A hybrid encrypted data transmission and storage method and system adapted to a creative environment

Abstract

This invention relates to the field of information security technology and proposes a hybrid encrypted data transmission and storage method and system adapted to the domestically developed IT environment. The method includes: training an adaptive neural network based on the hardware information of the domestically developed CPU to output a capability feature vector; using a cache prediction network based on the capability feature vector to predict the performance of national cryptographic algorithms and output a performance prediction matrix; using a deep reinforcement learning network based on the performance prediction matrix to detect resource conflicts between combinations of national cryptographic algorithms, filtering the combinations, and outputting the optimal combination of national cryptographic algorithms; encrypting the data to be transmitted and outputting encrypted transmission data; transmitting the encrypted transmission data to the target node, decrypting the encrypted transmission data, re-encrypting and storing it, and outputting securely stored data. This invention improves the security and adaptability of data transmission and storage in the domestically developed IT environment, and meets the security compliance requirements of the domestically developed IT environment through a complete transmission verification and storage re-encryption process.

Description

Technical Field

[0001] This invention relates to the field of information security technology, and in particular to a hybrid encrypted data transmission and storage method and system adapted to the domestic IT innovation environment. Background Technology

[0002] The "domestic information technology innovation environment" refers to an innovative ecosystem for information technology applications aimed at localization and independent control, encompassing basic software and hardware products such as CPU chips, operating systems, databases, and middleware. Within this environment, secure data transmission and storage face multiple technical challenges, including heterogeneous hardware platforms, standardized application of national cryptographic algorithms, and performance optimization. Domestically developed CPU architectures exhibit diverse characteristics, each possessing unique instruction set architectures and cryptographic acceleration capabilities. This heterogeneous environment introduces complexity to the unified deployment and performance optimization of national cryptographic algorithms.

[0003] In existing technologies, data encryption transmission and storage in the domestic IT innovation environment mainly adopts a single national cryptographic algorithm or a fixed combination of algorithms to achieve basic data security protection through preset encryption strategies. However, existing methods do not adequately consider the inherent adaptation mechanism between the heterogeneous architecture of domestic IT innovation CPUs and the performance characteristics of national cryptographic algorithms. They are unable to organically integrate objective hardware instruction set capability constraints with actual algorithm execution performance requirements, resulting in the inability to achieve high-performance, highly adaptable data security transmission and storage in the domestic IT innovation environment. Furthermore, they lack a national cryptographic algorithm performance prediction mechanism for heterogeneous domestic IT innovation CPUs and cannot dynamically select the optimal algorithm combination based on the characteristics of different CPU architectures. Summary of the Invention

[0004] In view of this, the present invention proposes a hybrid encrypted data transmission and storage method and system adapted to the domestic IT innovation environment. It solves the problems of existing methods not taking into account the inherent adaptation mechanism between the heterogeneous architecture of domestic IT innovation CPUs and the performance characteristics of national cryptographic algorithms, making it difficult to organically integrate the objective hardware instruction set capability constraints with the actual algorithm execution performance requirements. This results in the inability to achieve high-efficiency and highly adaptable secure data transmission and storage in the domestic IT innovation environment, and the lack of a national cryptographic algorithm performance prediction mechanism for heterogeneous domestic IT innovation CPUs, making it impossible to dynamically select the optimal algorithm combination according to the characteristics of different CPU architectures.

[0005] The technical solution of this invention is implemented as follows: On one hand, this invention provides a hybrid encrypted data transmission and storage method adapted to the domestic IT innovation environment, comprising the following steps: Obtain the hardware information of the domestically developed CPU, train the adaptive neural network based on the hardware information of the domestically developed CPU, learn the differences in the implementation of the national cryptographic algorithm instruction set by different domestically developed CPU architectures, and output the heterogeneous instruction set capability feature vector. Based on the feature vector of the heterogeneous instruction set capability, a cache prediction network is used to predict the performance of the national cryptographic algorithm and output a performance prediction matrix. Based on the performance prediction matrix, a deep reinforcement learning network is used to detect resource conflicts among combinations of Chinese cryptographic algorithms. The combinations of Chinese cryptographic algorithms are then filtered according to the resource conflicts, and the optimal combination of Chinese cryptographic algorithms is output. The optimal combination of national cryptographic algorithms is used to encrypt the data to be transmitted, and the encrypted transmitted data is output. The encrypted transmission data is transmitted to the target node, and the encrypted transmission data is decrypted and re-encrypted and stored using the optimal combination of national cryptographic algorithms, outputting securely stored data.

[0006] Based on the above technical solutions, preferably, the step of acquiring the hardware information of the domestically developed CPU, training the adaptive neural network based on the hardware information of the domestically developed CPU, learning the differences in the implementation of the national cryptographic algorithm instruction set by different domestically developed CPU architectures, and outputting a heterogeneous instruction set capability feature vector includes: The hardware identification information and CPUID instruction return value of the domestically developed CPU are obtained as the hardware information of the domestically developed CPU. The adaptive neural network is trained based on the hardware identification information and CPUID instruction return value to learn the implementation differences of the national cryptographic algorithm instruction set of different domestically developed CPU architectures. Micro-benchmark tests are performed to dynamically calibrate the actual cryptographic instruction set performance parameters of each domestically developed CPU. Based on the actual cryptographic instruction set performance parameters, a heterogeneous instruction set capability feature vector is generated. The adaptive neural network is established by fusing heterogeneous feature differences.

