A multiplexing architecture for a soft-decision SRAM PUF based on ECDSA cryptography protection

The soft-decision SRAMPUF multiplexing architecture protected by ECDSA cryptography solves the problems of original response leakage and auxiliary data tampering, realizes the synergistic integration of PUF and elliptic curve cryptography, improves authentication security and hardware resource utilization, and is suitable for security chips and trusted terminals.

CN123122939APending Publication Date: 2026-11-17BEIJING INST OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611355475.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2026-04-21
Filing Date
2026-09-03
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing soft-decision SRAMPUF solutions have the risk of original response leakage, auxiliary data is easily tampered with, the error correction process has poor controllability, and the PUF and elliptic curve cryptography modules are implemented separately, resulting in low hardware resource utilization and difficulty in meeting the integration requirements of security chips.

Method used

The soft-decision SRAMPUF reuse architecture based on ECDSA cryptographic protection is adopted. By sharing the underlying point operation module, combining the ECDH encryption module and the ECDSA signature verification module, and introducing the ECDSA signature protection mechanism, the data path and control organization are unified, realizing the integrated reuse of PUF authentication and elliptic curve cryptography.

Benefits of technology

It improves the integrity and reliability of auxiliary data, maintains the non-cloning nature of PUF, enhances resistance to reverse analysis, reduces hardware complexity, improves resource utilization, and enhances authentication security and system integration.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122939A_ABST
    Figure CN123122939A_ABST
Patent Text Reader

Abstract

A multiplexing architecture for a soft-decision SRAMPUF based on ECDSA cryptographic protection belongs to the fields of integrated circuit security, hardware cryptography, and chip authentication technology. It includes an input APB module, an AHB module, an SRAM module, an ECDH encryption module, an ECDSA signature verification module, a PUF module, and an output APB module. The input APB module can input the public key and base point to the ECDH encryption module, and can input the signature, public key, and base point to the ECDSA signature verification module. The input APB module can also send corresponding control signals to the ECDSA signature verification module, granting it control over the point operation module. This invention effectively improves the integrity and reliability of key auxiliary data in the soft-decision SRAMPUF, suppressing cloning attacks based on auxiliary data tampering at the source.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the fields of integrated circuit security, hardware cryptography and chip identity authentication technology, specifically a multiplexing architecture of soft-decision SRAMPUF based on ECDSA cryptographic protection. Background Technology

[0002] Physically Unclonable Functions (PUFs) are a hardware security technique that utilizes unavoidable random process variations during chip manufacturing to generate unique device responses. SRAMPUFs, in particular, directly utilize the initial power-on values ​​of SRAM cells as the chip's intrinsic fingerprint, requiring minimal additional hardware overhead. Therefore, they offer advantages such as simple structure, low implementation cost, and ease of integration with existing digital circuits, making them a commonly used PUF implementation method.

[0003] However, the initial power-on value of SRAM cells is affected by factors such as noise, temperature, voltage fluctuations, and process disturbances. The original responses obtained from different power-on cycles are usually not completely consistent. Therefore, existing technologies typically require error-correcting codes and auxiliary data to recover and correct the original responses. With the development of related research, soft-decision SRAMPUF, which incorporates error probability information into the recovery process, has gradually become the mainstream approach. This type of method improves the error correction success rate and response stability by introducing reliability information of each bit response during the recovery stage, and has better robustness and practicality compared to hard-decision schemes.

[0004] Meanwhile, in security chips, trusted terminals, and embedded cryptographic devices, PUFs not only perform device identity generation and authentication functions, but also often need to work in conjunction with cryptographic algorithm modules such as digital signatures and key negotiation. Therefore, in practical applications, soft-decision SRAMPUFs not only face the challenges of response recovery accuracy and security, but also the engineering requirement of how to efficiently collaborate with elliptic curve cryptography algorithms.

[0005] However, the existing soft-decision SRAMPUF scheme still has the following obvious problems: The raw SRAMPUF response is at risk of leakage, and the traditional premise of "keeping the raw response confidential by default" is unreliable in practical engineering. In many real-world chip systems, SRAM may be connected to the CPU, bus, or other accessible modules. Attackers may obtain the initial SRAM power-on value through software readout, debugging interfaces, probe testing, or even semi-invasive or invasive methods. In other words, many existing SRAMPUF solutions use "raw response unknowable" as a security premise, but this premise is often difficult to uphold in actual deployment.

[0006] The auxiliary data and error probability information introduced by soft-decision schemes can become new attack entry points, thereby compromising the unclonability of PUFs. Existing soft-decision SRAM PUF schemes based on error correction mechanisms such as RM codes typically store auxiliary data and error probability information related to the reliability of each bit in non-volatile memory to improve recovery accuracy. Although this information is intended to improve the recovery success rate, once an attacker obtains or tampers with this data, it may induce the error correction process to converge in the wrong direction, causing the recovered output to deviate from the original device response, thereby enabling the forgery or cloning of the PUF response. In other words, while soft-decision mechanisms enhance reliability, they also weaken the core "unclonability" property of PUFs to some extent.

[0007] Even with simple signature protection for auxiliary data or error probabilities, the "single-bit decision signal" remains a target for attacks. To prevent tampering with auxiliary data or error probabilities stored in non-volatile memory such as FLASH, an intuitive approach is to introduce a digital signature mechanism, such as using ECDSA to sign the relevant data and performing signature verification during the recovery phase. This approach can, to some extent, prevent attackers from forging legitimate data because they cannot obtain the signing private key.

