Home > Tech

Research About Optical Physical Unclonable Functions In Camera-Photo Identification

2015-12-04 12:18:26



In this literature review, we will discuss some existing methods discussed in the proposed papers about PUF functions, from the non-electric PUF to the Memory based on PUF, and their reliability and limits. Chapter 1 introduced the concept of PUF and explained the reason why it stood for the frontier in system security. Chapter 2 focused on the details and examples of PUFs. At first, it introduced a non-electrical PUF called optical PUF to help understand the concept behind this design, and then focused on the SRAM PUF which we will design a system by. Chapter 3 will briefly discuss the direction of our system design.

Key Words: Optical Physical Unclonable FunctionsInter-Distance and Intra DistanceMemory-Based PUFs and SRAM PUF ArchitectureOperating PUFSRAM-PUF System Mechanism.

1. Introduction
Device identification and hardware security have become critical issues with the rapid development in technology. Improvements had been made in the field of information security, but still facing the challenges of ever-dangerous attacks.

Traditional approaches focused on protecting the password in a certain way, but it might be vulnerable against invasive attacks. Measures should be made up to apply to counter this threat, but the hardware is still under the risk of different kinds of attacks due to its intrinsic system mechanism.

Physical Unclonable Functions(PUF) are now being proposed as the next generation hardware security solution due to the randomness, unique identity relied on the hardware itself. Unlike traditional information security approaches, which store a security password inside system memory for the authentication, PUF proposed a new method mostly rely on the physical characteristics of the hardware itself. By capturing the unique behaviors of different hardware, it will generate unique fingerprints for different hardware.

By now, there have been various kinds of PUFs, based on different concepts.These PUF design and applications will be reviewed in the later chapter, including analog PUFs, delay based PUFs, memory-based PUFs.[1]


2. Type of PUFs

PUF zoo contains a lot of categories; each uses different mechanism to generate the unique identity. Before introducing the SRAM PUFs, we will take a glance on  non-electronic PUFs as they are proper examples to show how this ‘unique identity’ is generated.The concept ‘non-electric’ only reflects the nature of the components in the system that contribute to the random structure which makes the PUF unique. It does not mean that all the procedures, measurements, processing stages are not involved with electronics. [2]

The electric PUFs, are often referred to Intrinsic PUFs. The word ‘intrinsic’ means, the PUF is completely embedded the device itself, rather than rely on any 3rd equipment. Also, the realization of the PUF should be based on the naturally manufacturing process of the device.

Before the discussion of the types of the PUFs, here we introduce some basic concepts of PUF that will be mentioned in the later chapters.

·             Challenge and Response:

Unlike saving a security key inside the memory, PUFs use challenge-response pairs. The challenge could be understood as an input, or a stimulus. When it is applied on the system, the system will react, in a repeatable but unpredictable way, like a fair coin. If we compare a microchip with a coin, due to the nature condition of the chip (the material, the flaws, manufacturing techniques, etc.), the chip would always ‘react’in the same way when given the same challenge (Here we assume it is under  the same environment conditions.), but we do not know why the chip acts like this .

·             Challenge Response Pairs(CRPs):

The challenges bring responses. It is usually a huge database.

·             Inter-Distance and Intra Distance

Inter-Distance: Same challenge, 2 PUFs. Applying the same challenge on 2 different PUFs, compare the difference between the responses of each.

Intra-Distance: Same challenge, 1 PUF, twice

Applying the challenges on the same PUF twice, and compare the difference between the two responses.

2.1 The Non-electric PUFs

Non-electric PUFs generates the unique information of the device from the nature of the components of the system.

a) Optical PUFs

Optical PUFs are termed physical one-way functions at early stages. It uses a transparent material with special light-scattering particles on it, so when the laser beam penetrates the particles, it would scatter. The placement of the particles on the transparent material is an uncontrollable and random process, and the way laser scattered by the particles is a very complicated process.

For this design, the challenge could be the arch of the input laser beam, the wavelength or the focus. And the response is the wave pattern after the laser penetrates through the transparent material.

As a result, it is impossible to reduplicate another same ‘transparent material’ even use the same manufacturing procedure, as the randomness could not be duplicated. Meanwhile, any attempts to calculate the particle distribution are impossible either.


Fig.1 Basic Operation of an optical PUF


Fig.2 How the particles in the transparent material operate (Pappu, 2001)

Optical PUF is a typical example of non-electric PUF applications, designed at an early stage in the development of PUFs. It is a genius design, and it perfectly generates the identity and at the same time makes it impossible to access the security information inside the transparent material. However, the disadvantage of this design also remains clearly. It will take huge amount of effort to set up the CRPs, as it requires every arch of the laser beam. It is a tough work for the laser positioning.

2.2 Memory-Based PUFs and SRAM PUF

A). SRAM PUF Architecture

Memory-based PUF is a kind of PUF that identify hardware from their primitive data when they started. In this chapter we will focus on the mechanism of SRAM PUF. SRAM is made of many component memory cells, and it is bitable, means there are only 2 states of each memory cell. When SRAM is powered up, each memory cell will be metastable for a short period of time, and then settled into a random state of 0 or 1.[4] The state of each memory cell is unpredictable and based on the primitive characteristics of itself. SRAM’s logic circuit is made of two inverters; these two inverters are back-to-back connected. Each inverter is made of a PMOS and a NMOS, shown in Fig.3. In manufacturing process, the two inverters are designed to be the same. In general, the inverters are not exactly the same due to the uncontrollable factors in manufacturing process and usually there would be a stronger one while a weaker one exists. This would let the final state settled into 1 or 0 due to the stronger inverter. In some cases, the inverters could be of the same strength and this would cause the unstable bits.[1] PUFs use these bits, caused from the manufacturing variability, to generate a key out of the chip and used as an authentication.[5]


Fig.3 SRAM Inverter[9]


Fig.4 SRAM Logic Cell[4]

In Fig.4, it illustrates the SRAM logic cell. In the middle it is the logic cell which uses two inverters to be connected and connect them back to back(The first inverter’s input is the output of the second one and its output is the input of the second one, vice versa.). It also includes the access transistors AXL and AXR, it is for reading and writing functions.

