SRAM-PUF pre-selection method and system based on two-dimensional space majority voting
Through the two-dimensional space majority voting algorithm and correlation matrix method, the problem of SRAM-PUF being sensitive to temperature and aging is solved, the stability and efficiency of key generation are improved, the hardware cost is reduced, and the efficient key reconstruction is achieved.
Patent Information
- Application Number
- CN202411135901.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-11-08
AI Technical Summary
Existing SRAM-PUF technology is sensitive to temperature, voltage and aging, resulting in higher error rates and noise, increasing hardware overhead, and increasing hardware costs for small platforms such as microcontrollers and IoT devices.
Using the SRAM-PUF pre-selecting method based on two-dimensional space majority voting, through the registration stage and reconstruction stage, the correlation matrix method and majority voting algorithm are used to generate and reconstruct stable response sequences, output auxiliary data to generate keys, simplify the pre-selecting process and improve the stability.
It improves the stability of PUF response and the success rate of key reconstruction, reduces resource consumption and time, simplifies hardware design, reduces bit error rate, and achieves higher efficiency and robustness compared with the prior art.
Smart Images

Figure CN118921174A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of physical unclonable function encryption, and in particular relates to a SRAM-PUF preselection method and system based on two-dimensional space majority voting. Background Art
[0002] Electronic devices are an integral part of our daily lives, such as smart cards and mobile phones, which are used to verify identity and store private data. Due to personal privacy, these devices are easy targets for attacks, which brings security issues. Most devices contain keys or information for identity verification, which must be stored securely to prevent leakage. Usually, keys are stored in non-volatile memory, but this is vulnerable to invasive attacks, resulting in key leakage. In order to store keys securely, electronic devices must be equipped with additional hardware security architectures to detect attacks. These hardware require additional design support, but adding such hardware will bring a large cost burden to small platforms such as microcontrollers (MCU, Microcontroller Unit) and IoT devices.
[0003] A physical unclonable function (PUF) is a technology that generates a unique output through tiny differences in the manufacturing process of a physical system. These differences make each device unique and cannot be exactly replicated even under the same manufacturing conditions. Once the integrated circuit is manufactured, these physical differences are fixed, and the resulting random output value is also deterministic and repeatable, so it can be used as the identity of the integrated circuit. Physical unclonable technology uses these differences in the manufacturing process to generate a unique and cryptographically usable random output for each integrated circuit.
[0004] Static random-access memory (SRAM) can be used to generate PUF. SRAM-PUF technology uses the initial value of the SRAM cell at power-on as the unique random output of the integrated circuit. The principle is that each SRAM cell has two stable states, representing 0 or 1. When the cell is powered on, the final state is unpredictable, but process differences during the manufacturing process cause small random changes between transistors, making each cell more inclined to 0 or 1. Therefore, the SRAM cells form a random array, which becomes the unique and unclonable output value of the integrated circuit. In recent years, SRAM-PUF has been widely studied as a form of hardware identification and key generation / storage. Its main advantage is that SRAM is common in most systems and no additional hardware is required. However, SRAM-PUF is sensitive to temperature, voltage and aging, and is prone to noise, resulting in a high error rate, so it is necessary to add error correction circuits, which increases hardware overhead. Summary of the invention
[0005] In view of the problems existing in the prior art, the present invention provides a SRAM-PUF preselection method and system based on two-dimensional space majority voting, which can replace the traditional test method for each point and greatly improve the efficiency of extracting PUF stable points.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a SRAM-PUF preselection method based on two-dimensional space majority voting, including a registration phase and a reconstruction phase; the registration phase includes the following steps S1 to S3:
[0007] S1. Input an excitation signal x to the physical unclonable function PUF, where the excitation signal is the address of the SRAM at the specified location;
[0008] S2. Execute the SRAM cell array organization determination method: read the power-on value of the SRAM, generate a probability vector q, and then determine the actual number of SRAM columns M according to the correlation matrix method; obtain an N-bit SRAM-PUF response sequence W according to the excitation signal x;
[0009] S3. Execute the relevant two-dimensional space majority voting algorithm: Based on the actual number of SRAM columns M, the SRAM-PUF response sequence W is modeled as a two-dimensional matrix, and then the matrix is divided into multiple pre-selected matrices, each of which is further divided into multiple sub-pre-selected matrices, and a stable response is extracted from each sub-pre-selected matrix; during the setting process, a majority subgroup is selected from each sub-pre-selected matrix, and then a majority vote is performed on the majority subgroup, and finally the auxiliary data h is output;
[0010] Reconstruction phase: Obtain an N-bit SRAM-PUF response sequence W' based on the excitation signal x, divide the response sequence W' into a matrix based on the actual number of SRAM columns M, model the SRAM-PUF response sequence W' as a two-dimensional matrix, and then divide the matrix into multiple pre-selected matrices, each of which is further divided into multiple sub-pre-selected matrices, and extract a stable response from each sub-pre-selected matrix; combine the voting threshold and auxiliary data h, and output the processed SRAM-PUF response for continuing to generate keys.