[0008] However, if a conventional ECDSA signature verification method is used, it typically outputs a single-bit decision signal similar to "verification passed / failed". If this single-bit signal directly determines the validity of the PUF output, it forms a relatively critical and highly centralized security control point in the circuit. Once an attacker locates and manipulates this signal through netlist analysis, layout analysis, fault injection, or partial tampering, they could bypass the signature verification logic, allowing erroneous or forged data to still be received by the PUF recovery process, thereby compromising the security of the entire authentication mechanism. In other words, the existing "signature protection + single-bit on / off control" structure still has significant vulnerabilities at the hardware implementation level.

[0009] Existing solutions lack a balance between algorithmic security and hardware implementation complexity, making it difficult to meet the integration requirements of security chips. For scenarios such as security chips and trusted terminals, PUF algorithms not only need to be theoretically feasible but also need to be efficiently mapped into hardware circuits to achieve collaborative operation with elliptic curve cryptography modules, storage modules, and control logic. Adding a separate, complete, and independent signature protection logic would incur additional overhead in terms of area, timing, and implementation complexity; while oversimplification could easily create new security vulnerabilities. Therefore, how to achieve a deeper integration of digital signature verification and the SRAMPUF soft-decision recovery process without significantly increasing hardware redundancy remains a problem that current technologies have not yet adequately solved.

[0010] Existing technologies lack an integrated, reusable hardware architecture for PUF authentication and elliptic curve cryptography collaborative processing.

[0011] Existing SRAMPUF schemes mostly focus on the response recovery and authentication process itself, while elliptic curve algorithms such as ECDH and ECDSA are usually implemented by independent cryptographic modules. These two approaches are generally separated in hardware implementation, lacking a unified module organization and underlying computational resource reuse mechanism. This not only increases chip area and interface control complexity but also hinders the efficient deployment of ECDSA-based soft-decision SRAMPUF authentication methods into secure chips. Therefore, it is necessary to propose an integrated, multiplexed hardware architecture that can simultaneously support PUF authentication and elliptic curve cryptography.

[0012] In summary, while existing soft-decision SRAMPUF technology has achieved some improvements in response stability and error correction recovery capabilities, it still suffers from significant security vulnerabilities when facing auxiliary data tampering, error probability forgery, and hardware-level bypass attacks. In particular, when a single-bit control node that can be located and modified exists in the authentication chain, traditional schemes struggle to fully guarantee the authenticity and non-cloning nature of the PUF authentication result. Furthermore, in existing technologies, PUF authentication and elliptic curve cryptography functions are typically implemented separately, lacking a unified, reusable hardware architecture, making it difficult to balance authentication security, hardware resource utilization, and system integration efficiency. Therefore, it remains necessary to propose a technical solution that can both enhance the security of soft-decision SRAMPUF authentication and achieve the collaborative integration of PUF authentication and elliptic curve cryptography. Summary of the Invention

[0013] This invention provides a multiplexing architecture for soft-decision SRAMPUF based on ECDSA cryptographic protection to address the shortcomings of existing technologies.

[0014] This invention is achieved through the following technical solution: A multiplexing architecture for soft-decision SRAMPUF based on ECDSA cryptographic protection includes an input APB module, an AHB module, an SRAM module, an ECDH encryption module, an ECDSA signature verification module, a PUF module, and an output APB module. The input APB module can input the public key and base point into the ECDH encryption module. The input APB module can input the signature, public key and base point into the ECDSA verification module. The input APB module can send corresponding control signals to the ECDSA verification module to give the ECDSA verification module control over the point operation module. The AHB module can input messages to the ECDSA signature verification module, and the AHB module can send signals to the FLASH to read data and output it. The SRAM module can output signals to the PUF module and interact with the PUF register. The ECDH encryption module receives the public key and base point from the APB module, the plaintext and random number from the AES module, completes ECDH encryption, and outputs the ciphertext to the output APB module. The ECDSA verification module verifies the signature, public key, and base point transmitted by the APB module and the message transmitted by the AHB module. It initiates a custom ECDSA verification request by inputting the APB module, transmits the corresponding control signals to enable the ECDSA verification module to have control of the dot operation module, completes the ECDSA verification, and outputs the verification result to the output APB module. The PUF module can receive read data from FLASH, data from the counter, output data from the SRAM module, and data from the PUF register; the PUF module can output a busy high-level signal to the ECDSA signature verification unit; the PUF module outputs data to an XOR gate; the ECDSA signature verification module outputs data to an XOR gate; and the XOR gate outputs a PUF value to the ECDSA signature verification unit. The output APB module receives the ciphertext output by the CDH encryption module and the signature verification result output by the ECDSA signature verification module.

[0015] The multiplexing architecture of SRAMPUF based on ECDSA cryptographic protection, as described above, uses a shared underlying point operation module for the ECDH encryption module and the ECDSA signature verification module, with control allocated through a multiplexer. Each module executes its respective top-level encryption algorithm by calling the underlying point operation module.

