Research On Power Analysis Attack Based On Modular Exponentiation Algorithm

The appearance of power analysis attack is breaking through the traditionalcryptography analysis, and also threatens the application security of the cipher products,especially the cipher hardware devices. Nowadays, RC4, DES and AES algorithms arealready broken by Power Analysis Attack in many research institutes. And this situationmake the security of the public key cryptosystems receive more and more attentions.For the bank cards usedwidely are magnetic cards which have many security problem,sochinese government is planning to replace themwith the smartcard during the12thFive-Year Plan. In a word, the security of smartcard is urgent issue, and play importantrole in people’s livelihood and national security.This dissertationreveals application principle of power analysis attacks byresearching on the power characteristic of public key cryptography and the law of theleaked power information aiming to the RSA cryptosystem which works on thesmartcard. The modular exponentiation is the essential part of the modern public keycryptography, which also is the central issue of the power analysis attack.We research on the modular exponentiation algorithm of the RSA cryptographysystem working on the smartcard, and achieving following creative results:(1) A new SPA attack method which based on the elaborately chosen message isproposed: the method combines traditional cryptographic analysis methods with poweranalysis. Comparing with the old chosen-message simple power analysis attack on RSA,there is a large amount of plaintextswhich can break the countermeasure which blocksthe specific messages. We can extract the secret key from one single power trace withthe100%accurate rate through the elaborately chosen message.(2) A new chosen message SPA auto-attack algorithm based on collision is given.Firstly, we analyze the attack problems caused by the alignment for power traces underthe real circumstance. Then, various auto-attacks for different collisions are proposed.Due to the auto-attack algorithms depending on a priori knowledge, a new chosenmessage SPA auto-attack algorithm based on the collision is proposed to avoiddependencyand we can get88%accurate rate. (3) Study and summary the affects of sampling resistance in the power analysisattack. Sampling resistance is indispensable part of the detecting board and plays aimportant role during the attack. We study the selection of sampling resistance throughtheoretical analysis and experiment. Finally the suitable selection range of the resistanceis shown.(4) Study and summary the methods of noise filtering. The step after getting thepower traces is to filter the noise in order to benefit to power analysis. There are variousnoise filtering methods, and we get a result that the moving average filter is suit for thepower analysis to handle with the noise in the power traces after comparing thosemethods.