[0011] Furthermore, the aforementioned step S2 includes the following sub-steps:
[0012] S2.1 evaluates stimulus x and obtains a power-on value n-bit PUF response w, repeating a preset number of times;
[0013] S2.2. For a preset number of w responses, count the probability q of 1 appearing in each of the n responses. 1 ,q 2 ,...,q n , forming a sequence q = {q 1 ,q2 ,……,q n}, each element qi (i=1,2,...,n) is a decimal with three significant digits between 0 and 1.
[0014] S2.3 The initial matrix column number cl is set to the data bus length of the SRAM, and the number of rows is rl. Arrange the sequence q into a two-dimensional matrix of rl×cl;
[0015] S2.4 sets the FHD threshold val, for each bit q of the n-bit vector q i , where i = 1, 2, ..., n, perform the following steps S2.4.1 to S2.4.4:
[0016] S2.4.1 If q i If it is located at the edge of the current matrix layout, it does not have enough neighboring cells and is ignored and not included in the traversal count;
[0017] S2.4.2 If q i Not on the edge, with q i As the center, get q i If each of the 8-bit neighboring cell vector values in the surrounding 3×3 matrix is less than the threshold val, it is confirmed that a correlation matrix has been found;
[0018] S2.4.3 After all bits are traversed, calculate the percentage of the correlation matrix to the total number of traversals Per cl ;
[0019] S2.4.4 Double the number of columns cl, change the shape, and repeat the above process until cl is greater than 128;
[0020] S2.5. Comparison sequence Per = {Per cl ,Per cl*2 ,……,Per 128}, find the maximum value Per i , at this time i is the actual number of columns M.
[0021] Furthermore, the aforementioned step S3 includes the following sub-steps:
[0022] S3.1. Evaluate the stimulus signal x and obtain the n-bit SRAM-PUF response sequence W.
[0023] S3.2. Arrange the W responses into a M×M matrix vector A: A 1 ,A 2 ,...,A a ,in S3.3. For each A i, where i = 1, ..., a, split it into b m × m matrix vectors B, where m is an odd number greater than 1, and B is: B 1 ,B 2 ,...,B b ,in b=t 2 .
[0024] S3.4, for each B i , where i = 1, ..., b, perform the following steps S3.4.1 to S3.4.2:
[0025] S3.4.1, let b i is the matrix B i The majority of Then b i There are at least k' bits, where k' is the size of the data in most subgroups.
[0026] S3.4.2, let h i is n, which marks the majority subgroup 2 Bit mask, i.e. b i The first k' bits are in B i Position in
[0027] S3.5. Output auxiliary data h=h 1 ,...,h ab .
