Inner product encryption scheme supporting privacy protection and traffic situation prediction expansion application

Abstract

The invention provides an inner product encryption scheme supporting privacy protection and a traffic situation prediction expansion application. A key generation center generates a system public parameter, a family public / private key and a decryption key; meanwhile, the decryption key is sent to the cloud server; a data owner uses an inner product encryption and data encryption algorithm to generate a corresponding ciphertext and upload the ciphertext to a cloud server; and the cloud server decrypts the data by using the decryption key to obtain a related inner product. On the basis of the data processing, the decrypted inner product information is input into a prediction model for safety data prediction. According to the invention, traffic situation prediction is realized by using an existing prediction model with good performance, a perfect traffic situation prediction system is constructed, under the condition of realizing the same prediction accuracy, storage space and computing resources are effectively saved, and the system is enabled to accord with a more emergency traffic scene.

Description

Technical Field

[0001] The present invention belongs to the field of data encryption, and specifically relates to an inner product encryption scheme supporting privacy protection and an expanded application of traffic situation prediction. Background Art

[0002] With the continuous improvement of economic levels, private car travel has become a common mode of transportation. However, with the surge in the number of vehicles, traffic congestion has become increasingly serious, causing numerous inconveniences for travelers, such as longer travel times and increased fuel consumption. Traffic management departments are constantly seeking possible solutions to alleviate travel pressures. In recent years, with the continuous development and maturity of machine learning, data prediction, such as neural network prediction, has provided an effective solution. Neural network prediction uses historical real-world data to train a model and then inputs current data for prediction. Based on this approach, traffic management departments can use existing data to predict traffic trends, achieve precise traffic management and prevention, and minimize traffic congestion.

[0003] However, the inherent nature of neural network predictions prevents users from controlling the data processing process. This transfer of physical control over data raises concerns about the security of sensitive data. Consequently, user concerns about data security have become a serious bottleneck hindering the application of neural network predictions.

[0004] To protect data security, there are currently two main solutions: one is to use distributed learning, such as federated learning, and the other is to utilize a "data processing + data prediction" approach, such as data encryption, differential privacy, and secure multi-party computation. Specifically, in the distributed learning approach, federated learning involves users training their own data locally and uploading model parameters to a central server, which then completes the final prediction. In the "data processing + data prediction" approach, differential privacy continuously adds noise to the data to achieve data security. Functional encryption and homomorphic encryption in data encryption use the method of encrypting plaintext data into ciphertext to hide private information in the data. Homomorphic encryption allows operations to be performed on ciphertext data to obtain encrypted results, while functional encryption allows calculations on ciphertext data to obtain function values ​​as return results.

[0005] Inner product encryption, as an important branch of function encryption, decrypts data into the inner product of the ciphertext corresponding to the plaintext and the key corresponding to a certain vector. The inner product result can be used in scenarios such as data analysis and machine learning. Therefore, it is very reasonable and necessary to combine inner product encryption with neural network prediction to achieve data security prediction. However, according to the inventor's understanding, in the process of application to traffic scenarios, due to the real-time variability of traffic scenarios, higher prediction efficiency is required to make a faster response to the current traffic situation. Therefore, improving prediction efficiency under the "data processing + data prediction" method is an inevitable choice for application in traffic scenarios.

[0006] As mentioned above, data security and prediction efficiency are both crucial practical requirements in traffic situational awareness. However, most current solutions fail to simultaneously address both of these requirements. This is primarily due to the following reasons: 1) Data privacy. Most current prediction models fail to adequately consider data security, failing to effectively protect the security of training and prediction data, leading to cyberattacks. 2) Data prediction efficiency. Combined with data processing, model prediction efficiency is significantly reduced, making timely responses impossible. 3) Long-term security is infeasible. Most current inner product encryption schemes are based on traditionally difficult problems in number theory, such as large integer prime factorization and discrete logarithms, and therefore cannot guarantee data security under quantum attacks. Summary of the Invention

[0007] The purpose of this invention is to design an inner product encryption scheme that supports privacy protection and its extended application in traffic situation prediction, which solves the problems of data privacy security and data prediction efficiency.