[0007] Based on the above technical solutions, preferably, the training process of the adaptive neural network includes: The hardware identification information and CPUID instruction return value are used as network input features. The heterogeneous feature differences of the domestically developed CPU architecture are extracted through a multilayer perceptron. The network weight parameters are updated using the backpropagation algorithm. The micro-benchmark test measures the instruction set execution time of different domestically developed CPUs by executing standard national cryptographic algorithm test vectors. The instruction set execution time is fused and mapped with the heterogeneous feature differences to generate the heterogeneous instruction set capability feature vector. The heterogeneous instruction set capability feature vector includes CPU architecture identification and performance parameters.

[0008] Based on the above technical solutions, preferably, the step of using a cache prediction network to predict the performance of the national cryptographic algorithm based on the heterogeneous instruction set capability feature vector, and outputting a performance prediction matrix, includes: Using the heterogeneous instruction set capability feature vector as input, and employing a cache prediction network sensitive to elliptic curve domain operations, performance prediction models for SM2 point operations, SM4 lookup operations, and SM9 bilinear pairing operations are established. The performance prediction models are used to predict the execution performance of national cryptographic algorithms on different domestically developed CPUs, generating a performance prediction matrix. The performance prediction matrix includes algorithm execution time and cache hit rate.

[0009] Based on the above technical solutions, preferably, the prediction process of the cache prediction network sensitive to elliptic curve domain operations includes: The heterogeneous instruction set capability feature vector is input into a multilayer neural network to extract memory access pattern features of elliptic curve domain operations. The cache behavior of different national cryptographic algorithms is predicted through the multilayer neural network. Based on the cache behavior, the doubling operation time of SM2 point operation, the table lookup delay of SM4 lookup operation, and the pairing calculation time of SM9 bilinear pairing operation are calculated. The performance prediction matrix is ​​established based on the doubling operation time, table lookup delay, and pairing calculation time.

[0010] Based on the above technical solutions, preferably, the step of using a deep reinforcement learning network to detect resource conflicts among combinations of national cryptographic algorithms based on the performance prediction matrix, filtering the combinations of national cryptographic algorithms according to the resource conflicts, and outputting the optimal combination of national cryptographic algorithms includes: Using the performance prediction matrix as input, a deep reinforcement learning network based on transmission and storage conflict awareness is used to detect transmission resource conflicts in the data transmission stage and storage resource conflicts in the data storage stage of the national cryptographic algorithm combination. Conflict evaluation index is constructed based on the transmission resource conflicts and storage resource conflicts. Candidate national cryptographic algorithm combinations are screened based on the conflict evaluation index, and the optimal national cryptographic algorithm combination is output.

[0011] Based on the above technical solutions, preferably, the selection process for the deep reinforcement learning network based on transmission-storage conflict awareness includes: The performance prediction matrix is ​​input into the deep reinforcement learning network to generate an environment state vector. Candidate national cryptographic algorithm combinations are selected based on the environment state vector. The transmission and storage conflict degree of the candidate national cryptographic algorithm combinations is evaluated. The conflict evaluation index is updated based on the transmission and storage conflict degree. The execution state generation process and the algorithm combination selection process are iterated until convergence. The candidate national cryptographic algorithm combination with the minimum conflict degree is selected to obtain the optimal national cryptographic algorithm combination.

[0012] Based on the above technical solutions, preferably, the step of encrypting the data to be transmitted using the optimal national cryptographic algorithm combination and outputting encrypted transmission data includes: The data to be transmitted is adaptively segmented based on the encryption characteristics of each national cryptographic algorithm in the optimal national cryptographic algorithm combination. Based on the adaptive segmentation result, the corresponding encryption parameter combination is configured, and the segmented data and encryption parameter combination are output. The optimal national cryptographic algorithm combination is used to perform multi-level concatenated encryption on the block data according to the encryption parameter combination. The concatenated encryption results of each block data are then reassembled to output the encrypted transmission data.

[0013] Based on the above technical solutions, preferably, the step of transmitting the encrypted data to the target node, decrypting the encrypted data using the optimal national cryptographic algorithm combination, re-encrypting and storing the data, and outputting securely stored data includes: The encrypted transmission data is transmitted to the target node through a secure transmission channel. During the transmission process, the optimal national cryptographic algorithm combination is used to generate a data integrity check code. The integrity of the encrypted transmission data after transmission is completed is verified based on the data integrity check code, and the received data that has passed the verification is output. At the target node, the received data is decrypted in reverse order using the optimal national cryptographic algorithm combination. Based on the security level requirements of the target node's storage environment, a new storage encryption mode is selected, and the decrypted data is re-encrypted using the new storage encryption mode to output the secure stored data.