[0028] Furthermore, the aforementioned reconstruction phase includes the following sub-steps:
[0029] S4.1. Evaluate the stimulus signal x and obtain the n-bit SRAM-PUF response sequence W',.
[0030] S4.2. Arrange the W' response sequence into a M×M matrix vector A: A 1 ,A 2 ,...,A a ,in S4.3. For each A i , where i = 1, ..., a, and it is divided into b m × m matrix vectors B, where m is the order of the sub-preselected matrix, which is an odd number and greater than 1; B is: B 1 ,B 2 ,...,B b ,in b=t 2 , t is the voting threshold;
[0031] S4.4 For each B i , where i = 1, ..., b, perform the following steps:
[0032] S4.4.1. i Set to use h i The number of 1s in the marked subgroup;
[0033] S4.4.2 If t i >t, then set the subgroup flag bit b i =1, otherwise set b i =0;
[0034] S4.5, output b 1 ,...,b ab as the processed PUF response.
[0035] Furthermore, the aforementioned step S2.1 evaluates the stimulus x and obtains the power-on value n-bit PUF response w, and repeats 1000 times to obtain w 1 ,w 2 ,w 3 ,...,w 1000 .
[0036] On the other hand, the present invention proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any one of the methods of the present invention when executing the computer program.
[0037] The present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any one of the methods of the present invention.
[0038] Compared with the prior art, the beneficial technical effects of the above technical solutions adopted by the present invention are as follows: the present invention does not require a pre-selection algorithm to be redesigned for the power supply voltage, has no requirements for the power supply voltage, and does not require the design of additional circuits. The present invention can generate SRAM-PUF responses in a shorter time and resource consumption, and achieves the technical effect of simplifying the screening process of SRAM unit pre-selection while ensuring high stability. Compared with the time majority voting algorithm represented by the pure algorithm in the prior art method, the present invention has higher efficiency, better robustness, and higher key reconstruction success rate. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 It is the algorithm flow chart.
[0040] Figure 2 It is a schematic diagram of the majority voting algorithm in two-dimensional space.
[0041] Figure 3 This is a test result diagram of the SRAM cell array organization determination algorithm.
[0042] Figure 4It is the spatial configuration diagram of the two-dimensional space majority voting algorithm under different t parameters. DETAILED DESCRIPTION
[0043] In order to better understand the technical content of the present invention, specific embodiments are given and described as follows in conjunction with the accompanying drawings.
[0044] Various aspects of the invention are described herein with reference to the accompanying drawings, in which many illustrative embodiments are shown. The embodiments of the invention are not limited to those described in the accompanying drawings. It should be understood that the invention is implemented by any of the various concepts and embodiments described above, as well as the concepts and embodiments described in detail below, because the concepts and embodiments disclosed in the invention are not limited to any implementation. In addition, some aspects disclosed in the invention may be used alone or in any appropriate combination with other aspects disclosed in the invention.
[0045] When a stimulus signal is input to PUF, PUF will output a unique corresponding response signal. The mapping relationship between the stimulus response pair (CRP, Challenge-Response Pair) is determined only by the physical characteristics of PUF itself. Ideally, the response output by PUF under the same stimulus signal is stable and unchanged. However, due to factors such as ambient temperature, power supply and circuit aging, the output response value is not stable, that is, the response value may flip from 0 to 1 during use. The difference between the two responses is called the noise rate.
[0046] Assuming that the various influencing factors of PUF are randomly distributed and each bit of the PUF response is independent, the average noise rate p of PUF is e Also called PUF bit error rate. Usually, the bit error rate of SRAM-PUF is high, between 0.05 and 0.15, and cannot be used directly as a key. It must be pre-selected with stable bits or corrected with error correction codes. This method is also applicable to this process.