[0008] To achieve the goal of data privacy protection, the present invention proposes an inner product encryption scheme, which includes the following steps:

[0009] Step A, Inner Product Encryption. System Establishment: The key generation center runs the algorithm to generate system public parameters, family public key, family private key and decryption key.

[0010] Step B, Inner Product Encryption. Key Extraction: After receiving the registration request from the user and the cloud server, the key generation center runs the algorithm to generate the decryption key and returns it to the cloud server.

[0011] Step C, Inner Product Encryption. Data Encryption: The user uses the family private key and the algorithm to generate the corresponding data ciphertext and upload it to the cloud server.

[0012] Step D, inner product encryption and data decryption: The user sends a decryption request, and the cloud server uses the decryption key to run the algorithm to decrypt the data and obtain the inner product of the relevant data.

[0013] Furthermore, the step A specifically includes:

[0014] The key generation center randomly selects a uniform vector 、 ,Label and a Gaussian distribution ;

[0015] run Algorithm output family private key and the family public key ;

[0016] Public family key And secretly store the family private key ;

[0017] in, Can be seen as , in which ; is the input vector and tags Algorithm for generating system family public keys and family private keys.

[0018] Furthermore, the step B specifically includes:

[0019] Set up two secure hash functions with fixed output lengths as well as ;

[0020] for ,calculate And combine all Get attribute strategy ;

[0021] Using logical functions ,calculate ;

[0022] According to the above results, run Get the decryption key And send it to the cloud server;

[0023] Among them, secure hash function and : Maps an input string of arbitrary length to a string of fixed length and range; Is the system identifier used for attribute strategy generation; logical function , which is stated as well as Represented as function vector and user attributes respectively; The algorithm is a pre-image sampling algorithm, which is mainly used to extract the private key.

[0024] Furthermore, the step C specifically includes:

[0025] Based on user attributes and equation , calculate its attribute hash value and The value of , and combine ;

[0026] Select Vector ,use Combined Message And calculate the ciphertext , upload the encrypted data to the cloud server;

[0027] Among them, for the attribute , It is through the hash function And the equation calculated; It is the remainder operation, which is mainly to fix the calculated value at between.

[0028] Furthermore, the step D specifically includes:

[0029] The cloud server receives the ciphertext , function vector And decryption key calculation ;

[0030] calculate , and the decrypted result Return to the user.

[0031] To achieve data prediction under data privacy protection, the present invention proposes the following traffic situation awareness system:

[0032] Step A, system initialization: The traffic management department runs the inner product encryption. The system establishes an algorithm to generate the corresponding family public key and family private key; then runs the inner product encryption. The key extraction algorithm generates the data decryption key and sends it to the cloud server.

[0033] Step B: Traffic data encryption: The traffic management department uses inner product encryption and data encryption algorithm to generate data ciphertext and send it to the cloud server.

[0034] Step C, traffic data security prediction: Based on the received data ciphertext, the cloud server uses the decryption key to run the inner product encryption. The data decryption algorithm produces the corresponding inner product decryption result; then, the cloud server inputs this result data into the trained neural network prediction model to complete the security prediction.

[0035] The traffic data security prediction is achieved by using the designed inner product encryption scheme to complete data processing and then relying on the excellent neural network prediction model proposed by Zhang et al. [1]. Reference [1] is M. Zhang, Z.-A. Li, and P. Zhang, “A secure and privacy-preserving word vector training schemebased on functional encryption with inner product predicates,” Computer Standards & Interfaces, vol. 86, p. 103734, 2023.

[0036] Furthermore, step A is completed using inner product encryption system establishment and inner product encryption key extraction algorithms.

[0037] Furthermore, step B is completed using inner product encryption and data encryption algorithm.

[0038] Furthermore, step C is divided into two parts: data decryption and data prediction. Data decryption is done using the inner product encryption algorithm, and data prediction is done using the prediction model designed in [1].

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] This paper proposes a data processing method to achieve secure data prediction and protect data privacy during the prediction process. This method allows users (or data senders) to maintain data privacy when using cloud servers for data prediction. No entity, including the cloud server, can extract useful information from the processed data, effectively meeting the data privacy requirements of real-world scenarios.