[0014] On the other hand, the present invention also provides a hybrid encrypted data transmission and storage system adapted to the domestic IT innovation environment, the system comprising: The heterogeneous instruction set adaptation module is used to acquire the hardware information of the domestically developed CPU, train the adaptive neural network based on the hardware information of the domestically developed CPU, learn the differences in the implementation of the national cryptographic algorithm instruction set by different domestically developed CPU architectures, and output the heterogeneous instruction set capability feature vector. The cache performance prediction module is used to predict the performance of the national cryptographic algorithm based on the heterogeneous instruction set capability feature vector and a cache prediction network, and output a performance prediction matrix. The algorithm combination optimization module is used to detect resource conflicts between combinations of Chinese cryptographic algorithms based on the performance prediction matrix using a deep reinforcement learning network, filter the combinations of Chinese cryptographic algorithms according to the resource conflicts, and output the optimal combination of Chinese cryptographic algorithms. The data encryption processing module is used to encrypt the data to be transmitted using the optimal national cryptographic algorithm combination and output encrypted transmission data. The secure transmission and storage module is used to transmit the encrypted transmission data to the target node, decrypt the encrypted transmission data using the optimal national cryptographic algorithm combination, re-encrypt and store the data, and output securely stored data.

[0015] The hybrid encrypted data transmission and storage method and system adapted to the domestic IT innovation environment of the present invention have the following advantages over the prior art: (1) By integrating adaptive neural networks, cache prediction networks and deep reinforcement learning networks, the adaptive neural network is used to learn the differences in the implementation of national cryptographic algorithm instruction sets by heterogeneous domestically developed CPU architectures. Combined with cache prediction networks sensitive to elliptic curve domain operations, the algorithm performance is dynamically predicted. The algorithm combination is screened according to the transmission and storage conflict perception mechanism, which improves the security and adaptability of data transmission and storage in the domestically developed environment. At the same time, the security and compliance requirements of the domestically developed environment are met through the complete transmission verification and storage re-encryption process. (2) By integrating hardware information acquisition and adaptive neural network training, micro-benchmark testing is used to dynamically calibrate the performance of actual cryptographic instruction sets and extract heterogeneous feature differences. A network model is dynamically established by combining multilayer perceptron and backpropagation algorithm. The heterogeneous feature differences are fused and mapped according to the instruction set execution time of the standard national cryptographic algorithm test vector, which improves the accuracy and adaptability of learning the differences in the national cryptographic algorithm instruction set under the heterogeneous information technology innovation environment. (3) By integrating the heterogeneous instruction set capability features with the cache prediction network sensitive to elliptic curve domain operations, multi-layer neural networks are used to extract memory access pattern features and predict cache behavior. The specific computational characteristics of SM2 point operation, SM4 lookup operation and SM9 bilinear pairing operation are combined to dynamically establish a performance prediction model. The performance prediction matrix is ​​accurately constructed based on the doubling operation time, table lookup delay and pairing calculation time, which improves the pertinence of national cryptographic algorithm performance prediction in heterogeneous information technology innovation environment. (4) By integrating the performance prediction matrix with the deep reinforcement learning network that senses transmission and storage conflicts, the candidate national cryptographic algorithm combination selection and conflict degree evaluation are carried out using the environmental state vector. The conflict evaluation index is dynamically constructed by combining the dual detection mechanism of transmission resource conflict and storage resource conflict. The state generation and algorithm combination selection are continuously improved according to the iterative optimization process, thereby improving the adaptability of national cryptographic algorithm combination optimization in the heterogeneous information technology innovation environment. Attached Figure Description

[0016] 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 only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a hybrid encrypted data transmission and storage method adapted to the domestic IT innovation environment according to the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 This invention provides a hybrid encrypted data transmission and storage method adapted to the domestic IT innovation environment, comprising the following steps: Obtain the hardware information of the domestically developed CPU, train the adaptive neural network based on the hardware information of the domestically developed CPU, learn the differences in the implementation of the national cryptographic algorithm instruction set by different domestically developed CPU architectures, and output the heterogeneous instruction set capability feature vector. Based on the feature vector of the heterogeneous instruction set capability, a cache prediction network is used to predict the performance of the national cryptographic algorithm and output a performance prediction matrix. Based on the performance prediction matrix, a deep reinforcement learning network is used to detect resource conflicts among combinations of Chinese cryptographic algorithms. The combinations of Chinese cryptographic algorithms are then filtered according to the resource conflicts, and the optimal combination of Chinese cryptographic algorithms is output. The optimal combination of national cryptographic algorithms is used to encrypt the data to be transmitted, and the encrypted transmitted data is output. The encrypted transmission data is transmitted to the target node, and the encrypted transmission data is decrypted and re-encrypted and stored using the optimal combination of national cryptographic algorithms, outputting securely stored data.

[0020] Specifically, this embodiment integrates adaptive neural networks, cache prediction networks, and deep reinforcement learning networks. The adaptive neural network learns the differences in the implementation of national cryptographic algorithm instruction sets by heterogeneous domestically developed CPU architectures. Combined with a cache prediction network sensitive to elliptic curve domain operations, it dynamically predicts algorithm performance and filters algorithm combinations based on a transmission-storage conflict awareness mechanism. This embodiment addresses the problems of inaccurate performance prediction of national cryptographic algorithms and suboptimal algorithm combination selection in heterogeneous domestically developed environments through multi-layered concatenated encryption processing and reverse decryption-re-encryption mechanisms. This improves the security and adaptability of data transmission and storage in domestically developed environments, while meeting the security and compliance requirements of these environments through a complete transmission verification and storage re-encryption process.

