DPA attack method for carrying out fitting by utilizing neural network, system and terminal

Abstract

The invention belongs to the field of cipher algorithm analysis and detection and discloses a DPA attack method for carrying out fitting by utilizing a neural network, a system and a terminal. In thetraining stage, a predictive neural network automatically learns and obtains a mapping relationship between an energy consumption characteristic vector and an intermediate combination value accordingto a training energy consumption set; and in high-order DPA attack, the predictive neural network is used to predict the intermediate combination value of the attack energy trace, and the correlationcoefficient between a predictive intermediate combination value and a conjecture intermediate combination is calculated, thus the device key is attacked. The method of the invention eliminates the requirement that the mask must be understood in the learning stage, simultaneously avoids the requirement of the high-order DPA on the energy consumption cross-combination, reduces the attack condition,and improves the attack efficiency. Experimental results show that the attack algorithm is feasible and efficient.

Description

technical field [0001] The invention belongs to the field of cryptographic algorithm analysis and detection technology, and in particular relates to a DPA attack method, a system and a terminal for fitting by using a neural network. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: [0003] In the side channel attack, the method of using the energy consumption change in the cryptographic device to attack the private information such as the key is called an energy analysis attack. Essentially, this attack exploits two types of energy consumption dependencies: data dependencies and operational dependencies. Based on this theory, Kocher et al. first proposed the concept and method of DPA attack in 1999, which made the security of cryptographic devices such as smart card chips suffer great challenges. The attack is based on the principle that the energy consumption of a cryptographic device depends on the intermed...

AI Technical Summary

Problems solved by technology

The attacker must have full control of the training device (be able to modify the encryption code in the training device) to learn the random mask, which is generally difficult to meet

Therefore, the applicable scenarios of high-order template attacks are very limited

Images