[0041] This invention combines the need for secure data prediction with inner product encryption to propose a data processing method for achieving secure data prediction. This invention uses inner product encryption to process data before inputting it into the prediction model, effectively protecting data privacy. Furthermore, this invention utilizes the advanced tool g-trapdoor to effectively reduce the size of ciphertext and keys, thereby improving data prediction efficiency.

[0042] This invention balances efficiency and security in its design. Specifically, it effectively resists selected attribute and message attacks in a random oracle model. Leveraging the inherent properties of lattice cryptography, it also offers security against quantum attacks. Furthermore, it dynamically adjusts the decryption inner product length to ensure prediction efficiency for varying application requirements. Compared to existing data processing methods, it achieves higher space and computational efficiency, effectively conserving system resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 Predict the overall architecture diagram for the neural network;

[0044] Figure 2 This is a model architecture diagram of the inner product encryption method of the present invention;

[0045] Figure 3 This is a model architecture diagram of the data prediction method of the present invention;

[0046] Figure 4 Comparison of the inner product encryption scheme of the present invention with existing results in terms of family public key length;

[0047] Figure 5 Comparison of the inner product encryption scheme of the present invention with existing results in terms of decryption key length;

[0048] Figure 6 Comparison of the inner product encryption scheme of the present invention with existing results in terms of ciphertext length;

[0049] Figure 7 Comparison of the inner product encryption scheme of the present invention with existing results in terms of plaintext and ciphertext expansion factors;

[0050] Figure 8 Comparison of the computational efficiency of the key extraction algorithm between the inner product encryption scheme of the present invention and existing results;

[0051] Figure 9 Comparison of the computational efficiency of the inner product encryption scheme of the present invention with existing results in data encryption algorithms;

[0052] Figure 10 Comparison of the computational efficiency of the data decryption algorithm between the inner product encryption scheme of the present invention and existing results;

[0053] Figure 11 This is a radar chart comparing the effects of the inner product encryption scheme of the present invention with existing achievements. DETAILED DESCRIPTION

[0054] The present invention will be fully and clearly described below in conjunction with the accompanying drawings and specific implementation cases in the embodiments of the present invention. This embodiment is implemented based on the technical solution of the present invention. It should be understood that the specific embodiments are only used to illustrate the present invention and not to limit the scope of the present invention.

[0055] The present invention provides an inner product encryption scheme supporting privacy protection and an expanded application of traffic situation prediction.

[0056] First, let’s explain the terms:

[0057] Inner product encryption: Inner product encryption technology is a cryptographic primitive that decrypts data into the inner product of the ciphertext corresponding to the plaintext and the key corresponding to a certain vector.

[0058] The basic principle of neural network prediction: The neural network prediction model is a machine learning model based on artificial neural networks. It realizes model training of past data and prediction of existing data by constructing a network structure composed of multiple neurons connected.

[0059] Ideal lattice: The ideal lattice is a generalization of the concept of cyclic lattice. It can effectively reduce the size of the lattice and is an effective tool for constructing lattice-based cryptographic systems.

[0060] The symbols required in this embodiment are described as shown in Table 1.

[0061] Table 1: Symbols

[0062] This example explains the concepts and design tools involved:

[0063] Grid: a group of dimensional linearly independent vectors Composition is dimensional matrix .but Vig can be defined as .

[0064] - Trapdoor: Assumption , and a trapdoor tag .vector One - A trapdoor is defined as satisfying A matrix .

[0065] Discrete Gaussian distribution: Given a vector ,parameter as well as Subsets on , then define the Gaussian distribution ; ; grid With parameters The discrete Gaussian distribution of is defined as follows: Above The center is The discrete Gaussian distribution of can be expressed as .

[0066] Gaussian sampling algorithm: input random vector , a label Trapdoor , a ring element and Gaussian parameters , the algorithm outputs the vector , satisfying the equation .

