A method for generating high-entropy random numbers in physically unclonable function hardware

By extracting noise entropy sources in IoT terminal devices using SRAM PUF and combining entropy accumulators and improved hash functions to generate high-entropy random number seeds, the security problem of random number generation in IoT terminal devices is solved, achieving low-cost and high-security random number generation and ensuring the security of device keys and data encryption.

CN116527238BActive Publication Date: 2026-11-17SOUTHEAST UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211418971.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-11-14
Publication Date
2026-11-17
Estimated Expiration
2042-11-14

AI Technical Summary

Technical Problem

Due to limited computing and storage resources, IoT terminal devices lack hardware security chips, making it difficult to establish secure and high-performance trusted security mechanisms. Furthermore, existing random number generators struggle to generate high-entropy random numbers to ensure the security of device keys and data encryption.

Method used

A hybrid random number generator (HRNG) is used to extract noise entropy sources through the SRAM physically unclonable function (PUF), and a high-entropy random number seed is generated by combining an entropy accumulator and an improved hash function. A true random number generator is then used to generate random numbers that meet security requirements.

Benefits of technology

Low-cost, high-security high-entropy random number generation has been implemented in IoT terminal devices to meet the requirements of device keys and data encryption, thereby improving the uniqueness and security of device identity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116527238B_ABST
    Figure CN116527238B_ABST
Patent Text Reader

Abstract

This invention relates to a method for generating high-entropy random numbers in physically unclonable function (PUF) hardware, comprising three parts: initial entropy source extraction, entropy accumulation, and high-entropy random number generation. In the entropy source extraction part, the IoT terminal device extracts a "noise" fingerprint, i.e., the initial power-on value, based on the physical characteristics of the SRAM memory, as the initial PUF entropy source. In the entropy accumulation part, the extracted PUF value is used as a challenge-response pair as a low-entropy source input, and a high-entropy random number seed is extracted from the PUF value using an entropy accumulator based on the HMAC algorithm, serving as the device key. In the high-entropy random number generation stage, a true random number seed is used as input, and a pseudo-random number generator based on cryptographic security generates high-entropy random numbers of different lengths using different high-entropy algorithm generators. This method utilizes the hardware fingerprint unique to the hardware device (PUF), reducing the computational and storage overhead of the IoT terminal device's trust system, while providing secure high-entropy random numbers for the terminal device's trust root.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the application of PUF hardware fingerprinting in a new field, specifically to a method for generating high-entropy random numbers in physically unclonable function hardware, belonging to the field of information security technology. Background Technology

[0002] Currently, IoT terminal devices generally suffer from limited computing and storage resources and a lack of hardware security chips, making it difficult to establish secure and high-performance trusted security mechanisms. The quality of random numbers is key to trusted security mechanisms and plays a central role in the encryption and protection of privacy data.

[0003] High-quality random numbers should meet three conditions: randomness, unpredictability, and non-repeatability. Randomness means that the occurrence of random numbers is random, appears to be uniformly distributed, and can pass randomness detection. Unpredictability means that even if an attacker obtains the software and hardware algorithms and principles of a true random number generator, they cannot predict the random numbers to be generated. Non-repeatability means that the generated random numbers are not cross-correlated with previously generated and subsequently generated random numbers, that is, it is impossible to predict subsequent random numbers based on prior knowledge.

[0004] Information entropy is a fundamental concept in information theory, representing the various possible uncertainties of an information source. Shannon proposed a quantitative mathematical expression for information entropy. Therefore, a high entropy value for random numbers implies high randomness. Furthermore, according to the "Security Technical Requirements for Cryptographic Modules" issued by the State Cryptography Administration, general security parameters require a minimum entropy of at least 256 bits. Therefore, when designing the key and random numbers for the root trust device, the key length should be at least 256 bits, and sufficient minimum entropy should be considered.