[0021] The process involves acquiring the hardware information of the domestically developed CPU, training an adaptive neural network based on this information, learning the differences in the implementation of national cryptographic algorithm instruction sets by different domestically developed CPU architectures, and outputting a heterogeneous instruction set capability feature vector, including: The hardware identification information and CPUID instruction return value of the domestically developed CPU are obtained as the hardware information of the domestically developed CPU. The adaptive neural network is trained based on the hardware identification information and CPUID instruction return value to learn the implementation differences of the national cryptographic algorithm instruction set of different domestically developed CPU architectures. Micro-benchmark tests are performed to dynamically calibrate the actual cryptographic instruction set performance parameters of each domestically developed CPU. Based on the actual cryptographic instruction set performance parameters, a heterogeneous instruction set capability feature vector is generated. The adaptive neural network is established by fusing heterogeneous feature differences.

[0022] The training process of the adaptive neural network includes: The hardware identification information and CPUID instruction return value are used as network input features. The heterogeneous feature differences of the domestically developed CPU architecture are extracted through a multilayer perceptron. The network weight parameters are updated using the backpropagation algorithm. The micro-benchmark test measures the instruction set execution time of different domestically developed CPUs by executing standard national cryptographic algorithm test vectors. The instruction set execution time is fused and mapped with the heterogeneous feature differences to generate the heterogeneous instruction set capability feature vector. The heterogeneous instruction set capability feature vector includes CPU architecture identification and performance parameters.

[0023] In one specific embodiment, the formula for calculating the heterogeneous instruction set capability feature vector is: ; ; in, This is a feature vector representing the heterogeneous instruction set capability of domestic IT innovation. The number of CPU architecture types for domestic IT innovation; For the first Adaptive weighting coefficients for each architecture; For dynamic micro-benchmark testing weights; This is a coefficient specific to the information technology innovation environment. This is a function for extracting differentiated features of the domestically developed CPU architecture. This is a hardware feature extraction function; Hardware identification information; The return value of the CPUID instruction; This is a micro-benchmark performance calibration function; For instruction set execution time; Differences in heterogeneous characteristics; for; For the first Sensitivity parameters of the architecture; For the first Sensitivity parameters of the architecture; Rating the CPU's overall performance; As a moderating factor for architectural differences; For the first Instruction set differences between architectures.

[0024] Specifically, unlike traditional heterogeneous instruction set feature vector generation methods that use fixed weight coefficients and cannot adapt to the performance differences of different domestically developed CPU architectures, this embodiment introduces an adaptive weight mechanism. The softmax function is used to dynamically adjust the weights of each architecture based on the overall CPU performance score, while also introducing instruction set differences. A correction factor is added to enable the feature vector to adaptively reflect the differences in the implementation of the national cryptographic algorithm instruction set by different domestically developed CPU architectures (such as ARM, RISC-V, x86, etc.). In addition, a new item specific to the domestically developed environment is added. It is specifically designed to handle the unique characteristics of domestically developed CPU architectures, solving the problem of insufficient adaptability of traditional methods to heterogeneous environments.

[0025] This embodiment integrates hardware information acquisition with adaptive neural network training, utilizes micro-benchmark testing for dynamic calibration of actual cryptographic instruction set performance and extraction of heterogeneous feature differences, dynamically establishes a network model using a multilayer perceptron and backpropagation algorithm, and performs fusion mapping of heterogeneous feature differences based on the instruction set execution time of the standard national cryptographic algorithm test vector. This embodiment addresses the issues of inaccurate identification of instruction set capabilities and imprecise acquisition of performance parameters for different domestically developed CPU architectures through a CPU architecture identification and performance parameter integration mechanism, thereby improving the adaptability of learning the differences in national cryptographic algorithm instruction sets under heterogeneous domestically developed environments.

[0026] In one specific embodiment, the weight update formula of the adaptive neural network is: ; in, For the first The network weight parameters for the next iteration; For the first The network weight parameters for the next iteration; The learning rate; The loss function; The regularization coefficient for heterogeneous features; For heterogeneous feature differentiation regularization function; The difference lies in the heterogeneous characteristics.

[0027] The process of predicting the performance of the national cryptographic algorithm using a cache prediction network based on the heterogeneous instruction set capability feature vector, and outputting a performance prediction matrix, includes: Using the heterogeneous instruction set capability feature vector as input, and employing a cache prediction network sensitive to elliptic curve domain operations, performance prediction models for SM2 point operations, SM4 lookup operations, and SM9 bilinear pairing operations are established. The performance prediction models are used to predict the execution performance of national cryptographic algorithms on different domestically developed CPUs, generating a performance prediction matrix. The performance prediction matrix includes algorithm execution time and cache hit rate.

[0028] The prediction process of the cache prediction network, which is sensitive to elliptic curve domain operations, includes: The heterogeneous instruction set capability feature vector is input into a multilayer neural network to extract memory access pattern features of elliptic curve domain operations. The cache behavior of different national cryptographic algorithms is predicted through the multilayer neural network. Based on the cache behavior, the doubling operation time of SM2 point operation, the table lookup delay of SM4 lookup operation, and the pairing calculation time of SM9 bilinear pairing operation are calculated. The performance prediction matrix is ​​established based on the doubling operation time, table lookup delay, and pairing calculation time.