[0047] refer to Figure 1 ,The present invention provides a SRAM-PUF pre-selection method based on two-dimensional space majority voting, including a registration phase, and a reconstruction phase;
[0048] The registration phase includes the following steps S1 to S3:
[0049] S1. Input an excitation signal x to the physical unclonable function PUF, where the excitation signal is the address of the SRAM at the specified location;
[0050] S2. Execute the SRAM cell array organization determination method: read the power-on value of the SRAM, generate a probability vector q, and then determine the actual number of SRAM columns M according to the correlation matrix method; obtain an N-bit SRAM-PUF response sequence W according to the excitation signal x;
[0051] S3. Execute the relevant two-dimensional space majority voting algorithm: Based on the actual number of SRAM columns M, the SRAM-PUF response sequence W is modeled as a two-dimensional matrix, and then the matrix is divided into multiple pre-selected matrices, each of which is further divided into multiple sub-pre-selected matrices, and a stable response is extracted from each sub-pre-selected matrix; during the setting process, a majority subgroup is selected from each sub-pre-selected matrix, and then a majority vote is performed on the majority subgroup, and finally the auxiliary data h is output;
[0052] Reconstruction phase: Obtain an N-bit SRAM-PUF response sequence W' based on the excitation signal x, perform matrix division on the response sequence W' in combination with the actual number of SRAM columns M, model the SRAM-PUF response sequence W' as a two-dimensional matrix, and then divide the matrix into multiple pre-selected matrices, each of which is further divided into multiple sub-pre-selected matrices, and extract a stable response from each sub-pre-selected matrix; output the processed SRAM-PUF response in combination with the voting threshold and auxiliary data h, and use the response to continue to generate the key k.
[0053] In step S2, the principle of the SRAM cell array organization determination method is that the mapping address of the SRAM in the program is not necessarily the real mapping of its physical address, which is manifested as two cells adjacent in the memory address are not necessarily adjacent in the physical address. The possible structure can be calculated by analyzing the correlation method. The present invention uses an SRAM cell array organization determination algorithm based on a correlation matrix, which is as follows.
[0054] Collect the n-bit SRAM spatial power-on distribution probability vector q(q 1 ,q 2 ,……,q n ), vector q i The value of indicates that the i-th bit will be i is the probability that the power-on value is 1. If the power-on value of this bit is 1 512 times in 1000 repeated power-on tests, then q i is 0.512. The probability vector q is then modeled as a two-dimensional matrix, where the number of columns is specified by the method and the number of rows is n divided by the number of columns. Changing the number of columns in the matrix will change the shape of the matrix and cause the physical neighbors of the cells to change. The correlation matrix (CM) is defined as some submatrices of the probability vector q matrix. If the submatrices meet some conditions, they can be regarded as a correlation matrix. Under different shapes, if the number of correlation matrices under a certain shape accounts for the largest proportion, then the correlation between the number of rows and columns corresponding to this shape should be closest to the actual arrangement.
[0055] The present invention uses the fractional Hamming distance (FHD) to describe the correlation of the correlation matrix. The Hamming distance (HD) is an indicator used to measure the number of different characters between two strings of equal length. FHD represents the ratio of the number of different characters in the two strings to the total length of the strings. In this method, the probability vector q i , can be approximately regarded as the FHD of the bit in the time dimension. The parameters of the method are the FHD threshold and the number of related neighbor SRAMs. i When the FHD of all related neighboring units on its two-dimensional structure is less than the threshold, it is defined as a positive correlation and is determined to be a correlation matrix. Otherwise, it is not a correlation matrix. All probability vectors q of the cyclic sequence are counted to calculate the correlation matrix ratio. Then the number of columns is changed to continue to obtain the correlation matrix ratio under different numbers of columns. Finally, the number of columns with the largest correlation matrix ratio is obtained, which is M, as the closest actual physical structure arrangement of SRAM.
[0056] Step S2 includes the following sub-steps:
[0057] S2.1 evaluates stimulus x and obtains the power-on value n-bit PUF response w, repeats 1000 times, and obtains w 1 ,w 2 ,w 3 ,...,w 1000 .
