Strong PUF (Physical Unclonable Function) machine learning attack resisting method based on matrix encryption

Abstract

The invention discloses a strong PUF (Physical Unclonable Function) machine learning attack resisting method based on matrix encryption, which comprises the following steps: taking response signals generated by acting multiple groups of different excitation signals on a strong PUF as to-be-encrypted information, arranging the response signals to form plaintext matrixes, then performing encryption operation, generating a ciphertext matrix by adopting two plaintext matrixes through matrix multiplication operation, elements in a conversion matrix obtained after binary conversion is carried out on the ciphertext matrix serving as final responses, being in one-to-one correspondence with the original excitation signals and serving as final CRP of the matrix encryption strong PUF. The method has the advantages that the machine learning attack resistance of the strong PUF can be greatly improved, the machine learning attack prediction rate can be reduced to about 50% and is close to random conjecture, and the risk that the strong PUF suffers from the machine learning attack is small.

Description

technical field [0001] The invention relates to a strong PUF anti-machine learning attack method, in particular to a matrix encryption-based strong PUF anti-machine learning attack method. Background technique [0002] Physically Unclonable Function (Physically Unclonable Function, PUF) is an emerging security application technology, which can be applied in the fields of secure key generation and low-cost authentication. It reduces the risk of information leakage by capturing differences in hardware manufacturing and generating unpredictable security information in a non-storage manner. The input of the PUF is called a stimulus, and the output is called a response. Each stimulus corresponds to a unique response, so we call the corresponding stimulus and response a Challenge Response Pair (CRP). According to the different ability of PUF to generate CRP, it can be divided into weak PUF and strong PUF. Since weak PUFs can only generate a limited number of CRPs, they are mainl...

AI Technical Summary

Problems solved by technology

Although the strong PUF proposed in the above two documents has a certain ability to resist machine learning attacks, with the further increase in the number of collected CRPs, the prediction accuracy rate is still very high with the model constructed by using a large number of CRPs, and there are still some problems. High Risk of Machine Learning Attacks

Images