[0005] To provide the random numbers required for encryption in trusted security functions, we need to construct a random number generator (RNG) in the terminal device. Currently, there are three types of random number generators: True Random Number Generator (TRNG), Pseudo Random Number Generator (PRNG), and Hybrid Random Number Generator (HRNG). Generally, PRNGs are used in computers. The principle is to select a fixed seed as input and use a deterministic algorithm to generate a random number sequence. However, a drawback is that the resulting random numbers will exhibit numerical cycles after a certain length. TRNGs use physical methods to generate random numbers, producing truly random and non-repeating random number sequences. HRNGs combine TRNGs and PRNGs. TRNGs generate a true random number seed from an entropy source, and this seed is then input into the PRNG to generate random numbers. In practical applications, HRNGs have the advantage of not being blocked by entropy source refreshes compared to TRNGs.

[0006] Physically unclonable functions (PUFs) arise from unavoidable process variations during the manufacturing of IC devices and are therefore unique. Different devices have different stimulus-response signal pairs (challenge-response pairs), which is equivalent to assigning each device a unique fingerprint. This unclonable fingerprint can be used for authentication. Currently, representative physically unclonable functions include optical PUFs, Arbiter PUFs, ROPUFs, and SRAMPUFs.

[0007] SRAMPUF (Signal-to-Person Response Fingerprint) is based on the fact that the difference in the transistor threshold voltage at each SRAM power-on results in a unique 0-1 state for each SRAM cell, which can serve as a device fingerprint. The PUF stimulus and feedback values ​​are referred to as "challenge-response pairs." Furthermore, the initial value of the SRAM at each power-on exhibits a certain range of random deviations due to the random differences in the transistor threshold voltage, i.e., fingerprint noise. This experiment extracts the random values ​​of the noise from the SRAMPUF fingerprint as an entropy source. Based on the physically non-cloning function property of PUF, the noise deviation present at each SRAMPUF power-on is generated by physical phenomena in the hardware circuit and possesses true randomness. It can be used as an entropy source to generate a true random number seed through entropy accumulation. Summary of the Invention

[0008] This invention addresses the problems existing in the prior art by providing a method for generating high-entropy random numbers in hardware with physically unclonable functions. The technical solution designs a hybrid random number generator (HRNG) to simultaneously generate random number seeds and high-entropy random numbers in terminal devices. It solves the problems of generating security device keys and high-entropy random numbers in IoT terminal devices to ensure the security of the device's secret ID. Simultaneously, the high-entropy true random numbers can be used for other security functions such as data encryption, providing a lightweight solution for generating high-entropy random numbers in IoT terminal devices.

[0009] To achieve the above objectives, the technical solution of the present invention is as follows: a method for generating high-entropy random numbers in physically unclonable function hardware, comprising three stages: an initial entropy source extraction stage, an entropy accumulation stage, and a high-entropy random number generation stage, with the specific steps as follows:

[0010] Step S1: In the initial entropy source extraction stage, the initial data is generated by a physically non-cloning function hardware, and then the 01 bit sequence of the obtained initial data is evaluated by different mathematical methods.

[0011] Step S2: In the entropy accumulation stage, the initial value of the PUF circuit is taken as input by the entropy accumulator. The initial value of the PUF is accumulated and extracted in the entropy accumulator by means of XOR function, message digest function, etc., so as to extract the true random number seed with high entropy value from the initial value of the PUF. The extracted true random number seed can be used as the security device key.

[0012] Step S3: In the random number generation stage, a high-entropy true random number seed is used as input. In the true random number generator, the seed is processed by a hash function, a symmetric encryption algorithm, etc., and high-entropy random numbers of different lengths are generated as needed.

[0013] As an improvement to the present invention, the initial entropy source extraction stage in step S1 further includes:

[0014] S11, the entropy source extractor extracts the "noise" in the SRAMPUF fingerprint. Since the initial value of SRAM will have a certain range of random deviation when it is powered on each time, SRAMPUF is a noisy fingerprint. Moreover, this noise is caused by physical principles and has true randomness. Based on the noise in the SRAMPUF fingerprint, this invention extracts the random value of the noise as the entropy source.

[0015] S12, for the 01 bit sequence of the challenge response pair obtained by extracting the PUF value, the Hamming distance method is used to evaluate the performance of the obtained PUF response value, including uniqueness and robustness, which represent the uniqueness feature of the PUF value hardware fingerprint and the robustness feature of the reliable hardware fingerprint, respectively; the minimum entropy method is used to impose corresponding restrictions on the relationship between the initial data input and output lengths in the entropy accumulation stage according to the magnitude of the minimum entropy.