[0058] S2.2. For a preset number of w responses, count the probability q of 1 appearing in each of the n responses. 1 ,q 2 ,...,q n , forming a sequence q = {q 1 ,q 2 ,……,q n}, each element qi (i=1,2,...,n) is a decimal with three significant digits between 0 and 1.
[0059] S2.3 The initial matrix column number cl is set to the data bus length of the SRAM, and the number of rows is rl, Arrange the sequence q into a two-dimensional matrix of rl×cl;
[0060] S2.4 sets the FHD threshold val, for each bit q of the n-bit vector q i , where i = 1, 2, ..., n, perform the following steps S2.4.1 to S2.4.4:
[0061] S2.4.1 If q iIf it is located at the edge of the current matrix layout, it does not have enough neighboring cells and is ignored and not included in the traversal count;
[0062] S2.4.2 If q i Not on the edge, with q i As the center, get q i If each of the 8-bit neighboring cell vector values in the surrounding 3×3 matrix is less than the threshold val, it is confirmed that a correlation matrix has been found;
[0063] S2.4.3 After all bits are traversed, calculate the percentage of the correlation matrix to the total number of traversals Per cl ;
[0064] S2.4.4 Double the number of columns cl, change the shape, and repeat the above process until cl is greater than 128;
[0065] S2.5. Comparison sequence Per = {Per cl ,Per cl*2 ,……,Per 128}, find the maximum value Per i , at this time i is the actual number of columns M.
[0066] The built-in SRAM of stm32f407 is used for testing. The chip has a built-in 192KB SRAM. The test area address is 0x2001 0000 to 0x2001 07FF, a total of 2KB, divided into two areas for experiment. First, the array parameter M of the SRAM must be determined. Use the SRAM cell array organization determination algorithm, repeat the power-on 1000 times, and then set the number of columns to 32 (based on the data bus width of stm32f407), and then to 256, and the FHD threshold val is from 0.1 to 0.3 (the change interval is 0.05), and the data of the two areas are averaged. The experimental results are as follows Figure 3 shown.
[0067] refer to Figure 2 , step S3 two-dimensional space majority voting algorithm, including the following sub-steps:
[0068] S3.1. Evaluate the stimulus signal x and obtain the n-bit SRAM-PUF response sequence W.
[0069] S3.2. Arrange the W responses into a M×M matrix vector A: A 1 ,A 2 ,...,A a ,in
[0070] S3.3. For each A i, where i = 1, ..., a, split it into b m × m matrix vectors B, where m is an odd number greater than 1, and B is: B 1 ,B 2 ,...,B b ,in b=t 2 .
[0071] S3.4, for each B i , where i = 1, ..., b, perform the following steps S3.4.1 to S3.4.2:
[0072] S3.4.1, let b i is the matrix B i The majority of Then b i There are at least k' bits, where k' is the size of the data in most subgroups.
[0073] S3.4.2, let h i is n, which marks the majority subgroup 2 Bit mask, i.e. b i The first k' bits are in B i Position in
[0074] S3.5. Output auxiliary data h=h 1 ,...,h ab .
[0075] As a preferred embodiment of the present invention, refer to Figure 4 , the reconstruction phase includes the following sub-steps:
[0076] S4.1. Evaluate the stimulus signal x and obtain the n-bit SRAM-PUF response sequence W',.
[0077] S4.2. Arrange the W' response sequence into a M×M matrix vector A: A 1 ,A 2 ,...,A a ,in
[0078] S4.3. For each A i , where i = 1, ..., a, and it is divided into b m × m matrix vectors B, where m is the order of the sub-preselected matrix, which is an odd number and greater than 1; B is: B 1 ,B 2 ,...,B b ,in b=t 2 , t is the voting threshold;
[0079] S4.4 For each B i, where i = 1, ..., b, perform the following steps:
[0080] S4.4.1. i Set to use h i The number of 1s in the marked subgroup;
[0081] S4.4.2 If t i >t, then set the subgroup flag bit b i =1, otherwise set b i =0;
[0082] S4.5, output b 1 ,...,b ab as the processed PUF response.