[0016] The multiplexing architecture of the soft-decision SRAMPUF based on ECDSA cryptographic protection described above uses an "instruction ROM + RAM + ALU" architecture for the ECDH encryption module and the ECDSA signature verification module. The instruction ROM stores the control signals for the ECDH encryption algorithm and the custom ECDSA signature verification algorithm. Each algorithm has two corresponding instruction ROMs, which are selected by switching signals. The RAM caches all variables and changes the read and write addresses according to the control signals in the instruction ROM. The write data is selected according to the multiplexer. The ALU can perform all calculations in the elliptic curve arithmetic layer and the modular arithmetic layer. The calculation type is controlled by the instruction ROM. The ALU result can be output outside the module or written back to RAM.

[0017] As described above, in the soft-decision SRAMPUF multiplexing architecture based on ECDSA cryptographic protection, the ECDH encryption module and the ECDSA signature verification module each have their own unique parts responsible for data input and output and for completing their own unique operations. The data in the unique parts will eventually be written into the shared RAM. The switching signal controls the RAM write multiplexer to control the currently running algorithm. When calculating the ECDH encryption algorithm, the RAM data is written from the ECDH encryption-specific part. The ECDH instruction ROM address starts to increment automatically. A counter is used to concatenate plaintext to expand the redundant values ​​in the plaintext embedding algorithm. All steps in the remaining ECDH encryption algorithm and plaintext embedding algorithm can be decomposed into elliptic curve point operations and modular operations. Therefore, all data is written to RAM through a multiplexer, and all calculations are completed by the point operation module. The instruction output from the instruction ROM controls the address written to RAM and the calculation type of the point operation module. When calculating the custom ECDSA signature verification algorithm, the RAM write data comes from the ECDSA signature verification-specific part. The ECDSA instruction ROM address starts to increment automatically. First, the unique HASH module is used to calculate z. The input is selected through a multiplexer. If the second layer of the triple hybrid encryption system is being calculated, the PUF value and the software program read from FLASH are concatenated as a message. Depending on whether the user is in the user registration stage or the user is in the usage stage, the developer's fixed public key or the user's authorized public key is selected. Then, the unique modular multiplier is used to calculate u1 and u2. The modular multiplication calculation is only performed twice in the algorithm. The modular multiplication calculation time is less than that of the dot product calculation. The modular multiplier does not need high bit width and high speed. Instead, a low bit width and low area design should be adopted. A 41-bit multiplier is used to perform 163 multiplications in segments and multiple cycles. A subtractor is used to repeatedly subtract to reduce the modulo n.

[0018] The above describes a multiplexing architecture for a soft-decision SRAMPUF based on ECDSA cryptography protection, wherein the PUF module includes an auxiliary data submodule and an RM code soft-decision submodule. The auxiliary data section includes XOR gates at the input and output of the arithmetic logic unit, a comparator at the output of the arithmetic logic unit, and a lookup table used to simplify the calculation of the likelihood estimate. The three XOR gates correspond to the three XOR operations in the algorithm's recovery phase. The comparator converts the calculated likelihood estimate into codeword c. The lookup table simplifies the calculation of the formula in the likelihood estimate through mapping. Hardware is not easy to calculate The logarithm of the error probability P in FLASH memory is allocated 6 bits of storage space. P has a finite of 64 values, so a pre-calculated lookup table is used to map each value of P to the corresponding value. To achieve fast and convenient calculations, the mapping relationship of the lookup table is deliberately changed to random order when using the lookup table; The RM code soft decision submodule includes an ALU, an address generator, a stack, and a state machine. The ALU provides the F function, G function, and accumulation calculation functions in the soft decision algorithm, and can also output directly without performing any operation. The address generator can generate SRAM read and write addresses according to the stack and state machine control. The stack and state machine are used to complete the control logic for recursive operations.

[0019] As described above, in a multiplexing architecture of a soft-decision SRAMPUF based on ECDSA cryptographic protection, the data path of the auxiliary data submodule includes: FLASH output w, x P Z, P, SRAM output y′; XOR gate inputs w and x P Z obtains w′, and the XOR gate inputs y′, w′, and x. P Z yields c′; the lookup table input P yields... The likelihood estimate L, formed by c′, is used as the input to the ALU to execute a recursive algorithm; the comparator input is the result of the recursive algorithm. Once the correct codeword c is obtained, it is input into an XOR gate along with w′. The output is then XORed with the signature verification result X from outside the module to obtain the ECDSA-PUF value y.