[0016] Step S1 includes the following sub-steps:

[0017] S11, set challenge values ​​of different ranges and lengths, challenge the memory device and obtain its response value, obtain challenge response pairs, including obtaining PUF values ​​of different address ranges under the same length, and obtaining PUF values ​​of different times under the same address range;

[0018] S12, calculate the Hamming distance n times for each pairwise different range of PUFs under the same length D, and obtain the average Hamming distance. The uniqueness of PUF values ​​can be determined based on This indicates that, under ideal conditions, the uniqueness value is 50%; for each pair of PUFs of the same length D with different numbers of iterations, the Hamming distance is calculated n times to obtain the average Hamming distance. The stability of PUF value can be determined based on This indicates that if the PUF value is within a 20% error correction range of the BCH error correction function, it can be considered stable; n acquisitions are performed on the same length D, using... The minimum entropy is used to represent the randomness of the PUF value as an entropy source.

[0019] As an improvement to the present invention, the entropy accumulation stage in step S2 further includes:

[0020] S21, In the process of generating random number seeds, an entropy accumulator is used to accumulate and extract sufficiently high entropy results from the PUF entropy source according to the principle of message digest function, thus avoiding the design of a complex entropy estimation part;

[0021] S22: Using a sufficient number of values ​​from different iterations under the same PUF as input, calculate the minimum length required for the input data based on the minimum entropy and the length of the true random number seed output.

[0022] S23, use the random number seed as the device key, without having to generate the device key through the original equipment manufacturer;

[0023] S24 employs an improved iterative HMAC-SHA256 algorithm for function computation, offering better security and ensuring that the true random number seed has a sufficiently high entropy value.

[0024] As an improvement to the present invention, the high-entropy random number generation stage in step S3 further includes:

[0025] S31 employs a Hybrid Random Number Generator (HRNG) approach. The principle involves using an entropy accumulator as a True Random Number Generator (TRNG) to obtain a true random number seed, and designing a high-entropy random number generator as a pseudo-random number generator (PRNG). It uses the true random number seed obtained from the TRNG as input and generates secure random numbers efficiently through a specific algorithm. This hybrid random number generator combines the features of TRNG and PRNG. In practical applications, HRNG has the advantage of not being blocked by entropy source refresh compared to TRNG.

[0026] S32, based on the requirements of the trusted security function for high-entropy true random numbers, adopts the improved iterative CTR-AES algorithm with high security, and outputs random number results of different lengths as required.

[0027] Compared with the prior art, the present invention has the following advantages: (1) When the computing and storage resources of IoT terminal devices are limited, the SRAM physical non-cloning function hardware used in the present invention is low-cost and readily available, which facilitates its promotion and deployment in practical applications.

[0028] (2) Using an improved iterative algorithm entropy accumulator, a full-entropy true random number seed was obtained. The random number passed 15 tests of the NIST standard, has higher security, and can be used as a security device key.

[0029] (3) An efficient and secure improved iterative true random number generator is implemented. Since the full-entropy true random number seed is used as input, the resulting random number has better security and can be better used in other security functions such as encrypting sensitive data in practical applications. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the structure of the present invention;

[0031] Figure 2 This is a schematic diagram of the entropy accumulator of the present invention;

[0032] Figure 3 This is a schematic diagram of the random number generator of the present invention;

[0033] Figure 4 A schematic diagram of the noise in a physically non-clonable function fingerprint;

[0034] Figure 5 This is a flowchart of the generation method of the present invention. Detailed Implementation

[0035] To enhance understanding of the present invention, the embodiments will be described in detail below with reference to the accompanying drawings.

[0036] Example 1: The symbols and their definitions in this scheme are shown in Table 1:

[0037] Table 1:

[0038]

[0039] See Figure 1This invention proposes a method for generating high-entropy random numbers in hardware with physically unclonable functions. This method leverages the uniqueness and stability of physically unclonable functions to ensure the security of the physical fingerprint of a trusted module. In IoT terminal devices, widely available SRAM memory is used as the PUF value. This SRAM PUF possesses high uniqueness and stability, ensuring a unique identifier for each IoT terminal device. Furthermore, the unclonability of the SRAM PUF hardware fingerprint guarantees the unclonability of the IoT terminal device's identity.