[0029] In one specific embodiment, the multidimensional cache-aware prediction formula for the elliptic curve domain operation is: ; ; in, Cached prediction results for elliptic curve domain operations; For the algorithm Elliptic curve domain sensitive weights; This refers to elliptic curve public-key cryptography algorithms. It is a block cipher algorithm; An identity-based cryptographic algorithm; For activation functions; For the algorithm Memory access pattern feature vector; For the algorithm Temporal characteristics; The characteristic matrix for operations on the elliptic curve domain; This is the spatial locality weight vector; Basic weighting coefficients; For the algorithm The algorithm specificity coefficient; The computational complexity function for elliptic curve domain operations; The number of elliptic curve operations that were cache hits; This represents the total number of elliptic curve operations. It is an exponential function.

[0030] Specifically, unlike traditional cache prediction methods that ignore the special memory access patterns of elliptic curve domain operations, leading to inaccurate predictions of cache behavior for national cryptographic algorithms such as SM2, SM4, and SM9, this embodiment introduces elliptic curve domain sensitive weights. Based on the computational complexity function of elliptic curve domain Ratio of cache hit rate The weights of each algorithm are dynamically adjusted to enable the prediction model to perceive the different caching characteristics of SM2's elliptic curve point operations, SM4's table lookup operations, and SM9's bilinear pairing operations. Simultaneously, the elliptic curve domain operation feature matrix is ​​integrated. and spatial locality weight vector This enhances the ability to perceive the memory access patterns of elliptic curve domain operations.

[0031] This embodiment integrates heterogeneous instruction set capabilities with an elliptic curve domain (ECD) sensitive cache prediction network. It utilizes a multi-layer neural network for memory access pattern feature extraction and cache behavior prediction, dynamically establishing a performance prediction model based on the specific computational characteristics of SM2 point operations, SM4 lookup operations, and SM9 bilinear pairing operations. The performance prediction matrix is ​​then precisely constructed based on doubling operation time, lookup table latency, and pairing computation time. This embodiment addresses the shortcomings of traditional cache prediction methods in adapting to the special operational modes of Chinese cryptographic algorithms and achieving low performance prediction accuracy through an ECD mechanism, thereby improving the relevance of Chinese cryptographic algorithm performance prediction in heterogeneous information technology environments.

[0032] Based on the performance prediction matrix, a deep reinforcement learning network is used to detect resource conflicts among combinations of national cryptographic algorithms. The combinations of national cryptographic algorithms are then filtered according to the resource conflicts, and the optimal combination of national cryptographic algorithms is output, including: Using the performance prediction matrix as input, a deep reinforcement learning network based on transmission and storage conflict awareness is used to detect transmission resource conflicts in the data transmission stage and storage resource conflicts in the data storage stage of the national cryptographic algorithm combination. Conflict evaluation index is constructed based on the transmission resource conflicts and storage resource conflicts. Candidate national cryptographic algorithm combinations are screened based on the conflict evaluation index, and the optimal national cryptographic algorithm combination is output.

[0033] The selection process for the deep reinforcement learning network based on transmission-to-storage conflict awareness includes: The performance prediction matrix is ​​input into the deep reinforcement learning network to generate an environment state vector. Candidate national cryptographic algorithm combinations are selected based on the environment state vector. The transmission and storage conflict degree of the candidate national cryptographic algorithm combinations is evaluated. The conflict evaluation index is updated based on the transmission and storage conflict degree. The execution state generation process and the algorithm combination selection process are iterated until convergence. The candidate national cryptographic algorithm combination with the minimum conflict degree is selected to obtain the optimal national cryptographic algorithm combination.

[0034] In one specific embodiment, the conflict assessment index is calculated as follows: ; ; in, For timing-aware evaluation of transmission and storage conflicts; This represents the total number of candidate national cryptographic algorithm combinations; For the first Time-aware priority weights for each combination; and The first The combined transmission and storage resource utilization rate; For the first Total resource capacity of each combination; For the first Resource overlap of each combination; For the first Resource interference level of each combination; and These are the resource overlap conflict trade-off coefficient and the resource interference conflict trade-off coefficient, respectively. Basic priority weight; This is the adjustment coefficient for the time-series impact; This is the time-series decay factor; and The first The transmission and storage times of each combination; This is the conflict prediction adjustment function; For the first The prediction time window for each combination.

[0035] Specifically, unlike traditional conflict assessment methods that only consider static resource occupancy and ignore the timing dependence of transmission and storage operations and dynamic conflict changes, this embodiment introduces time-aware priority weights. Using the exponential decay function The impact of time differences in quantified transmission and storage operations is considered; the smaller the time difference, the greater the impact of the conflict. Additionally, new resource interference is introduced. This paper comprehensively considers the mutual interference between algorithm combinations and introduces a conflict prediction adjustment function. By dynamically adjusting the evaluation weights based on the prediction time window, the system achieves time-series awareness and predictive evaluation of transmission and storage conflicts.

