Research On Key Technologies Of Constructing Physically Unclonable Function By Dynamic Reconfigurable Computing Array

Physical unclonable function(PUF)can provide a high-security and low-cost key generation and authentication scheme and has received extensive attention and research in the field of information security in recent years.Dynamic reconfigurable computing array can support multiple cryptographic algorithms by a single chip.At the same time,during the encryption operation,the idle Processing Elements(PE)on the array can be used to perform spurious operations to improve the physical attack resistance of the chip.Therefore,dynamic reconfigurable computing array is widely used in crypto chips.The use of PUF for key generation and authentication for cryptographic chips has higher security and less resource overhead.Therefore,it is of great significance and value to study the construction of PUF by dynamic reconfigurable computing array.At the same time,the PUF attack method has been continuously matured in recent years.Therefore,designing a PUF with higher security is very challenging.This thesis aims to design a highly secure PUF based on dynamic reconfigurable computating array.This thesis proposed the structure of PE PUF,which uses the time difference of the same PEs to process the same data to generate PUF output.At the same time,this thesis designed the router unit of the PUF and used the router unit to cascade multiple PEs on the dynamic reconfigurable computing array to implement PE PUF,which is used to increase the entropy of PUF.Then,this paper proposed PE proximity rule to reduce the impact of systematic process variation on PUF uniqueness and uniformity.Then,this thesis designed the output network for PE PUF to combine the outputs of multiple PE chains to generate the final PUF response,which is used to enhance the resistance of PE PUF to cryptanalysis attacks and modeling attacks.Finally,this thesis designed a True-Random Number Generator(TRNG)based on PE PUF.In this thesis,PE PUF is modeled and analyzed to prove that PE PUF is a time-delayed strong PUF.At the same time,through simulation experiments,this thesis verified that the router unit can increase the difference between the outputs of PE PUF produced by different inputs by about 4 times.In this thesis,through modeling of systematic process variation and simulation of PUF uniqueness and uniformity,it was verified that PE proximity rule can improve PUF uniqueness and uniformity by 17.14%and 40.8%,respectively.This thesis verified the Avalanche Effect(AE)and Strict Avalanche Effect(SAE)of PE PUF with output network through the simulation experiments and proved the resistance of PE PUF to cryptanalysis attacks.At the same time,this thesis proved the resistance of PE PUF to modeling attacks through theoretical analysis.In this thesis,the uniqueness,reliability and uniformity of PE PUF were tested by simulation,which were 49.99%,91.76%and 50.49%,respectively.The error correction model of PE PUF was designed based on the simulation results of reliability.In order to verify the feasibility of PE PUF,this thesis designed a Coarse-Grained Reconfigurable Array(CGRA)with an array size of 8?8 and used Verilog to implement CGRA.This thesis used Modelsim to verify the CGRA and synthesized the CGRA using TSMC 65?m technology.The overall area is1.12?10~5?m~2.The timing is 500MHz in the calculation mode and 3.8MHz in the PUF mode.Finally,the randomness test of the sequences generated by PE PUF-based TRNG was performed using the NIST sp800-22 test sets.It was verified that the sequences generated by PE PUF-based TRNG is random sequences.