基于多项式拟合频率重构的物理不可克隆函数优化

近年来硬件安全不断受到挑战，具有不可预测性、随机性等特性的环形振荡器物理不可克隆函数（Ring Oscillator Physical Unclonable Function，RO PUF）可作为硬件安全重要的加密密钥方式，但通常原始RO PUF不满足加密密钥对随机性的要求。因此，提出了基于多项式拟合频率重构的PUF优化方法。首先，实现RO电路的硬宏设计并在现场可编程门阵列(Field Programmable Gate Array，FPGA)上进行实例化，从而获得RO阵列的频率数据；其次，针对原始响应的随机性较差的情况，通过统计分析其分布特征，利用多项式拟合法优化重构RO阵列频率；最后，采用熵密度值评估RO PUF响应的随机性。选用型号为Xilinx Artix 7103的FPGA板进行实验测试评估，结果表明所提方法不仅比原始RO PUF响应的随机性强，而且与随机补丁混合法(Random Patch Mixer,RPM)和基于回归的熵蒸馏法相比也具备更好的随机性。

Optimization of physical unclonable function based on polynomial fitting frequency reconstruction

In recent years,hardware security has been continuously challenged.Ring oscillator physical unclonable function(RO PUF) with unpredictability,randomness and other characteristics can be used as an important encryption key for hardware security,but usually the original RO PUF does not meet the randomness requirements of encryption keys.In this regard,a PUF optimization method based on polynomial fitting frequency reconstruction is proposed.First,the hard macro design of the RO circuit is implemented and instantiated on the field programmable gate array(FPGA) to obtain the frequency of the RO array.In the case of poor randomness of the original response,the distribution characteristics is statistically analyzed,and the frequency of the RO array is optimized using the polynomial fitting method.Finally,the entropy density value is used to evaluate the randomness of the RO PUF response.The FPGA board model Xilinx Artix 7103 is selected for experimental testing and evaluation,and the result shows that the proposed method is not only more random than the original RO PUF response,but also has a better randomness than the Random Patch Mixer (RPM) method and regression-based entropy distillation method.