[0067] RLWE problem: given parameters and a Gaussian error distribution .one -LWE problem instance is to query the challenge oracle , the oracle contains the vector Pseudo-random sampling oracle and a truly random sampling oracle . Then the LWE problem is defined as follows: :The samples output by this oracle come from the distribution ,in ; :The oracle outputs a uniform random sample from the distribution ; -LWE problem instances can be repeatedly answered by the oracle For any vector , the adversary can distinguish LWE instances if the advantage Cannot be ignored.

[0068] The overall architecture of this embodiment consists of two parts. First, the data is processed using inner product encryption, and then the processed data is input into the prediction model to obtain the prediction results. The architecture diagram of the data processing stage using inner product encryption is as follows: Figure 2As shown, it specifically includes four algorithms: inner product encryption. system establishment, inner product encryption. key generation, inner product encryption. data encryption, and inner product encryption. data decryption. Specifically, inner product encryption. system establishment refers to the key generation center running the algorithm to generate system public parameters, family public keys, and family private keys. Inner product encryption. key extraction phase refers to the key generation center running the key extraction algorithm to generate a decryption key and return it to the cloud server. In the inner product encryption. data encryption phase, the user uses the data encryption algorithm to generate the corresponding data ciphertext and upload it to the cloud server. In the inner product encryption. data decryption phase, the user sends a decryption request, and the cloud server uses the decryption key to run the data decryption algorithm to decrypt the data and obtain the inner product of the relevant data. The decrypted inner product data is input into the prediction model for secure data prediction. The specific process includes three stages: system initialization, traffic data encryption, and traffic data security prediction. Figure 3 In short, the system initialization and data encryption phases rely primarily on the inner product encryption system setup, inner product encryption key extraction, and inner product encryption data encryption algorithms. In the security prediction phase, in addition to utilizing the inner product encryption data decryption algorithm to generate the inner product decryption result, the cloud server needs to feed this result data into a trained neural network prediction model to complete the security prediction.

[0069] The embodiment of the present invention takes the cloud computing environment as the research background and applies the proposed inner product encryption to construct a traffic situation awareness model, which includes a cloud server and a traffic management department.

[0070] System Initialization: This phase is performed by the traffic management department and consists of two parts. Specifically, the traffic management department runs the inner product encryption system establishment algorithm and the inner product encryption key generation algorithm to generate the family public key / family private key and the decryption key.

[0071] Generation of family public key / family private key: The traffic management department runs inner product encryption. The system establishes an algorithm to generate family public key and family private key, among which the family public key is made public and the family private key is stored secretly.

[0072] Among them, the specific details of the inner product encryption system establishment algorithm are:

[0073] The key generation center randomly selects a uniform vector 、 ,Label and a Gaussian distribution ;

[0074] run Algorithm output family private key and the family public key ;

[0075] Public family key And secretly store the family private key ;

[0076] in, Can be seen as , in which ; is the input vector and tags Algorithm for generating system family public keys and family private keys.

[0077] Decryption Key Generation: The traffic management department runs the inner product encryption and key extraction algorithm to generate the decryption key and send it to the cloud server. The specific process is as follows:

[0078] Traffic management department uses the logic function defined in inner product encryption ,calculate .

[0079] Traffic management department execution algorithm To generate the decryption key , for all .

[0080] Sent to the cloud server to complete data decryption.

[0081] Among them, the specific details of the inner product encryption key extraction algorithm are:

[0082] Set up two secure hash functions with fixed output lengths as well as ;

[0083] for ,calculate And combine all Get attribute strategy ;

[0084] Using logical functions ,calculate ;

[0085] According to the above results, run Get the decryption key And send it to the cloud server;

[0086] Among them, secure hash function and : Maps an input string of arbitrary length to a string of fixed length and range; Is the system identifier used for attribute strategy generation; logical function , which is stated as well as Represented as function vector and user attributes respectively; The algorithm is a pre-image sampling algorithm, which is mainly used to extract the private key.

[0087] Traffic data encryption: This stage is completed by the traffic management department, using the inner product encryption. Data encryption algorithm generates data ciphertext and sends it to the cloud server. Specifically,

[0088] Traffic management departments use the properties of cloud servers Encrypting individual data features To obtain the ciphertext .

