FPGA-based PUF Structure Design And RNG Application Analysis

In the digital information era,the transmission of information and data of any device is closely related to the information security,and the security of smart devices has gained the key attention of researchers.The critical data in traditional security systems cannot meet the needs of today's security systems because it is stored in non-volatile memory,which exits the drawbacks of low security and excessive design area and power consumption.Physically Unclonable Functions(PUF)have become an alternative way to store critical data and have many applications in the lightweight aspects due to their unclonable and unpredictable physical nature.This paper mainly studies the design of structure and application analysis of PUFs in FPGAs,and the details are as follows.(1)To address the problem that strong PUFs cannot resist machine learning modeling attacks,this paper proposes a kind of PUF structure that can resist machine learning modeling attacks.It increases the complexity of the PUF structure using two methods which are structural nonlinearization and challenge-response obfuscation.The PUF is a dynamically variable DF PUF structure that can dynamically configure the operating mode of the PUF deepending on the state of challenge.Besides,it uses a bitwise XOR obfuscation mechanism to obfuscate the relationship of the challenge and the response in two operating states,so that the attacker cannot obtain the internal mapping mode of the PUF.Then,this paper analyzes the performance of the PUF in terms of security,reliability and uniqueness from the aspects of software and hardware respectively,and the results show that the DF PUF structure proposed in this paper can resist machine learning modeling attacks without reducing the performance.(2)Taking advantage of the unpredictability and physically properties of PUF itself and using its responses as a seed of random number generators(RNG),this paper proposes a PUF-based random number generator.It uses the chaotic behavior of logistic map when the parameters are reasonable to amplify the small differences DF PUF structure,and uses a XOR module to combine the initial value sensitivity of chaotic systems with the unclonability of PUF,and designs a PUF-based random number generator.The structure of PUF-based random number generator is designed and implemented in FPGA.The differences of the output sequence under different conditions are explored and the random characteristics of the output are analyzed.The experimental results show that the random number generator designed in this paper has unpredictable and unclonable properties of PUF and has good statistical properties.