[0036] This embodiment utilizes a deep reinforcement learning network that integrates performance prediction matrices with transmission-storage conflict awareness. It employs environmental state vectors for candidate national cryptographic algorithm combinations and conflict severity assessment. A conflict assessment index is dynamically constructed using a dual detection mechanism for transmission and storage resource conflicts. The state generation and algorithm combination selection are continuously improved through iterative optimization. This embodiment addresses the shortcomings of traditional algorithm combination selection methods in considering resource conflicts in transmission and storage scenarios and their low efficiency in algorithm collaboration through a conflict severity minimization screening mechanism, thereby improving the adaptability of national cryptographic algorithm combination optimization in heterogeneous information technology innovation environments.

[0037] The process of encrypting the data to be transmitted using the optimal national cryptographic algorithm combination and outputting encrypted transmission data includes: The data to be transmitted is adaptively segmented based on the encryption characteristics of each national cryptographic algorithm in the optimal national cryptographic algorithm combination. Based on the adaptive segmentation result, the corresponding encryption parameter combination is configured, and the segmented data and encryption parameter combination are output.

[0038] In one specific embodiment, the data type and data size of the data to be transmitted are analyzed. Based on the key length of SM2 elliptic curve encryption, the block size of SM4 block encryption, and the encryption mode of SM9 identifier cipher in the optimal national cryptographic algorithm combination, the data to be transmitted is divided into adaptive data blocks corresponding to the processing capabilities of each national cryptographic algorithm. Key parameters, initialization vectors, and encryption rounds are configured for each of the adaptive data blocks to generate the block data and encryption parameter combination.

[0039] The optimal national cryptographic algorithm combination is used to perform multi-level concatenated encryption on the block data according to the encryption parameter combination. The concatenated encryption results of each block data are then reassembled to output the encrypted transmission data.

[0040] In one specific embodiment, the block data is sequentially input into each national cryptographic algorithm in the optimal national cryptographic algorithm combination according to the encryption parameter combination. The block data is initially encrypted using the first layer SM4 block encryption, the initial encryption result is asymmetrically encrypted using the second layer SM2 elliptic curve encryption, and the asymmetric encryption result is identified and encrypted using the third layer SM9 identifier cipher. The three-layer encryption results of each block are concatenated and recombined according to the original data order to generate the encrypted transmission data.

[0041] Specifically, this embodiment integrates the optimal combination of national cryptographic algorithms with an adaptive block-segmentation mechanism. It utilizes data type and scale analysis to match algorithm characteristics and configure encryption parameters. It dynamically divides data into adaptive blocks using encryption modes of SM2 elliptic curve cryptography, SM4 block cipher, and SM9 identifier cipher, and precisely controls the multi-layered concatenated encryption process based on key parameters, initialization vectors, and the number of encryption rounds. This embodiment addresses the issues of insufficient security strength and poor adaptability of fixed block-segmentation patterns in traditional single encryption algorithms through a three-layer progressive encryption and data reassembly mechanism, thereby improving the security and flexibility of data encryption processing in a domestically developed information technology environment.

[0042] The process of transmitting the encrypted data to the target node, decrypting the encrypted data using the optimal national cryptographic algorithm combination, re-encrypting and storing the data, and outputting securely stored data includes: The encrypted transmission data is transmitted to the target node through a secure transmission channel. During the transmission process, the optimal national cryptographic algorithm combination is used to generate a data integrity check code. Based on the data integrity check code, the integrity of the transmitted encrypted data is verified, and the received data that has passed the verification is output.

[0043] In one specific embodiment, a secure transmission channel based on the optimal combination of national cryptographic algorithms is established. The encrypted transmission data is fragmented and transmitted according to a preset transmission fragment size. A fragment check code generated using the SM3 hash algorithm is embedded in each transmission fragment. After the target node receives the data, the transmission fragments are reassembled and the fragment check code is verified. After successful verification, a transmission integrity confirmation identifier is generated, and the received data is output. The received data includes the transmission integrity confirmation identifier.

[0044] At the target node, the received data is decrypted in reverse order using the optimal national cryptographic algorithm combination. Based on the security level requirements of the target node's storage environment, a new storage encryption mode is selected, and the decrypted data is re-encrypted using the new storage encryption mode to output the secure stored data.

[0045] In one specific embodiment, the received data is decrypted in three layers of reverse order using the SM9 identifier cipher, SM2 elliptic curve encryption, and SM4 block encryption in the optimal national cryptographic algorithm combination in the reverse order of encryption. After obtaining the original data, the corresponding storage encryption algorithm is selected according to the storage security policy of the target node. The original data is then re-encrypted using the storage encryption algorithm combined with the newly generated storage key. A storage timestamp and access permission identifier are added to generate the secure storage data.

[0046] Specifically, this embodiment integrates a secure transmission channel with a data integrity verification mechanism. It utilizes the SM3 hash algorithm for fragment checksum generation and transmission integrity verification, combines reverse decryption processing with dynamic selection of storage encryption mode based on the target node's storage environment security level, and re-encrypts the original data and adds permission identifiers according to the storage security policy. Through a three-layer reverse decryption and storage re-encryption mechanism, it solves the problems of insufficient integrity verification and inadequate storage security adaptability in traditional data transmission and storage processes, thereby improving the security and reliability of data transmission and storage in the domestic IT innovation environment. At the same time, it meets the security control and audit traceability requirements of the target node's storage environment through storage timestamps and access permission identifiers.