[0089] Among them, the process of inner product encryption and data encryption algorithm is:

[0090] Based on user attributes and equation , calculate its attribute hash value and The value of , and combine ;

[0091] Select Vector ,use Combined Message And calculate the ciphertext , upload the encrypted data to the cloud server;

[0092] Among them, for the attribute , It is through the hash function And the equation calculated; It is the remainder operation, which is mainly to fix the calculated value at between.

[0093] Traffic data security prediction: This stage is completed by the cloud server and consists of two parts: data decryption completed by the inner product encryption and data decryption algorithm, and data prediction completed by the prediction algorithm.

[0094] Data decryption: After receiving the ciphertext from the traffic management department, the cloud server uses the inner product encryption and data decryption algorithm to decrypt the ciphertext and obtain the data inner product. Specifically, for .

[0095] Among them, the specific process of the inner product encryption and data decryption algorithm is:

[0096] The cloud server receives the ciphertext , function vector And decryption key calculation ;

[0097] calculate , and the decrypted result Return to the user.

[0098] Data prediction: based on inner product results , deviation vector And the activation function , the cloud server completes this stage as follows:

[0099] for , using the bias vector Calculation of the value .

[0100] enter To the activation function Get results , then the result can be expressed as .

[0101] Run the data prediction algorithm to get .

[0102] Compare The value in to find the coordinates of the maximum value of the vector, then predict the label Sent to traffic management.

[0103] Among them, the neural network prediction algorithm FNPP proposed by Zhang et al. can be briefly described as:

[0104] Input the first layer calculation results of the prediction model , The weight vector, bias vector and function in each layer ,run get , the specific details are:

[0105] for ,calculate ,

[0106] if ,otherwise .

[0107] Calculated , the output result is expressed as .

[0108] The data security prediction model provided in this embodiment has the following improvements:

[0109] The data prediction solution proposed in this paper is completed in three stages: system initialization, data encryption, and secure data prediction. Prior to data prediction, an inner product encryption algorithm is embedded to process the raw data. This model enables secure data prediction while protecting data privacy. No entity, including the cloud server, can obtain useful information from the processed data.

[0110] The inner product encryption scheme proposed in this paper includes system establishment, key extraction, data encryption, and data decryption algorithms. To achieve efficient data analysis, the decrypted inner product length can be dynamically adjusted to meet the prediction efficiency requirements of different applications and save system computing and storage resources.

[0111] The solution provided in this embodiment has the following functions:

[0112] This paper utilizes the proposed inner product encryption scheme in a neural network prediction model to protect data security while achieving data prediction. This method not only ensures data prediction results but also prevents data privacy leaks caused by plaintext data prediction. This prevents any entity, including the cloud server, from accessing the predicted data, effectively protecting the privacy of the predicted data.

[0113] Improved data prediction efficiency. The proposed inner product encryption scheme's decrypted inner product length is independent of the scheme's security parameters. This decrypted inner product length can be dynamically adjusted based on application requirements, optimizing the efficiency of the data prediction model. This feature is more adaptable to different application scenarios, fully leveraging the computational advantages of inner product encryption data processing to meet more urgent data prediction needs.

[0114] Effectively resisting quantum attacks. Since 1994, when Shor proposed the Shor algorithm, cryptographic systems designed based on traditional difficult problems such as large integer prime factorization and discrete logarithms cannot guarantee security under quantum attacks. Research has found that lattice-based cryptography can effectively resist quantum attacks. Therefore, this invention utilizes lattice cryptographic tools to complete the design scheme, ensuring the long-term security of the design.

[0115] Safety demonstration of this embodiment:

[0116] The present invention utilizes the proposed inner product encryption scheme to complete data processing and ensure the security of data in the traffic situation awareness model. Therefore, the present invention utilizes the security of the inner product encryption scheme to meet the security requirements of the traffic situation prediction system.