[0040] The trusted module needs to include true random number generation functionality. In this solution, the trusted module overcomes the difficulty of generating true random numbers by producing pseudo-random numbers using computer algorithms. It constructs an entropy accumulator based on changes in the PUF value caused by physical phenomena in the hardware circuitry and extracts true random number seeds using entropy extraction methods. A high-quality true random number seed is used as the device key, ensuring high security and uniqueness. Furthermore, the true random numbers generated by the generator will be used for other security functions, such as encrypting sensitive data.

[0041] The technical solution disclosed in this invention will be further described in detail below through specific steps:

[0042] like Figure 1 As shown, this paper provides a method for generating high-entropy random numbers in physically non-clonable function hardware, including the following steps:

[0043] (1) In the initial entropy source extraction stage, the initial data is generated by a physical non-cloning function hardware and the resulting 01 bit sequence is initially evaluated.

[0044] Step (1) includes the following sub-steps:

[0045] (1.1) Set challenge values ​​of different ranges and lengths, challenge the memory device and obtain its response value, obtain challenge response pairs, including obtaining PUF values ​​of different address ranges under the same length, and obtaining PUF values ​​of different times under the same address range.

[0046] (1.2) Calculate the Hamming distance n times for each pairwise PUF with different ranges under the same length D, and obtain the average Hamming distance. The uniqueness of PUF values ​​can be determined based on This indicates that, under ideal conditions, the uniqueness value is 50%; for each pair of PUFs of the same length D with different numbers of iterations, the Hamming distance is calculated n times to obtain the average Hamming distance. The stability of PUF value can be determined based on This indicates that if the PUF value is within a 20% error correction range of the BCH error correction function, it can be considered stable; n acquisitions are performed on the same length D, using... The minimum entropy is used to represent the randomness of the PUF value as an entropy source.

[0047] (2) In the entropy accumulation stage, the initial value of the PUF circuit is taken as input by the entropy accumulator. The initial value of the PUF is extracted by entropy accumulation in the entropy accumulator through XOR function, message digest function and other methods. This realizes the extraction of a high-entropy true random number seed from the PUF value, which can be used as a security device key.

[0048] Step (2) includes the following sub-steps:

[0049] (2.1) The first algorithm used is the HMAC-SHA256 algorithm, which produces a 256-bit message digest for plaintext input of any length, and therefore can be used for information entropy accumulation. For example... Figure 2 The diagram illustrates an entropy accumulator based on the HMAC-SHA256 algorithm. The algorithm can be represented as HMAC-SHA256 = (refresh, finalize). In the diagram, F represents HMAC-SHA256(key, message), where x1, x2, ..., x... p It is the original entropy source. The Refresh process means that in the first HMAC-SHA256 algorithm, the key input uses 64 bytes {0x3636……} as internal padding, and in each subsequent round, x... i (1≤i≤p) is used as the input to the key part, with a key length of 512 bits. The input part uses a 0-bit string in the first round, and the input in each subsequent iteration comes from the output of the previous HMAC-SHA256 algorithm. During p iterations, the output can be obtained from x... i Accumulate sufficient entropy. Simultaneously, using a 0-bit string as input, and 64 bytes {0x5C5C……} as external padding, the result obtained after one HMAC-SHA256 algorithm and the output obtained after p iterations are used as input. After a final HMAC-SHA256 algorithm processing, the final true random number result is obtained.

[0050] Assuming the input entropy source has a minimum entropy of 8%, and a 256-bit full entropy seed is required, at least 256 / 8% = 3200 bits of entropy source input are needed. Therefore, in this invention, p = 8 is selected to ensure that the obtained true random number seed accumulates sufficient entropy.

[0051] (2.2) In the finalize phase, the required r bits are truncated from y according to different required lengths. For example, when a 256-bit random number is needed as the device key, the first 256 bits of the 512 bits obtained by the entropy accumulation algorithm are extracted as the secure device key.