[0020] The advantages of this invention are: (1) The present invention can effectively improve the integrity and reliability of key auxiliary data of soft decision SRAMPUF, and suppress cloning attacks based on auxiliary data tampering from the source; Traditional RM code soft-decision SRAMPUF requires storing auxiliary data and error probability information in non-volatile memory to achieve stable recovery. However, once this data is read and tampered with by an attacker, the error correction process may be artificially guided to incorrect codewords, causing the PUF output to deviate from the actual device response and compromising "unclonability." To address this issue, this invention introduces an ECDSA signature protection mechanism for the auxiliary data and error probability information stored in FLASH. Since the signature is generated by a private key, and the recovery phase requires verification of the stored data, even if an attacker can access or modify the relevant content in FLASH, it is difficult to simultaneously construct a legitimate signature corresponding to the tampered data. Therefore, this invention transforms the auxiliary information, which could be arbitrarily rewritten, into protected data that "can only participate in the recovery after being confirmed by a legitimate signature." This significantly improves the integrity and reliability of key parameters in the soft-decision SRAMPUF recovery chain and reduces the risk of PUF forgery based on auxiliary data tampering. (2) The present invention can maintain the core attribute of PUF being "unclonable" and make up for the damage to the inherent security of PUF caused by existing soft decision schemes; The most important value of PUF technology lies in the fact that even if an attacker observes a response result, it is difficult to replicate the same physical characteristics by creating differences. However, existing soft-decision schemes rely on error probability information for recovery, giving attackers a new manipulation entry point. They can force the error correction output to deviate in the direction the attacker expects by modifying the error probability. In other words, while traditional soft-decision mechanisms improve the recovery success rate, they weaken the fundamental unclonability of PUF to some extent. This invention protects auxiliary data and error probabilities with signatures and incorporates signature verification into the recovery process, making it difficult for attackers to forge the target PUF output by simply tampering with stored data. Thus, this invention not only improves the security of the recovery process, but more importantly, it repairs the damage to the "unclonable" property of PUF caused by soft-decision mechanisms, allowing SRAMPUF to regain a trustworthy foundation as a basis for hardware identity authentication. (3) This invention avoids the security vulnerability of the traditional "signature verification pass / fail" single-bit control method and improves the ability to resist reverse engineering and bypass attacks; If conventional methods are used to perform ECDSA verification on FLASH data and a single-bit pass signal is used to control the validity of the PUF output, this single-bit signal will become the most critical and vulnerable control node in the entire authentication chain. Once an attacker finds and manipulates this node through netlist analysis, layout location, fault injection, or partial tampering, even if the front-end verification logic itself is not compromised, authentication constraints may still be bypassed, resulting in the receipt of forged data. This invention does not use this simple single-bit control method, but instead mixes the relevant intermediate results in ECDSA verification with the PUF recovery process, so that "whether the verification is correct" is no longer directly determined by an isolated control line, but is reflected in the calculation results of the subsequent recovery chain. It can be deduced that this invention disperses the originally highly concentrated single-point control risk into a more complex data operation relationship, increasing the difficulty for attackers to locate key nodes and implement hardware bypass attacks, thus having stronger resistance to reverse analysis and tampering. (4) The present invention improves security while avoiding the introduction of excessive additional hardware costs by integrating the signature verification process and the recovery process. Adding a completely independent signature protection unit and setting up additional verification, judgment, and control logic outside the PUF recovery process, while achieving "signature protection" in form, would incur additional overhead in terms of area, timing, and control complexity, reducing the cost-effectiveness of security chip integration. This invention does not employ a simple superposition of a "PUF module + independent external signature verification module," but rather directly incorporates the ECDSA signature verification-related calculation results into the PUF recovery process, creating a closer link between the signature verification function and the recovery mechanism. In this way, signature protection is no longer an isolated external unit but becomes part of the authentication method itself. This enhances security without significantly increasing redundant logic, balancing authentication strength and implementation complexity, and offering better engineering feasibility. (5) The optimized ECDSA-PUF fast algorithm of this invention takes into account both authentication security and operating efficiency, and is more suitable for actual chip deployment; In the ECDSA-based soft-decision SRAMPUF scheme, strictly following the conventional ECDSA signature verification process introduces additional computational delays, impacting authentication speed and system startup efficiency. To address this, this invention proposes an optimized ECDSA-PUF fast algorithm. By preprocessing and recombining key intermediate quantities, it improves authentication efficiency while ensuring the validity of the authentication relationship and maintaining security. The logic is that if certain intermediate quantities can be pre-calculated or redundant calculations can be reduced through equivalent transformations, the real-time computation required in the recovery phase decreases, resulting in shorter authentication latency. Therefore, this invention not only enhances security but also balances startup speed and implementation efficiency, making it more suitable for deployment in latency-sensitive security chips, trusted terminals, and embedded systems. (6) This invention realizes the integrated reuse of PUF authentication and elliptic curve cryptography, which significantly improves the utilization of hardware resources; Existing implementations typically design the PUF authentication module separately from elliptic curve cryptography modules such as ECDH and ECDSA, leading to redundant configuration of underlying computing resources, increased chip area, and