[0117] First, the security of the inner product encryption scheme is demonstrated:

[0118] Under the random oracle model, the inner product encryption algorithm proposed in this paper can achieve security under selected attributes and selected keywords. At the same time, due to the design of the lattice cryptosystem, this inner product encryption scheme can also meet quantum security requirements. The brief security proof is as follows:

[0119] Assume that the adversary can break the indistinguishability of the chosen plaintext attack of the inner product encryption scheme with probability Then the challenger uses the adversary as a subroutine to try to distinguish between two distributions (LWE distribution and random distribution), and the probability is also Specifically, the challenger responds to the adversary's inquiries about the system identifier, the hash value of the user attribute, and the decryption key. After the adversary obtains enough information, the challenger initiates a challenge based on the previously determined challenge message. The adversary makes a guess based on the challenge ciphertext. The adversary's guessing process is the process of distinguishing the two distributions, which can also be seen as the process of cracking the LWE difficulty problem. In this process, the challenger will challenge the distribution A part of is disguised as the family public key, thereby embedding the LWE problem. If the adversary successfully guesses it, it proves that it has the ability to distinguish between LWE distribution and random distribution.

[0120] Based on the security of the inner product encryption scheme, the security of the traffic situational awareness system can be effectively guaranteed. In this system, two main types of adversaries are considered: an honest but curious cloud server and an external attacker. The honest but curious cloud server's primary responsibility is to store all ciphertext data and perform neural network predictions. However, after data processing, the cloud server only obtains the ciphertext of the traffic data and the decrypted inner product. During the data prediction process, the cloud server only obtains the predicted label and no other valuable information. External attackers can only obtain the ciphertext and decrypt the inner product by eavesdropping on the channel. Therefore, neither the cloud server nor the external attacker can obtain any useful information from the data.

[0121] Functional demonstration of this embodiment:

[0122] Functional comparison of the inner product encryption scheme of the present invention and related comparative literature: As shown in Table 2, the inner product encryption scheme proposed in the present invention has advantages in encryption and decryption algorithm running time and post-quantum security, and is more in line with the security and efficiency requirements of traffic prediction scenarios.

[0123] Among them, [2] X. Qian, H. Li, M. Hao, G. Xu, H. Wang, and Y. Fang, “Decentralized multi-client functional encryption for inner product with applications to federated learning,” IEEE Transactions on Dependable and Secure Computing, pp. 1–16, 2024.; [3] H. Yang, Y. Su, J. Qin, and H. Wang, “Privacy-preserving outsourced inner product computation on encrypted database,” IEEE Transactions on Dependable and Secure Computing, vol. 19, no. 2, pp. 1320–1337, 2020.; [4] J. Han, L. Chen, W. Susilo, L. Chen, and G. Wu, “Dfe-ip: Delegatable functional encryption for inner product,” Information Sciences, vol. 647, p. 119425, 2023.; [5] J. Han, L. Chen, A. Hu, L. Chen, and J. Li, “Privacy-preserving decentralised functional encryption for inner product,” IEEE Transactions on Dependable and Secure Computing, vol. 21, no. 4, pp. 1680–1694, 2023.; [6] H. Yang and C. Peng, “Cp-ipfe: Ciphertext-policy based inner product functional encryption,” IEEE Transactions on Information Forensics and Security, vol. 19, pp. 5419–5433, 2024.;[7]M. Abdalla, F.Bourse, H. Marival, D. Pointcheval, A. Soleimanian, and H. Waldner, “Multi-client inner-product functional encryption in the random-oracle model,” inSecurity and Cryptography for Networks: 12th International Conference, SCN2020, Amalfi, Italy, September 14 16, 2020, Proceedings 12. Springer, 2020,pp. 525–545.;[8] K. Yun and R. Xue, “Chosen ciphertext attacks secure inner-product functional encryption from learning with errors assumption,” inInternational Conference on Information Security and Cryptology. Springer,2020, pp. 133–147.;[9]V. Cini, S. Ramacher, D. Slamanig, C. Striecks, and E.Tairi, “(inner product) functional encryption with updatable ciphertexts,”Journal of Cryptology, vol. 37, no. 1, p. 8, 2024.;

[10] K. Yun, X. Wang, andR. Xue, “Identity-based functional encryption for quadratic functions fromlattices,” in International Conference on Information and CommunicationsSecurity. Springer, 2018, pp. 409 425.;

[11] N. Yang, S. Yang, Y. Zhao, W. Wu,and X.Wang, “Inner product encryption from middle-product learning witherrors,” Computer Standards&Interfaces, vol. 87, p. 103755, 2024. Table 2: Functional comparison of inner product encryption schemes

[0124] in, They represent the number of users, plaintext message length, and attribute length respectively; encryption complexity and decryption complexity refer to the running time of the encryption and decryption algorithms.

