Power consumption attack method based on convolutional neural network and selection message method

Abstract

The invention discloses a power consumption attack method based on a convolutional neural network and a selection message method, and the method comprises the steps: constructing a data set based on apublic key encryption algorithm, wherein the data set comprises a training set and a test set; constructing a power consumption analysis model based on a convolutional neural network method; traininga power consumption analysis model based on a convolutional neural network method; and testing by using the trained power consumption analysis model based on the convolutional neural network method.The RSA encryption scheme based on the blind Boscher modular exponentiation algorithm is realized, and the attack efficiency is far higher than that of other existing schemes on the basis of simple realization. The method provided by the invention is not only suitable for RSA encryption systems based on a blind Boscher modular exponentiation algorithm, but also suitable for most RSA encryption systems based on modular multiplication modular square loop iteration at present.

Description

technical field [0001] The invention relates to a power consumption attack method. In particular, it involves a power consumption attack method based on convolutional neural network and chosen message method. Background technique [0002] With the development of economic globalization and informatization, information transmission is applied to various fields such as finance, commerce, and military affairs. The leakage of sensitive information will not only cause huge losses to personal and corporate property, but even threaten national security in severe cases. , which poses a challenge to the construction of information security and confidentiality. [0003] The current side-channel attacks are mainly divided into two categories, one is vertical side-channel attacks, and the other is horizontal side-channel attacks. Vertical side-channel attacks mainly use statistical tools to measure a large number of power consumption curve samples and extract data features. In order t...

AI Technical Summary

Problems solved by technology

However, after a series of blinding, coupled with interference such as noise, random delay, and random clock glitches in the actual hardware environment, it is difficult to directly observe the characteristics of the power consumption curve through SPA

So far, many machine learning methods have been applied to RSA, but they all need to go through various complicated preprocessing and denoising

Images