interface control complexity. The proposed ECDH-ECDSA-PUF reused hardware architecture allows the fast ECDSA-PUF algorithm and the custom ECDSA signature verification algorithm to reuse the same ECDSA signature verification module. Furthermore, it allows the ECDSA signature verification module and the ECDH encryption module to reuse the same ECC point operation module. From this, it can be directly deduced that the multiple sets of high-cost modules that originally required separate configurations are unified and converged onto the same set of underlying resources in this invention, thereby reducing the independent configuration requirements of a set of ECDSA signature verification modules and a set of ECC point operation modules. Since these two types of modules constitute a large portion of the area in elliptic curve cryptography algorithms, this invention can significantly reduce the overall hardware area, improve resource reuse, and enhance structural compactness. (7) The unified data path and control organization method of this invention reduces the difficulty of system integration and improves the scalability of the architecture; The multiplexing hardware architecture of this invention does not simply place multiple functional modules side by side. Instead, it organizes the execution process of different algorithms through multiplexers, control signals, and a unified data path. In the ECC-related parts, it adopts a CPU-like structure of "instruction ROM + RAM + ALU" to complete step-by-step operations. Its advantage is that although the multiple algorithms have different functions, their underlying resource calling methods are consistent, and data can be uniformly written to RAM and scheduled through addresses and control signals. As a result, this invention can reduce the additional logical overhead caused by complex and huge state machines, make the module boundaries clearer, and make the interface scheduling more unified. This reduces the complexity of design and verification and provides a good foundation for subsequent expansion of other elliptic curve protocols or authentication processes. (8) This invention is beneficial to improving the overall integration of security chips, and further expands SRAMPUF from "key material" to "trusted authentication mechanism"; Traditional SRAMPUF is often viewed as a key derivation or key recovery method. However, in practical engineering, its direct use as a trusted authentication basis is significantly limited due to the risk of original response leakage and soft-decision attacks. This invention, by introducing ECDSA protection, a recovery link fusion design, and a reused hardware architecture, enables SRAMPUF to move beyond the fragile premise of "the original response must be kept confidential" and complete identity authentication in more complex environments closer to real-world engineering scenarios. Thus, this invention substantially expands the application boundaries of SRAMPUF, making it more reliable for use in security chips for device authentication, trusted boot, and anti-cloning protection, thereby enhancing the practical value of SRAMPUF in engineering. (9) The present invention forms an overall solution that takes into account both "method security" and "hardware reusability", and its overall performance is better than the existing discrete implementation approach; Existing technologies often only offer localized optimizations in a single dimension, such as improving only the reliability of soft-decision recovery, implementing only ECDSA protection alone, or pursuing only hardware reuse. However, these decentralized improvements cannot simultaneously address the three types of problems: data tampering, hardware bypass, and modular implementation. This invention organically combines a soft-decision SRAMPUF authentication method based on ECDSA protection, a fast algorithm, and an ECDH-ECDSA-PUF reuse hardware architecture. This allows security enhancement measures to directly serve engineering implementation, while the hardware reuse design, in turn, supports the low-overhead deployment of the authentication method. Therefore, the advantage of this invention is not an improvement on a single metric, but rather a comprehensive enhancement of authentication security, resource utilization, system integration, and engineering deployability. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the structure of the present invention; Figure 2 This is a schematic diagram of the structure of the ECDH encryption module and the ECDSA signature verification module of the present invention; Figure 3 This is a schematic diagram of the PUF module of the present invention; Figure 4 This is a schematic diagram of the RM code soft decision state machine of the present invention; Figure 5 This is a schematic diagram of the FLASH address and bit allocation of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] like Figure 1 As shown, a multiplexing architecture of soft-decision SRAM PUF based on ECDSA cryptographic protection includes an input APB module, an AHB module, an SRAM module, an ECDH encryption module, an ECDSA signature verification module, a PUF module, and an output APB module. The input APB module can input the public key and base point into the ECDH encryption module. The input APB module can input the signature, public key and base point into the ECDSA verification module. The input APB module can send corresponding control signals to the ECDSA verification module to give the ECDSA verification module control over the point operation module. The AHB module can input messages to the ECDSA signature verification module, and the AHB module can send signals to the FLASH to read data and output it. The SRAM module can output signals to the PUF module and interact with the PUF register. The ECDH encryption module receives the public key and base point from the APB module, the plaintext and random number from the AES module, completes ECDH encryption, and outputs the ciphertext to the output APB module. The ECDSA verification module verifies the signature, public key, and base point transmitted by the APB module and the message transmitted by the AHB module. It initiates a custom ECDSA verification request by inputting the APB module, transmits the corresponding control signals to enable the ECDSA verification module to have control of the dot operation module, completes the ECDSA verification, and outputs the verification result to the output APB module. The PUF module can receive read data from FLASH, data from the counter, output data from the SRAM module, and data from the PUF register; the PUF module can output a busy high-level signal to the ECDSA signature verification unit; the PUF module outputs data to an XOR gate; the ECDSA signature verification module outputs data to an XOR gate; and the XOR gate outputs a PUF value to the ECDSA signature verification unit. The output APB module receives the ciphertext output by the CDH encryption module and the signature verification result output by the ECDSA signature verification module.