B. SRAM Requirements For Operating PUF

For a SRAM-based PUF, several system requirements should be taken into consideration about the SRAM itself.

a). Correlation between two chips

The correlation between two SRAM chips is a very important characteristic that should be taken into consideration. This protects the unique identification of each chip.

b). Error Rate

High error rate is a main disadvantage of the SRAM-based PUF approach. This could be improved and should be tightly controlled during the design. The error rate is so high in SRAM-based PUF design (more than 10%) that it would influence the PUF response output if not handled properly.

c). Mean Value

This is a parameter that measures the number of 1s and 0s of the start up value of SRAM cell. It is obvious that after powering up the chip and take all cells into calculation, the mean value at last should be very closed to 0.5. If not, this SRAM cell is not suitable for PUFs.

d). Correlation between bits

This parameter measures the data layout and the influence of the neighboring cells. A single PUF output bit cannot predict the rest of the output bits.[6]


C. SRAM-PUF System Mechanism

To conclude the above, the SRAM-based PUF, which uses the SRAM randomness as the PUF source, was proposed in [4]. The author also mentioned in SRAM PUFs, the PUF is usually very noisy and a fuzzy extractor which should be implemented. The Fuzzy Extractor will be discussed in the later chapters. The construction of a SRAM PUF was proposed in [8]. The author proposed a method. The input is a series of binary bits. PUF uses these input as the address and the challenge. Then it takes the start up data of the SRAM to generates the response.[9]


As discussed before, PUF is full of noise and thus a fuzzy extractor is needed in the design. Fuzzy Extractor performs two functions. The first function is the information reconciliation, which correct the data bits and turns it into a stable and reliable output. The second function is privacy amplification which uses universal hash function to obtain a entropy key.

3. Conclusions

The discussion above introduced the concept of PUF and focused on the SRAM-PUF. SRAM has become very popular right now and can be found in a majority of the electronic devices.As a result, it is possible and valuable to make any new PUF-related designs in the future. In our design, we are planning to produce a SRAM PUF-based system operated on a digital camera, using the SRAM embedded inside as the PUF source. The PUF will encrypt the photos which taken by the camera, so that a third software could be used to decide if the photo is taken by this exact camera.



[1] Khoshroo, Sasan. Design and Evaluation of FPGA-based Hybrid Physically Unclonable Functions.Diss. Western University London, 2013.

[2] JOUINI, Zouha CHERIF, Jean-Luc DANGER, and Lilian BOSSUET. "Characterisation of Physically Unclonable Functions at Design Stage."

[3] Van der Leest V, Preneel B, Van der Sluis E. Soft decision error correction for compact memory-based PUFs using a single enrollment[M]//Cryptographic Hardware and Embedded Systems–CHES 2012. Springer Berlin Heidelberg, 2012: 268-282.

[4] Guajardo J, Kumar S S, Schrijen G J, et al. FPGA intrinsic PUFs and their use for IP protection[M]. Springer Berlin Heidelberg, 2007.

[5] Khoshroo S. Design and Evaluation of FPGA-based Hybrid Physically UnclonableFunctions[D]. Western University London, 2013.

[6] Bohm C, Hofer M, Pribyl W. A microcontroller sram-puf[C]//Network and System Security (NSS), 2011 5th International Conference on. IEEE, 2011: 269-273.

[7]Sunar B, Martin W J, Stinson D R. A provably secure true random number generator with built-in tolerance to active attacks[J]. Computers, IEEE Transactions on, 2007, 56(1): 109-119.

[8] Maes R, Tuyls P, Verbauwhede I. Low-overhead implementation of a soft decision helper data algorithm for SRAM PUFs[M]//Cryptographic Hardware and Embedded Systems-CHES 2009. Springer Berlin Heidelberg, 2009: 332-347.

[9] Maes R, Tuyls P, Verbauwhede I. A soft decision helper data algorithm for SRAM PUFs[C]//Information Theory, 2009.ISIT 2009.IEEE International Symposium on. IEEE, 2009: 2101-2105.

Author brief introduction:

Hu Xu, Technical Director,JiangsuHengtong Optical Network Technology Co.,Ltd.

Matte-kexu, Technical Engineer, Jiangsu Hengtong Optical Network Technology Co.,Ltd.

Add: No.88,HengtongRoad,WujiangDistrict,SuzhouArea,JiangsuProvince,China

Postcode: 215200

Recommended information
IT Minister of Egypt Meets Joint Delegation of Chinese YOFC, Huawei
IT Minister of Egypt
A Cable Joint Venture of YOFC and PT FOTI Established in Indonesia
A Cable Joint Ventur
China Eyes Fiber Optic Use to Expand Communications Systems
China Eyes Fiber Opt
China Sets up US$14.6 billion Internet Investment Fund: Xinhua
China Sets up US$14.