A nobel key-search method for side channel attacks based on pattern recognition

Differential power analysis (DPA) has been known as an efficient attack for finding secret keys of cryptosystems but its efficiency may be lowered due to the misalignment of the acquired signals. Though the misalignment problem has been now solvable by various successful approaches in DPA, a lot of power traces are still required to find correct keys. Since the required number of power traces is directly connected with the efficiency of SCAs, we propose a key-search method even with relatively reduced number of power traces based on recognizing special patterns of the signal caused by cryptographic operations. Experimental results show that the proposed method is able to search correct keys with much smaller number of traces than the minimum number of traces with which the conventional methods of the energy-based DPA and frequency-based DPA succeed in finding keys.