[0025] Preferably, in this embodiment, the ECDH encryption module and the ECDSA signature verification module share a common underlying point operation module, and control is allocated using a multiplexer. Each module executes its respective top-level encryption algorithm by calling the underlying point operation module.

[0026] like Figure 2 As shown, preferably, the ECDH encryption module and ECDSA signature verification module described in this embodiment use an "instruction ROM + RAM + ALU" architecture. The instruction ROM stores the control signals for the ECDH encryption algorithm and the custom ECDSA signature verification algorithm. Each algorithm has two corresponding instruction ROMs, which are selected by switching signals. The RAM caches all variables and changes the read / write address according to the control signals in the instruction ROM. Data writes are selected based on the multiplexer. The ALU can perform all calculations in the elliptic curve arithmetic layer and the modular arithmetic layer. The calculation type is controlled by the instruction ROM. The ALU result can be output outside the module or written back to RAM. The "instruction ROM + RAM + ALU" hardware structure saves the complex and huge state machine required to implement the algorithm, and all data is uniformly written to RAM and scheduled through RAM addresses, reducing the logic circuit overhead of the data path.

[0027] Preferably, in this embodiment, the ECDH encryption module and the ECDSA signature verification module have their own unique parts responsible for data input and output and for completing their own unique calculations. The data of the unique parts will eventually be written into the shared RAM. The switching signal controls the RAM data writing multiplexer to control the currently running algorithm. When calculating the ECDH encryption algorithm, the RAM data is written from the ECDH encryption-specific part. The ECDH instruction ROM address starts to increment automatically. A counter is used to concatenate plaintext to expand the redundant values ​​in the plaintext embedding algorithm. All steps in the remaining ECDH encryption algorithm and plaintext embedding algorithm can be decomposed into elliptic curve point operations and modular operations. Therefore, all data is written to RAM through a multiplexer, and all calculations are completed by the point operation module. The instruction output from the instruction ROM controls the address written to RAM and the calculation type of the point operation module. When calculating a custom ECDSA signature verification algorithm, RAM write data comes from the ECDSA signature-specific part. The ECDSA instruction ROM address starts incrementing. First, a unique HASH module is used to calculate z. The input is selected via a multiplexer. If the second layer of a triple hybrid encryption system is being calculated, the PUF value and the software program read from FLASH are concatenated as a message. Depending on whether the user is currently in the user registration or user usage phase, the developer's fixed public key or the user's authorized public key is selected. Then, a unique modular multiplier is used to calculate u1 and u2. The modular multiplication is performed only twice in the algorithm. Modulo-n multiplication takes less time than dot-matrix multiplication. Modulo-n multipliers do not require high bit width and high speed; instead, they should adopt a low bit width and low area design. A 41-bit multiplier is used to perform 163 multiplications in segments over multiple cycles. A subtractor is used to repeatedly subtract and reduce the modulo-n value. All remaining steps can be decomposed into elliptic curve point operations and modular operations. Therefore, all data is written to RAM through a multiplexer, and all calculations are completed by the point operation module. The address written to RAM and the calculation type of the point operation module are controlled by instructions output from the instruction ROM. Finally, the point operation module outputs X and xPZ, and the verification result is obtained through a comparator.

[0028] Since the PUF generation speed depends on which is slower, the RM code soft decision or the ECDSA signature verification, and the RM code soft decision hardware already uses the fastest design, if the ECDSA signature verification calculation speed is faster than the speed of four RM code soft decisions, it will not improve the overall PUF generation speed. On the contrary, the area added to improve speed will increase the overall resource consumption. Therefore, the ECDSA signature verification should be designed with the calculation time of four RM code soft decisions as the target, so that the point operation module can just meet the speed requirement, to achieve the optimal match between speed and area.

[0029] Since most of the computation time of ECDSA is spent on dot product calculations, the speed of dot product calculations can affect the overall ECDSA operation. Since PUF requires 40960 cycles to complete the calculation, the ideal state for ECDSA is to complete the calculation synchronously. Too fast will waste resources, while too slow will affect the overall computation.

[0030] like Figure 3 As shown, preferably, the PUF module described in this embodiment includes an auxiliary data submodule and an RM code soft decision submodule; The auxiliary data section includes XOR gates at the input and output of the Arithmetic Logic Unit (ALU), a comparator at the output of the ALU, and a lookup table used to simplify the calculation of the likelihood estimate. The three XOR gates correspond to the three XOR operations in the algorithm's recovery phase. The comparator converts the calculated likelihood estimate into codeword 'c'. The lookup table simplifies the calculation of the formulas in the likelihood estimate through mapping. Hardware is not easy to calculate The logarithm of the error probability P in FLASH memory is allocated 6 bits of storage space. P has a finite of 64 values, so a pre-calculated lookup table is used to map each value of P to the corresponding value. To achieve fast and convenient calculations, the mapping relationship of the lookup table is deliberately changed to a random order when using the lookup table, which increases the difficulty for attackers to crack it; The RM code soft decision submodule includes an ALU, an address generator, a stack, and a state machine. The ALU provides the F function, G function, and accumulation calculation functions in the soft decision algorithm, and can also output directly without any operation. The address generator generates SRAM read / write addresses based on the stack and state machine control. The stack and state machine are used to complete the control logic for recursive operations. The hardware recursive circuit is implemented using software design methods. By adding a stack to cache the parameters and local variables in each round of recursion, the state machine in this implementation significantly reduces hardware resource consumption. It only needs to control the current recursive round, while the recursive control of other rounds is implemented by the parameters in the stack. The specific logic is as follows: Figure 4 As shown.

[0031] L(1) push, L(2) push, and pop are jump states that control the push and pop operations in the recursive algorithm. The push state indicates that the next round of recursion is about to begin, and the stack saves the parameters r, m, and the current recursion state. The state machine is reset to the idle state. The pop state indicates that the stack needs to read the r, m, and the current recursion state parameters from the previous round of recursion and continue to execute the current round of recursion.

[0032] Each computational state corresponds to a computational step for a specific variable in the algorithm. All computations are performed within the ALU. The ALU selects the appropriate F function, G function, or SoftDecisionDecode_Rep function for accumulation based on the computational type signal of the current state. The data path for the RM code soft decision part is as follows: starting from L in the ALU's auxiliary data input section, the state machine recursively processes the data. When the current state is computational, data is read from the SRAM and input into the ALU for corresponding computation, and the result is output to the SRAM cache. If the current state is a transition state, the data in the SRAM is directly stored in another address through the ALU. The computational state and transition state are repeatedly recursively processed, ultimately leading to the final result... Complete RM code soft judgment error correction.

