Title Page
Abstract
국문요약
Contents
Chapter 1. Introduction 34
1.1. Physical unclonable functions (PUFs) 34
1.2. Characteristics of PUFs 35
1.2.1. Challenges and Responses 35
1.2.2. Theoretical performance of PUFs 36
1.3. Analysis of the correlation between bit configuration and inter-HD through simulation models 40
1.4. Type of PUFs 44
1.4.1. Optical PUFs 44
1.4.2. Arbiter PUFs 45
1.4.3. SRAM PUFs 46
1.4.4. Nanotechnologies PUFs 47
1.4.5. Advantages and disadvantages of each type of PUFs 48
1.5. Practical optical PUFs 50
1.6. Objective and Outline 52
Chapter 2. Engineered optical PUFs with conventional type 54
2.1. Optical PUFs based on Scatter media 54
2.1.1. Conventional optical PUFs based on scatter media 54
2.1.2. Optical PUF with added gain media 55
2.2. Optical PUFs based on Reflective media 64
2.2.1. Conventional optical PUFs based on Reflective media 64
2.2.2. Optical PUF with added edible function 66
2.3. Summary 71
Chapter 3. Lensless, portable, optical PUF with fibrous medium 72
3.1. Introduction: lens-free optical physical unclonable functions 74
3.2. Stochastic random holes for 'self-focusing' in native silk 75
3.3. An analysis of the self-focusing property using the modeling of fibrous media 79
3.4. Integrated lens-less optical PUF system 86
3.5. Reliability of LOP-PUFs 93
3.6. The performance of LOP-PUFs 103
3.7. The authentication and encoding capability of the LOP-PUF 109
3.8. Summary 115
Chapter 4. Conclusion and Future directions 117
4.1. Conclusion 117
4.2. Future directions 118
References 120
CURRICULUM VITAE 132
Table 1.1. A confusion matrix. 39
Table 1.2. Design parameters of the four generated cases 42
Table 1.3. Advantages and disadvantages of each type of PUFs. Reprinted with permission from Ref.[1] 49
Table 3.1. Summary of the randomness tests of binary sequences generated from the LOP-PUFs. aNIST tests are performed using 135 sequences of 128 bits each such that...[이미지참조] 110
Figure 1.1. Proposed physically one way function for hardware security applications. Reprinted with permission from Ref.[3] 35
Figure 1.2. Challenge and response pair in PUFs device and datacenter. 36
Figure 1.3. Characteristics of PUFs. Reprinted with permission from Ref. [4] 36
Figure 1.4. Hamming distance calculation method for PUFs performance evaluation. 37
Figure 1.5. The decision threshold in the authentication process. 39
Figure 1.6. Generation method of virtual PUF response to confirm PUF performance change according to bit sequence characteristics. 41
Figure 1.7. Virtually generated bit sequences in 4 cases to compare the performance of PUFs. 42
Figure 1.8. Graph of uniformity and inter-HD of 4 cases. 43
Figure 1.9. Classification according to fabrication method of PUFs and capacity of CRPs. 44
Figure 1.10. Operating principle of optical PUFs according to the type of optical taggants. 45
Figure 1.11. Structure and operation method of arbiter PUFs 46
Figure 1.12. Structure and operation method of SRAM PUFs. 47
Figure 1.13. Structure and operation method of Nanotechnologies PUFs. Reprinted with permission from Ref. [5] 47
Figure 1.14. Components of optical PUFs using scatter and reflective media. 50
Figure 1.15. Method to generate more CRPs in optical PUF. Reprinted with permission from Ref. [6] 51
Figure 1.16. Availability of optical PUFs depending on the material of the optical particle. Reprinted with permission from Ref. [7] 51
Figure 1.17. Novel optical PUFs utilizing fibrous media. Reprinted with permission from Ref.[2] 51
Figure 2.1. The operation principle of conventional optical PUFs using scattering medium 54
Figure 2.2. Random lasing that can occur in scattering media with gain media added. 55
Figure 2.3. Method for manufacturing zinc oxide nanowircs using hydrothermal synthesis. 56
Figure 2.4. Controllable ZnO nanorod structures under different synthesis conditions. (A-D) The molar concentration of zinc nitride and hexamethylenetetramine (HMT)... 57
Figure 2.5. A top-view SEM image shows the disordered morphology of ZnO nanorod-sand a side-view SEM image shows the vertical orientation of ZnO nanorods with a... 58
Figure 2.6. Scalable nanofabrication of ZnO nanorod devices in a disordered geometry by facile self-organized synthesis. (A) Binary images are calculated from top-view SEM... 58
Figure 2.7. (A) The emission intensity summed over the emission spectral range has threshold behavior indicative of lasing action. Immediately above the threshold (Eth... 59
