis (SCA) targets implementations by some leakage information. Correlation Power Analysis (CPA) is type of SCA. CPA has successfully compromised many implementations on smart cards, 32 bit architecture and microcontroller platforms. For CPA attack adversary need precise hypothetical power consumption model to extract secret parameters. In this research leakage points are identified in six different modes of encryption i-e ECB, CBC, CFB, OFB, Counter and GCM based upon previous researches. The leakage points can be further used to generate hypothetical vectors for attack. A new model is also defined using Modulo 2 oracle which segregates hamming weights with respect to odd and even values. For proof of concept of extraction of keybytes from ECB mode practical CPA is simulated on AES-128 in 8 Bit PIC Microcontroller. Total 3000 traces are acquired to extract 3 keybytes. Another contribution of this thesis is the analysis of GCM the authenticated mode of encryption against CPA attack. Vulnerability of field multiplication via Look-up tables discovered by Lopez [1] against CPA attack is studied with respect to GCM mode. S-Box role in attack is studied in detail from power analysis point of view. Countermeasures for Counter mode and GCM mode are proposed to resist power analysis attack. This research will help in addressing side channel (Power) vulnerabilities of encryption modes.