[0033] Preferably, the data path of the auxiliary data submodule described in this embodiment includes: FLASH output w, x P Z, P, SRAM output y′; XOR gate inputs w and x P Z obtains w′, and the XOR gate inputs y′, w′, and x. P Z yields c′; the lookup table input P yields... The likelihood estimate L, formed by c′, is used as the input to the ALU to execute a recursive algorithm; the comparator input is the result of the recursive algorithm. Once the correct codeword c is obtained, it is input into an XOR gate along with w′. The output is then XORed with the signature verification result X from outside the module to obtain the ECDSA-PUF value y.

[0034] Preferably, the algorithms in this embodiment include: a fast ECDSA-PUF algorithm, a custom ECDSA signature verification algorithm, and an ECDH encryption algorithm. Since the three algorithms are designed for reuse, a single ECDH-ECDSA-PUF module can execute all three algorithms: the fast ECDSA-PUF algorithm and the custom ECDSA signature verification algorithm reuse the same ECDSA signature verification module; the ECDSA signature verification module and the ECDH encryption module reuse the same ECC point calculation module. Compared to the implementation scheme where each algorithm has its own module, this saves one set of ECDSA signature verification modules and one set of ECC point calculation modules, which are the main components of the hardware area in the algorithms, thus greatly saving hardware space. The overall structure of the ECDH-ECDSA-PUF module using the reuse approach is as follows: Figure 4 As shown.

[0035] The redesigned ECDH-ECDSA-PUF module can compute three algorithms and uses a large number of multiplexers and control signals to select the data path when different algorithms are selected.

[0036] When calculating the custom ECDSA signature verification algorithm, the RM code in the PUF module is in a non-operating state, outputting a busy low-level signal to control the multiplexer. The ECDSA signature verification module verifies the signature, public key, and base point transmitted by the APB, and the message transmitted by the AHB. Since the public key and base point are not multiplied simultaneously, the same APB bus can be reused. Afterwards, the APB bus initiates a custom ECDSA signature verification request, transmitting the corresponding control signals to grant the ECDSA signature verification module control of the dot operation module, completing the ECDSA signature verification, and outputting the verification result to the APB bus.

[0037] When calculating the ECDH encryption algorithm, the APB bus initiates an ECDH encryption request and transmits the corresponding control signals to give the ECDH encryption module control over the point operation module. It receives the public key and base point from the APB, the plaintext and random number from the AES module, completes the ECDH encryption, and outputs the ciphertext to the APB bus.

[0038] When calculating the fast ECDSA-PUF algorithm, the RM code in the PUF module is active, outputting a busy high-level signal to control the multiplexer. The ECDSA signature verification module verifies all auxiliary data w, error probability P, and signature x stored in the FLASH memory. P x P Z, s -1 For signature verification, the ECDSA verification module calculates X using the fixed public key embedded in the developer's hardware. Simultaneously, the PUF module receives auxiliary data w, error probability P, and SRAM value y′ to perform soft-decision analysis on the RM code, calculating c⊕w′. This result is then XORed with X to generate an SRAM PUF, obfuscating the RM code and signature. This process eliminates single-bit critical signals or obvious boundaries between modules during placement and routing. Furthermore, the RM code has low codeword entropy, requiring multiple iterations to obtain a high-entropy PUF. Therefore, for a 256-bit RM code with m = 8 and entropy of 37, the PUF module receives four segments of auxiliary data w, error probability P, and SRAM value y′, calculates four c⊕w′ values, and XORs them with X to obtain a 1024-bit PUF value y with an entropy greater than 100. These multiple calculations also employ the concept of reuse, using a counter to select the auxiliary data w, error probability P, and signature x for each input. P x P Z, s -1 The SRAM value y′ needs to be cached in an additional PUF register; otherwise, during the first RM code soft decision, the initial value in the SRAM will be overwritten by the intermediate calculation results, losing the remaining initial value. The auxiliary data w, error probability P, and signature x... P x P Z, s -1Each 8-bit segment is divided into 4 segments distributed across the 32-bit width of the FLASH memory, as follows: Figure 5 , where w[i], x P [i]、x P Z[i]、s -1 [i] is 1 bit, P[i] is 6 bits, x P Z is expanded to a total of 256 bits using a random bit widening algorithm 26.