Figure 2.8. A device consisting of densely packed with self-organized ZnO nanorods is challenged (Cn,m) by optical excitation in a specific location (n, m) to generate a...[이미지참조] 60
Figure 2.9. Method of generating a key using the obtained random laser wavelength data. 61
Figure 2.10. Binary bitmaps extracted from four representative wavelengths within the same device after von Neumann debiasing. 62
Figure 2.11. Method of generating a key using the obtained random laser wavelength data. 62
Figure 2.12. Basic performance metrics of random laser PUFs by inter-distance calculations. (A) Inter-wavelength Hamming Distance (HD) within the same device and... 63
Figure 2.13. A schematic of invisible and covert PUFs mounted to the surface of medications. The PUFs gadget is made entirely of edible, digestible fluorescent protein... 65
Figure 2.14. Ondose authentication method The pharmaceutical company integrates an edible PUF device with each unique medication in a solid oral dose form. By accessing... 66
Figure 2.15. Silk and fluorescent proteins are used to create all protein-based edible PUFs. (A) Regenerated particulate eCFP, eGFP. eYFP. and mKate2 silk made us-... 67
Figure 2.16. Photograph shows an edible PUF device that has a thin silk layer that is embedded with fiuorescent silk microparticles. There is a 2 mm scale bar. An picture of... 67
Figure 2.17. An entropy source is provided via fluorescent silk microparticles that are dispersed randomly. An excitation and emission band set that is matched to each kind... 69
Figure 2.18. (A) Bit uniformity obtained from 30 distinct PUFs demonstrates an unbiased distribution of 0 and 1 states after von Neumann debiasing, with a mean (μ)... 70
Figure 3.1. Illustration of natural silk with diffraction-induced "self-focusing" property. The diffraction appeared differently depending on the fiber densities; low and high fiber... 76
Figure 3.2. Schematic representation of a transparent fiber (left) and an opaque fiber (right) resulting from the nanofibrillar structures present in natural silk fabric.... 77
Figure 3.3. Operating principle of the self-focusing effect. Randomly distributed fibers formed a stochastic pinhole, which caused diffraction, 'dIS', 'w', and 'λ' are the distance...[이미지참조] 78
Figure 3.4. Description of the bit extraction process, involving: (i) raw data acquisition, (ii) noise reduction and producing a cut-off image, (iii) binning, (iv) binning and... 79
Figure 3.5. Optical simulation of an opaque fiber with a diameter of 35 μm under a wavelength of 645 nm. Fraunhofer diffraction formed an intense spot within a certain... 80
Figure 3.6. Calculated spot information, including 'Spotstart', 'Spotend', 'Spotmax', and the 'Spotregion', depending on D. The 'Spot region' is marked in blue. The active pixel...[이미지참조] 81
Figure 3.7. Schematic illustration of the image sensor (MT9J003, On semiconductor, USA). (A) The top view of used image sensor. (B) The cross sectional view of the... 82
Figure 3.8. Scanning electron microscope images of the silk fibers. Reprinted with permission from Ref. [2] 83
Figure 3.9. (A) Schematic of the simulation domain. The silk medium, which was 25 μm-thick, is located at the zero of the Z position. The monitor domain is 200 μm-... 83
Figure 3.10. (A) Refractive index profile of three media with different hole densities (i.e., 0, 5, and 10%). The black and bluish green colors refer to the refractive indices... 84
Figure 3.11. Cross-sectional amplitude profiles of three results. Reprinted with permis sion from Ref. [2] 84
Figure 3.12. (A, B) Photographs of the experimental setup for the self-focusing effect. The motorized stage is used for detecting the Fraunhofer region. The 1x objective... 85
Figure 3.13. (A) The measured data of the silk sample according to depth (0, 0.1, 0.3, and 0.5 mm). (B) The intensity of the measured data. The maximum intensity of 226... 85
Figure 3.14. (A) Schematic illustration of the proposed LOP-PUF module. (B) An optical image of the sealed LOP-PUF. The inset shows a silk identification (ID) card.... 86
Figure 3.15. Optical simulation of fibrous medium. (A) Optical simulations for a density of 80 % with three different wavelengths, i.c., 635, 530, and 435 nm, corresponding to... 88