[0083] Under the calculation of this method, the optimal number of columns of the built-in unit structure of the chip SRAM is 32. Then, a section of the memory address is selected, and its size is 512B. After obtaining the 4096-bit PUF response. The voting threshold is selected as 6, and the PUF response is divided into 4 32×32 pre-selected blocks. For each pre-selected block, m=5 is selected, and the pre-selected block is divided into 36 5×5 sub-pre-selected blocks, and the excess bits will be ignored. For each sub-pre-selected block, it has a 25-bit binary value. The value greater than 12 is selected (assuming it is 1), and the voting result is 1. The first 13 bits with this value are recorded as the majority sub-group, and its position is recorded and output as help data. Wait until the reconstruction stage, and vote on the value of this position. As long as the flipped bits in the majority sub-group are no more than 7, the voting result remains unchanged and is still 1. The result is output as the key. The 4096-bit SRAM-PUF power-on value, under the setting of this parameter, can generate a maximum of 144 bits of key.
[0084] Based on 40,000 key generation experiments, the stability data of the existing method and the present method in restoring the 128-bit key were tested. The key reconstruction success rate data of the improved present method and the existing method that does not consider the spatial configuration are compared as shown in Table 1: The key reconstruction success rate table of the existing method and the present method is shown in Table 1:
[0085] Table 1
[0086]
[0087] The experimental results are shown in Table 1. When t is 1, 2, and 3, it can be seen that this method has a significant reconstruction stability advantage over the existing method. However, under this t condition, the bit error rate is still high and cannot be used directly. When t = 6, the bit error rate of this method is 1.22×10 -4 , the existing technology method is 6.64×10 -4, which reduces the bit error rate by 82% compared with the existing technology. At the same time, the commonly used time majority voting (TMV) algorithm was tested. When the stable number parameter of TMV was set to 1000, its bit error rate was 3.9×10 -3 Compared with them, this method has advantages in bit error rate correction. In summary, this method can be effectively used for SRAM-PUF key generation.
[0088] Although the present invention has been described above with preferred embodiments, it is not intended to limit the present invention. A person skilled in the art of the present invention may make various modifications and improvements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the definition of the claims.
Claims
1. A SRAM-PUF pre-selection method based on two-dimensional space majority voting, characterized in that: It includes a registration phase and a reconstruction phase; the registration phase includes the following steps S1 to S3: S1. Input an excitation signal x to the physical unclonable function PUF, where the excitation signal is the address of the SRAM at the specified location; S2. Execute the SRAM cell array organization determination method: read the power-on value of the SRAM, generate a probability vector q, and then determine the actual number of SRAM columns M according to the correlation matrix method; obtain an N-bit SRAM-PUF response sequence W according to the excitation signal x; S3. Execute the relevant two-dimensional space majority voting algorithm: Based on the actual number of SRAM columns M, the SRAM-PUF response sequence W is modeled as a two-dimensional matrix, and then the matrix is divided into multiple pre-selected matrices, each of which is further divided into multiple sub-pre-selected matrices, and a stable response is extracted from each sub-pre-selected matrix; during the setting process, a majority subgroup is selected from each sub-pre-selected matrix, and then a majority vote is performed on the majority subgroup, and finally the auxiliary data h is output; Reconstruction phase: Obtain an N-bit SRAM-PUF response sequence W' based on the excitation signal x, divide the response sequence W' into a matrix based on the actual number of SRAM columns M, model the SRAM-PUF response sequence W' as a two-dimensional matrix, and then divide the matrix into multiple pre-selected matrices, each of which is further divided into multiple sub-pre-selected matrices, and extract a stable response from each sub-pre-selected matrix; combine the voting threshold and auxiliary data h, and output the processed SRAM-PUF response for continuing to generate keys.