[0047] This invention also provides a hybrid encrypted data transmission and storage system adapted to the domestic IT innovation environment, the system comprising: The heterogeneous instruction set adaptation module is used to acquire the hardware information of the domestically developed CPU, train the adaptive neural network based on the hardware information of the domestically developed CPU, learn the differences in the implementation of the national cryptographic algorithm instruction set by different domestically developed CPU architectures, and output the heterogeneous instruction set capability feature vector. The cache performance prediction module is used to predict the performance of the national cryptographic algorithm based on the heterogeneous instruction set capability feature vector and a cache prediction network, and output a performance prediction matrix. The algorithm combination optimization module is used to detect resource conflicts between combinations of Chinese cryptographic algorithms based on the performance prediction matrix using a deep reinforcement learning network, filter the combinations of Chinese cryptographic algorithms according to the resource conflicts, and output the optimal combination of Chinese cryptographic algorithms. The data encryption processing module is used to encrypt the data to be transmitted using the optimal national cryptographic algorithm combination and output encrypted transmission data. The secure transmission and storage module is used to transmit the encrypted transmission data to the target node, decrypt the encrypted transmission data using the optimal national cryptographic algorithm combination, re-encrypt and store the data, and output securely stored data.

[0048] Specifically, this embodiment of a hybrid encrypted data transmission and storage system adapted to the domestic IT innovation environment integrates a heterogeneous instruction set adaptive identification and a cache performance intelligent prediction mechanism. It utilizes an adaptive neural network to learn the implementation differences of national cryptographic algorithm instruction sets for different domestic IT innovation CPU architectures and the performance characteristics of cache prediction network modeling algorithms. Combined with a deep reinforcement learning network, it dynamically detects resource conflicts between combinations of national cryptographic algorithms and optimizes the selection of algorithm combinations, thereby improving the intelligence and security performance of hybrid encrypted data transmission and storage in the domestic IT innovation environment.

[0049] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A hybrid encrypted data transmission and storage method adapted to the domestic IT innovation environment, characterized in that, Includes the following steps: Obtain the hardware information of the domestically developed CPU, train the adaptive neural network based on the hardware information of the domestically developed CPU, learn the differences in the implementation of the national cryptographic algorithm instruction set by different domestically developed CPU architectures, and output the heterogeneous instruction set capability feature vector. Based on the feature vector of the heterogeneous instruction set capability, a cache prediction network is used to predict the performance of the national cryptographic algorithm and output a performance prediction matrix. Based on the performance prediction matrix, a deep reinforcement learning network is used to detect resource conflicts among combinations of Chinese cryptographic algorithms. The combinations of Chinese cryptographic algorithms are then filtered according to the resource conflicts, and the optimal combination of Chinese cryptographic algorithms is output. The optimal combination of national cryptographic algorithms is used to encrypt the data to be transmitted, and the encrypted transmitted data is output. The encrypted transmission data is transmitted to the target node, and the encrypted transmission data is decrypted and re-encrypted and stored using the optimal combination of national cryptographic algorithms, outputting securely stored data.

2. The hybrid encrypted data transmission and storage method adapted to the information technology innovation environment as described in claim 1, characterized in that, The process involves acquiring the hardware information of the domestically developed CPU, training an adaptive neural network based on this information, learning the differences in the implementation of national cryptographic algorithm instruction sets by different domestically developed CPU architectures, and outputting a heterogeneous instruction set capability feature vector, including: The hardware identification information and CPUID instruction return value of the domestically developed CPU are obtained as the hardware information of the domestically developed CPU. The adaptive neural network is trained based on the hardware identification information and CPUID instruction return value to learn the implementation differences of the national cryptographic algorithm instruction set of different domestically developed CPU architectures. Micro-benchmark tests are performed to dynamically calibrate the actual cryptographic instruction set performance parameters of each domestically developed CPU. Based on the actual cryptographic instruction set performance parameters, a heterogeneous instruction set capability feature vector is generated. The adaptive neural network is established by fusing heterogeneous feature differences.

3. The hybrid encrypted data transmission and storage method adapted to the information technology innovation environment as described in claim 2, characterized in that, The training process of the adaptive neural network includes: The hardware identification information and CPUID instruction return value are used as network input features. The heterogeneous feature differences of the domestically developed CPU architecture are extracted through a multilayer perceptron. The network weight parameters are updated using the backpropagation algorithm. The micro-benchmark test measures the instruction set execution time of different domestically developed CPUs by executing standard national cryptographic algorithm test vectors. The instruction set execution time is fused and mapped with the heterogeneous feature differences to generate the heterogeneous instruction set capability feature vector. The heterogeneous instruction set capability feature vector includes CPU architecture identification and performance parameters.

4. The hybrid encrypted data transmission and storage method adapted to the information technology innovation environment as described in claim 1, characterized in that, The process of predicting the performance of the national cryptographic algorithm using a cache prediction network based on the heterogeneous instruction set capability feature vector, and outputting a performance prediction matrix, includes: Using the heterogeneous instruction set capability feature vector as input, and employing a cache prediction network sensitive to elliptic curve domain operations, performance prediction models for SM2 point operations, SM4 lookup operations, and SM9 bilinear pairing operations are established. The performance prediction models are used to predict the execution performance of national cryptographic algorithms on different domestically developed CPUs, generating a performance prediction matrix. The performance prediction matrix includes algorithm execution time and cache hit rate.

