A Key Generation Mechanism Based On BR-PUFs

Physical Unclonable Functions?PUFs?is unpredictable and unclonable due to the uncontrollable random manufacturing differences of integrated circuits.It can be used for secret key generation and security authentication.There is random noise in the PUFs response.In order to ensure that the secret keys generated at each time are identical,the fuzzy extraction algorithm is widely used.In the fuzzy extraction algorithm,the key generation needs random number.However,the traditional framework of fuzzy extraction algorithm has some shortcomings in dealing with random numbers.One is that there is recorded security risk when random numbers are injected externally,and the other is that the internal implementation of random number generator will consume more hardware resources.In this thesis,we propose to implement True Random Number Generator?TRNG?by multiplexing PUFs in the framework of fuzzy extraction algorithm to avoid external injection of random numbers.TRNG is implemented by multiplexing PUFs,so the improved scheme proposed in this thesis only consumes less hardware resources.In addition,the basic Bistable Ring PUFs?BR-PUFs?have a long setup time,the average setup time of 128-bit basic BR-PUFs is 23.08us,which limits its application in secret key generation scenarios.In order to reduce the setup time,this thesis proposes to truncate the long inverter ring of basic BR-PUFs into several short inverter ring.The experimental results show that the oscillation time of the improved BR-PUFs is only 20 ns,and the randomness,reliability and uniqueness of it are obviously improved.Finally,a key generation system is implemented based on the improved fuzzy extraction algorithm framework and the improved BR-PUFs.The secret key generation system is implemented on Xilinx LX110T FPGA development board.The whole system consumes 970 SLICELs and 6 block RAMs,and the maximum clock frequency can reach 127.081 MHz.The experimental results show that the throughput of TRNG is 23.8 Mbps,and it has passed the NIST randomness test.The theoretical false rejection rate?FRR?of the secret key is 1.7257×10^-12,and the number of false keys in more than 1.2 million keys is 0.The secret key generation system realizes the combination of the secret key generator and TRNG.When there is no secret key request,the true random number can be generated for other security applications,so it has a wide application prospect.