Design And Implementation Of CMOS Image Sensor With Physical Unclonable Function

The Physical Unclonable Function(PUF)based CMOS image sensor relays on the unpredictable and unclonable process mismatches to produce the physical unclonable information so as to replace the current security solution that bases on the authentication code stored in the Non-Volatile Memory(NVM)and to improve the security of image sensors.Since the first PUF circuit was proposed in 2000,this kind of circuits receive attention quickly in the area of information security,and become a hot technique in this area.Nowadays,the authentication scheme of image sensor is based on the authentic information code that stored in the NVM.However,the drawback of the NVM-based solution is the NVM suffers from hacker’s attack and the data stored in the memory that could be easily modification and copy,which leads the solution exposing more and more shortages in current applications.To enhance the security of the CMOS image sensor,in the thesis,a PUF based image sensor security solution is proposed.Owing to the physical unclonable information comes from the intrinsic physical structure of the CMOS image sensor array,produced by the fabrication mismatch,this information cannot be copied and cracked.Therefore,the security of the proposed CMOS image sensor with PUF is greatly boosted.The contributions of this thesis are listed in the follows.1.A detail and comprehensive analysis of the principle of PUF based CMOS image sensor system is introduced including basic CMOS image sensor modules,noise module of CMOS image sensor and the extraction solution physical unclonable information.The basic CMOS image sensor modules include optoelectronic sensing frontend circuit,correlated double sampling denoise circuit,analog to digital circuit,digital to analog circuit,and bandgap reference circuit.The noise theory of the CMOS image sensor and the corresponding denoise method are systematically analyzed including pattern noise,kTC noise,photon shot noise and thermal noise.Based on the analysis of the noise model,the physical unclonable information using the dark signal non uniformity noise(DSNU noise)is constructed.The extraction scheme of physical unclonable information in the image sensing array is accomplished.2.The extracted physical unclonable information is investigated in three properties including uniqueness,bit error rate and randomness.The stability of these three parameters is also measured under variable operating conditions including the temperature varying from-55 ℃ to 125 ℃,power supply voltage changing from 1.62 V to 1.98 V and the intensity of the incident light varying from 0.001 lux to 109 lux.The test results show that the bit error rate is not stable enough against temperature and supply voltage variation.Then,a threshold-based selection method is proposed to lower the bit error rate and boost the reliability.Finally,the bit error rate is reduced to zero at the given conditions.