[0039] In summary, the ECDH-ECDSA-PUF module uses a reuse design approach to schedule three algorithm sub-modules: the PUF module, the ECDH encryption module, and the ECDSA signature verification module, thereby implementing the operation of the three algorithms.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multiplexing architecture for a soft-decision SRAMPUF based on ECDSA cryptographic protection, characterized in that: It includes an input APB module, an AHB module, an SRAM module, an ECDH encryption module, an ECDSA signature verification module, a PUF module, and an output APB module; The input APB module can input the public key and base point into the ECDH encryption module, the input APB module can input the signature, public key and base point into the ECDSA verification module, and the input APB module can send corresponding control signals to the ECDSA verification module to give the ECDSA verification module control over the point operation module. The AHB module can input messages to the ECDSA signature verification module, and the AHB module can send signals to the FLASH to read data and output it. The SRAM module can output signals to the PUF module and interact with the PUF register. The ECDH encryption module receives the public key and base point from the APB module, the plaintext and random number from the AES module, completes ECDH encryption, and outputs the ciphertext to the output APB module. The ECDSA verification module verifies the signature, public key, and base point transmitted by the APB module and the message transmitted by the AHB module. It initiates a custom ECDSA verification request by inputting the APB module, transmits the corresponding control signals to enable the ECDSA verification module to have control of the dot operation module, completes the ECDSA verification, and outputs the verification result to the output APB module. The PUF module can receive read data from FLASH, data from the counter, output data from the SRAM module, and data from the PUF register; the PUF module can output a busy high-level signal to the ECDSA signature verification unit; the PUF module outputs data to an XOR gate; the ECDSA signature verification module outputs data to an XOR gate; and the XOR gate outputs a PUF value to the ECDSA signature verification unit. The output APB module receives the ciphertext output by the CDH encryption module and the signature verification result output by the ECDSA signature verification module.

2. The multiplexing architecture of a soft-decision SRAMPUF based on ECDSA cryptographic protection according to claim 1, characterized in that: The ECDH encryption module and ECDSA signature verification module share the same underlying point operation module. Control is allocated by using a multiplexer, and each module completes its own top-level encryption algorithm by calling the underlying point operation module.

3. The multiplexing architecture of a soft-decision SRAMPUF based on ECDSA cryptographic protection according to claim 1, characterized in that: The ECDH encryption module and ECDSA signature verification module use an "instruction ROM + RAM + ALU" architecture. The instruction ROM stores the control signals for the ECDH encryption algorithm and the custom ECDSA signature verification algorithm. Each algorithm has two corresponding instruction ROMs, which are selected by switching signals. The RAM caches all variables and changes the read and write addresses according to the control signals in the instruction ROM. The write data is selected by the multiplexer. The ALU can perform all calculations in the elliptic curve arithmetic layer and the modular arithmetic layer. The calculation type is controlled by the instruction ROM. The ALU result can be output outside the module or written back to RAM.

4. The multiplexing architecture of a soft-decision SRAMPUF based on ECDSA cryptographic protection according to claim 3, characterized in that: The ECDH encryption module and ECDSA signature verification module have unique parts responsible for data input and output and completing their unique calculations. The data of the unique parts will eventually be written into the shared RAM. The switching signal controls the RAM write multiplexer to control the current algorithm. When calculating the ECDH encryption algorithm, the RAM data is written from the ECDH encryption-specific part. The ECDH instruction ROM address starts to increment automatically. A counter is used to concatenate plaintext to expand the redundant values ​​in the plaintext embedding algorithm. All steps in the remaining ECDH encryption algorithm and plaintext embedding algorithm can be decomposed into elliptic curve point operations and modular operations. Therefore, all data is written to RAM through a multiplexer, and all calculations are completed by the point operation module. The instruction output from the instruction ROM controls the address written to RAM and the calculation type of the point operation module. When calculating a custom ECDSA signature verification algorithm, the RAM write data comes from the ECDSA signature verification-specific part. The ECDSA instruction ROM address starts to increment automatically. First, the unique HASH module is used to calculate z. The input is selected through the multiplexer. If the second layer of the triple hybrid encryption system is being calculated, the PUF value is concatenated with the software program read from FLASH as a message. Depending on the current user registration stage or user usage stage, the developer's fixed public key or the user's authorized public key is selected.

5. The multiplexing architecture of a soft-decision SRAMPUF based on ECDSA cryptographic protection according to claim 1, characterized in that: The PUF module includes an auxiliary data submodule and an RM code soft decision submodule; The auxiliary data section includes XOR gates at the input and output of the arithmetic logic unit, a comparator at the output of the arithmetic logic unit, and a lookup table used to simplify the calculation of the likelihood estimate. The three XOR gates correspond to the three XOR operations in the algorithm's recovery phase. The comparator converts the calculated likelihood estimate into codeword c. The lookup table simplifies the calculation of the formula in the likelihood estimate through mapping. Hardware is not easy to calculate The logarithm of the error probability P in FLASH memory is allocated 6 bits of storage space. P has a finite of 64 values, so a pre-calculated lookup table is used to map each value of P to the corresponding value. To achieve fast and convenient calculations, the mapping relationship of the lookup table is deliberately changed to random order when using the lookup table; The RM code soft decision submodule includes an ALU, an address generator, a stack, and a state machine. The ALU provides the F function, G function, and accumulation calculation functions in the soft decision algorithm, and can also output directly without performing any operation. The address generator can generate SRAM read and write addresses according to the stack and state machine control. The stack and state machine are used to complete the control logic for recursive operations.

6. The multiplexing architecture of a soft-decision SRAMPUF based on ECDSA cryptographic protection according to claim 5, characterized in that: The data path of the auxiliary data submodule includes: FLASH output w, x P Z, P, SRAM output y′; XOR gate inputs w and x P Z obtains w′, and the XOR gate inputs y′, w′, and x. P Z yields c′; the lookup table input P yields... The likelihood estimate L, formed by c′, is used as the input to the ALU to execute a recursive algorithm; the comparator input is the result of the recursive algorithm. Once the correct codeword c is obtained, it is input into an XOR gate along with w′. The output is then XORed with the signature verification result X from outside the module to obtain the ECDSA-PUF value y.