5. The hybrid encrypted data transmission and storage method adapted to the information technology innovation environment as described in claim 4, characterized in that, The prediction process of the cache prediction network, which is sensitive to elliptic curve domain operations, includes: The heterogeneous instruction set capability feature vector is input into a multilayer neural network to extract memory access pattern features of elliptic curve domain operations. The cache behavior of different national cryptographic algorithms is predicted through the multilayer neural network. Based on the cache behavior, the doubling operation time of SM2 point operation, the table lookup delay of SM4 lookup operation, and the pairing calculation time of SM9 bilinear pairing operation are calculated. The performance prediction matrix is ​​established based on the doubling operation time, table lookup delay, and pairing calculation time.

6. The hybrid encrypted data transmission and storage method adapted to the information technology innovation environment as described in claim 1, characterized in that, Based on the performance prediction matrix, a deep reinforcement learning network is used to detect resource conflicts among combinations of national cryptographic algorithms. The combinations of national cryptographic algorithms are then filtered according to the resource conflicts, and the optimal combination of national cryptographic algorithms is output, including: Using the performance prediction matrix as input, a deep reinforcement learning network based on transmission and storage conflict awareness is used to detect transmission resource conflicts in the data transmission stage and storage resource conflicts in the data storage stage of the national cryptographic algorithm combination. Conflict evaluation index is constructed based on the transmission resource conflicts and storage resource conflicts. Candidate national cryptographic algorithm combinations are screened based on the conflict evaluation index, and the optimal national cryptographic algorithm combination is output.

7. A hybrid encrypted data transmission and storage method adapted to the information technology innovation environment as described in claim 6, characterized in that, The selection process for the deep reinforcement learning network based on transmission-to-storage conflict awareness includes: The performance prediction matrix is ​​input into the deep reinforcement learning network to generate an environment state vector. Candidate national cryptographic algorithm combinations are selected based on the environment state vector. The transmission and storage conflict degree of the candidate national cryptographic algorithm combinations is evaluated. The conflict evaluation index is updated based on the transmission and storage conflict degree. The execution state generation process and the algorithm combination selection process are iterated until convergence. The candidate national cryptographic algorithm combination with the minimum conflict degree is selected to obtain the optimal national cryptographic algorithm combination.

8. The hybrid encrypted data transmission and storage method adapted to the information technology innovation environment as described in claim 1, characterized in that, The process of encrypting the data to be transmitted using the optimal national cryptographic algorithm combination and outputting encrypted transmission data includes: The data to be transmitted is adaptively segmented based on the encryption characteristics of each national cryptographic algorithm in the optimal national cryptographic algorithm combination. Based on the adaptive segmentation result, the corresponding encryption parameter combination is configured, and the segmented data and encryption parameter combination are output. The optimal national cryptographic algorithm combination is used to perform multi-level concatenated encryption on the block data according to the encryption parameter combination. The concatenated encryption results of each block data are then reassembled to output the encrypted transmission data.

9. A hybrid encrypted data transmission and storage method adapted to the information technology innovation environment as described in claim 1, characterized in that, The process of transmitting the encrypted data to the target node, decrypting the encrypted data using the optimal national cryptographic algorithm combination, re-encrypting and storing the data, and outputting securely stored data includes: The encrypted transmission data is transmitted to the target node through a secure transmission channel. During the transmission process, the optimal national cryptographic algorithm combination is used to generate a data integrity check code. The integrity of the encrypted transmission data after transmission is completed is verified based on the data integrity check code, and the received data that has passed the verification is output. At the target node, the received data is decrypted in reverse order using the optimal national cryptographic algorithm combination. Based on the security level requirements of the target node's storage environment, a new storage encryption mode is selected, and the decrypted data is re-encrypted using the new storage encryption mode to output the secure stored data.

10. A hybrid encrypted data transmission and storage system adapted to the domestic IT innovation environment, used to execute the hybrid encrypted data transmission and storage method adapted to the domestic IT innovation environment as described in any one of claims 1-9, characterized in that, The system includes: The heterogeneous instruction set adaptation module is used to acquire the hardware information of the domestically developed CPU, train the adaptive neural network based on the hardware information of the domestically developed CPU, learn the differences in the implementation of the national cryptographic algorithm instruction set by different domestically developed CPU architectures, and output the heterogeneous instruction set capability feature vector. The cache performance prediction module is used to predict the performance of the national cryptographic algorithm based on the heterogeneous instruction set capability feature vector and a cache prediction network, and output a performance prediction matrix. The algorithm combination optimization module is used to detect resource conflicts between combinations of Chinese cryptographic algorithms based on the performance prediction matrix using a deep reinforcement learning network, filter the combinations of Chinese cryptographic algorithms according to the resource conflicts, and output the optimal combination of Chinese cryptographic algorithms. The data encryption processing module is used to encrypt the data to be transmitted using the optimal national cryptographic algorithm combination and output encrypted transmission data. The secure transmission and storage module is used to transmit the encrypted transmission data to the target node, decrypt the encrypted transmission data using the optimal national cryptographic algorithm combination, re-encrypt and store the data, and output securely stored data.

Images