2. A SRAM-PUF pre-selection method based on two-dimensional space majority voting according to claim 1, characterized in that: Step S2 includes the following sub-steps: S2.1 evaluates stimulus x and obtains a power-on value n-bit PUF response w, repeating a preset number of times; S2.
2. For a preset number of w responses, count the probability of 1 appearing in each of the n-bit responses, q1,q2,...,q n , forming a sequence q={q1,q2,……,q n }, each element qi (i=1,2,...,n) is a decimal with three significant digits between 0 and 1. S2.3 The initial matrix column number cl is set to the data bus length of the SRAM, and the number of rows is rl, Arrange the sequence q into a two-dimensional matrix of rl×cl; S2.4 sets the FHD threshold val, for each bit q of the n-bit vector q i , where i = 1, 2, ..., n, perform the following steps S2.4.1 to S2.4.4: S2.4.1 If q i If it is located at the edge of the current matrix layout, it does not have enough neighboring cells and is ignored and not included in the traversal count; S2.4.2 If q i Not on the edge, with q i As the center, get q i If each of the 8-bit neighboring cell vector values in the surrounding 3×3 matrix is less than the threshold val, it is confirmed that a correlation matrix has been found; S2.4.3 After all bits are traversed, calculate the percentage of the correlation matrix to the total number of traversals Per cl ; S2.4.4 Double the number of columns cl, change the shape, and repeat the above process until cl is greater than 128; S2.5, comparison sequence Per = {Per cl ,Per cl*2 ,……,Per 128 }, find the maximum value Per i , at this time i is the actual number of columns M.
3. The SRAM-PUF pre-selection method based on two-dimensional space majority voting according to claim 1 is characterized in that: Step S3 includes the following sub-steps: S3.
1. Evaluate the stimulus signal x and obtain the n-bit SRAM-PUF response sequence W. S3.
2. Arrange the W responses into a M×M matrix vector A: A1, A2, ..., A a ,in S3.
3. For each A i , where i = 1, ..., a, split it into b m × m matrix vectors B, where m is an odd number greater than 1, and B is: B1, B2, ..., B b ,in b=t 2 . S3.4, for each B i , where i = 1, ..., b, perform the following steps S3.4.1 to S3.4.2: S3.4.1, let b i is the matrix B i The majority of Then b i There are at least k' bits, where k' is the size of the data in most subgroups. S3.4.2, let h i is n, which marks the majority subgroup 2 Bit mask, i.e. b i The first k' bits are in B i Position in S3.
5. Output auxiliary data h=h1,...,h ab .
4. The SRAM-PUF pre-selection method based on two-dimensional space majority voting according to claim 1 is characterized in that: The reconstruction phase consists of the following sub-steps: S4.
1. Evaluate the stimulus signal x and obtain the n-bit SRAM-PUF response sequence W',. S4.
2. Arrange the W' response sequence into a M×M matrix vector A: A1, A2, ..., A a ,in S4.
3. For each A i , where i = 1, ..., a, and it is divided into b m × m matrix vectors B, where m is the order of the sub-preselected matrix, which is an odd number and greater than 1; B is: B1, B2, ..., B b ,in b=t 2 , t is the voting threshold; S4.4 For each B i , where i = 1, ..., b, perform the following steps: S4.4.
1. i Set to use h i The number of 1s in the marked subgroup; S4.4.2 If t i >t, then set the subgroup flag bit b i =1, otherwise set b i =0; S4.5, output b1,...,b ab as the processed PUF response.
5. The SRAM-PUF pre-selection method based on two-dimensional space majority voting according to claim 2, characterized in that: Step S2.1 evaluates the stimulus x and obtains the power-on value n-bit PUF response w, repeats 1000 times, and obtains w1, w2, w3, ..., w 1000 .
6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.
7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.
Citation Information
Cited By
SRAM-PUF pre-selection method and device, medium and product
CN122417107A