Figure 3.16. (A) Three virtual fiber media with densities of 70, 80, and 90%. The dashed lines represent the one-dimensional simulation domain for Figure3.16D. (B)-(C) Hole... 89
Figure 3.17. Morphologies and reproducibility of the virtually generated fibrous media. The morphologies of the virtually generated fibrous media of three densities (70%,80%,... 90
Figure 3.18. (A) Obtained raw data from the LOP-PUF using silk fabrics with different densities. In the red dashed box, the self-focused spots are shown. Other areas... 91
Figure 3.19. Threshold setting for image process. (A) Grayscale image of raw data obtained using the LOP-PUF. (B) Histogram of the grayscale image which shows a... 91
Figure 3.20. (A) Lens-less images (top) and cross-sectional intensity profiles (bottom) under illumination by three different color LEDs from three different angles (-15°,... 92
Figure 3.21. The flow chart of signal-to-noise rate (SNR) obtained by introducing additive white Gaussian noise (AWGN) computationally. Reprinted with permission from Ref. [2] 94
Figure 3.22. Effect of various white noise. The obtained raw data with the LOP-PUF system and artificially noise added images with different SNRs (i.e., 0, 2, 4. 6, 8, 10,... 95
Figure 3.23. The graph of bit error rate according to SNR. Reprinted with permission from Ref. [2] 95
Figure 3.24. The illustration of the LOP-PUF module with a cooling fan, which can reduce thermal noise. Reprinted with permission from Ref. [2] 96
Figure 3.25. (top) The obtained responses at various temperature conditions (i.e., 30, 31, 32, 33, 34, 35, and 36 ℃). (bottom) The ratio of number of corrected and error... 96
Figure 3.26. (A) Temperature variation of an image sensor as a function of the operating time of a LOP-PUF module with and without the blue light illumination with the... 97
Figure 3.27. Extracted bit sequences obtained by a LOP-PUF module with the temperature control (i.e., 27 ℃). Reprinted with permission from Ref. [2] 98
Figure 3.28. (A) Schematic of the measurement setup for humidity control. The LOP-PUF module is placed in the enclosed chamber with nebulizer, hygrometer, and silica... 98
Figure 3.29. Raw data (left) and bit response (right) obtained with the LOP-PUF over seven days. The bit responses show '1' bit error on the sixth day in the measure-... 99
Figure 3.30. The ratio of the number of correct and error bits over a week under room temperature and 30% humidity. Reprinted with permission from Ref. [2] 100
Figure 3.31. Mechanical stability test. (A) Schematic illustration of the measurement setup for bending test. (B) Photographs of observed silk with alignment keys (left) and... 101
Figure 3.32. Thermal stability measurement of silk (A) Photograph of measurement setup for self-focusing effect under heating. (B) Temperature change of the silk. A... 102
Figure 3.33. Generated bits and features of the LOP-PUF. Thirty bit sequences obtained by the LOP-PUF for three spectrally separate LEDs. The three LEDs were po-... 104
Figure 3.34. Silk ID card fabrication. (A) The fabrication process of the silk ID cards. (B) Photograph of the thirty silk cards for bit extraction with the LOP-PUF module.... 105
Figure 3.35. (A) Bit uniformity of the response bit data given by the ratio of '0' and '1' obtained by counting the number of digits from each silk ID card. (B) Inter-device HD... 106
Figure 3.36. Bitmap extracted from 30 silk ID cards via a LOP-PUF module through nine challenge-response pairs (i.e., Red +15°, Green +15°, Blue +15°, Red 0°, Green 0°,... 108
Figure 3.37. Power consumption of the LOP-PUF. (A) Schematic for power consumption measurement of the LOP-PUF. Two USB-type power meters were used to measure... 109
Figure 3.38. Schematic illustration for user authentication. 'Enrollment' is the step where the manufacturer registers the PUF keys from silk ID cards in the data center.... 110
Figure 3.39. (A) An illustration of a Brute Force attack test for estimating the defense ability of the authentication system by substituting a random key for fake authentica-... 111
Figure 3.40. Description for the data encryption applications of the LOP-PUF. Successful data encoding and decoding with encryption were enabled only when software... 112
Figure 3.41. Data encryption and decryption utilizing five distinct replies with '1' prediction ratios of 10, 30, 50, 70, and 90% in the materials shown in (A). Using silk,... 113