[0052] (3) In the random number generation stage, a high-entropy true random number seed is used as input through a random number generator. In the true random number generator, high-entropy random numbers of different lengths are generated as needed by using hash functions, symmetric encryption algorithms, etc.

[0053] Step (3) includes the following sub-steps:

[0054] (3.1) such as Figure 3 The diagram shows the structure of a random number generator, RNG = (setup, refresh, next), which is implemented based on the CTR-AES algorithm. In the setup phase, the initial key is {0, ..., ...} k To improve the randomness of the generated key, we use the high-entropy value TRS generated by the fuzzy extractor as input to generate the key K; similar to the refresh process in the entropy accumulator, this RNG accumulates enough entropy by using TRS as the key for each round of n iterations during the refresh process.

[0055] (3.2) In the next stage, according to the required random number length, truncate the output as the random number R. For example, if a 256-bit random number is needed as the session key, the random number generator algorithm key length is selected as 256 bits. Using 4 message blocks B1 to B4 (128*4=512 bits), we use the 2-time entropy accumulator algorithm to generate a 512-bit true random number seed TRS. During the setup and refresh processes, it is XORed with the ciphertext block, and the first 256 bits are truncated as the secure session key.

[0056] The technical means disclosed in this invention are not limited to the technical means and systems disclosed in the above embodiments. The scope of protection of this invention also includes technical solutions and systems composed of any combination of the above technical features, as well as variations or changes that can be easily conceived within the technical scope described in this invention.

[0057] It should be noted that the above embodiments are not intended to limit the scope of protection of the present invention. Equivalent transformations or substitutions made based on the above technical solutions all fall within the scope of protection of the claims of the present invention.

Claims

1. A method for generating high-entropy random numbers in physically unclonable function hardware, characterized in that, It includes three stages: the initial entropy source extraction stage, the entropy accumulation stage, and the high-entropy random number generation stage. The specific steps are as follows: Step S1: In the initial entropy source extraction stage, the initial data is generated by a physically non-cloning function hardware, and then the obtained 01 bit sequence is evaluated by different mathematical methods. Step S2, in the entropy accumulation stage, multiple initial values ​​read from the PUF circuit are processed by the entropy accumulator. As input, the iterative HMAC-SHA256 algorithm is used in the entropy accumulator to accumulate and extract the entropy of the multiple initial values ​​to obtain a high-entropy true random number seed; Step S3: In the random number generation stage, a high-entropy true random number seed is used as input. In the pseudo-random number generator, the true random number seed is processed by the CTR-AES algorithm, and high-entropy random numbers of different lengths are generated as needed.

2. The method according to claim 1, characterized in that, The initial entropy source extraction stage in step S1 further includes: S11, set challenge values ​​of different ranges and lengths, challenge the memory device and obtain its response value, obtain challenge response pairs, including obtaining PUF values ​​of different address ranges under the same length, and obtaining PUF values ​​of different times under the same address range; S12, composed of the same length Next, perform PUF in pairs of different ranges. The Hamming distance is calculated to obtain the average Hamming distance. The uniqueness of PUF values ​​is based on Indicates; the same length PUF with different numbers of pairs The Hamming distance is calculated to obtain the average Hamming distance d̅. The stability of the PUF value is based on... Indicates; the same length conduct Second collection, using This represents the minimum entropy, used to indicate the degree of randomness of the PUF value as an entropy source.

3. The method according to claim 1, characterized in that... The entropy accumulation stage in step S2 further includes: S21: Collect data from the same PUF hardware. The response values ​​under different challenges were obtained. One original entropy source Each It is 512 bits; The value of is calculated based on the minimum entropy of PUF and the required length of the true random number seed, and is at least ceil(256 / minimum entropy); S22: Employ the iterative HMAC-SHA256 algorithm to accumulate entropy in the following manner: (1) Initialization: ,in It is 64 bytes {0x3636…36}; (2) To to : ; (3) Final result: , in It is 64 bytes {0x5C5C…5C}; Where F represents the HMAC-SHA256 compression function, S23: Use the obtained TRS as a 256-bit true random number seed, i.e., the device key, without having to generate it through the original equipment manufacturer; S24: If different lengths are needed, cut from TRS.