[0125] Space efficiency analysis of the inner product encryption scheme designed by this invention: As shown in Table 3, the algorithm designed by this invention has significant advantages in terms of family public key length and plaintext and ciphertext expansion factor, effectively conserving system resources. Furthermore, the data processing method proposed in this invention is designed based on an ideal lattice. Compared to random lattices, it improves the construction of trapdoors and other methods, further improving efficiency.

[0126] Table 3: Comparison of space efficiency of inner product encryption schemes

[0127] in, is the matrix dimension, The matrix dimension refers to the number of matrix columns, is the infinite norm of the message, is the message length, Refers to the common parameters in [2] and the length of a single message, It refers to the dimension and degree of the polynomial in

[11] .

[0128] Comparison of the computational efficiency of the inner product encryption scheme proposed in this invention: As shown in Table 4, this scheme is designed using only hash operations, matrix multiplication, and Gaussian sampling algorithms. The computational cost during the operation is relatively low. Compared with other schemes, it has potential computational advantages and is more suitable for traffic data prediction scenarios.

[0129] Table 4: Comparison of computational efficiency of inner product encryption schemes

[0130] Functional verification of this embodiment:

[0131] Take the values ​​of each parameter in the scheme, calculate the space efficiency and time efficiency of the inner product encryption scheme, and intuitively reflect the advantages of this scheme compared with existing schemes. The specific parameters are set as .

[0132] Comparison of the inner product encryption scheme designed by the present invention with the relevant references in terms of family public key: Figure 4 As shown in the figure, compared with other existing literature, the family public key length of the designed scheme has obvious advantages and can effectively reduce the consumption of storage space.

[0133] Comparison of the inner product encryption scheme designed by the present invention with the relevant references in terms of decryption keys: Figure 5 As shown in the figure, compared with the existing literature [8][9]

[10] , the decryption key length of the designed scheme has obvious advantages and can effectively reduce the consumption of storage space. However, compared with the literature

[11] , the space storage efficiency of the proposed scheme is slightly lower.

[0134] Comparison of the inner product encryption scheme designed by the present invention with the relevant references in terms of ciphertext: Figure 6 As shown in the figure, compared with the existing literature [8]

[10]

[11] , the ciphertext length of the designed scheme has potential advantages and can effectively reduce the consumption of storage space. However, compared with the literature [9], the ciphertext length of the proposed scheme is slightly longer.

[0135] Comparison of the inner product encryption scheme designed by the present invention with the related references in terms of plaintext and ciphertext expansion factors: Figure 7 As shown in the figure, compared with other existing literature, the designed scheme has obvious advantages in the comparison of plaintext and ciphertext expansion factors. When encrypting the same message length, the ciphertext length increases less.

[0136] In summary, the inner product encryption scheme designed by the present invention has less overall storage space and is more suitable for the data processing stage of the neural network prediction model.

[0137] Comparison of the inner product encryption scheme designed by the present invention with the relevant references in terms of key extraction: Figure 8 As shown in the figure, the efficiency of the key extraction algorithm of the designed scheme has obvious advantages compared with other existing literature [9]

[10] , and the efficiency of literature [8]

[11] is slightly higher than that of this scheme.

[0138] Comparison of the inner product encryption scheme designed by the present invention with the relevant references in data encryption: Figure 9 As shown in the figure, compared with other existing literature, the efficiency of the data encryption algorithm of the designed scheme has obvious advantages.

[0139] Comparison of the inner product encryption scheme designed by the present invention with the relevant references in data decryption: Figure 10 As shown in the figure, the efficiency of the data decryption algorithm of the designed scheme has obvious advantages compared with other existing literature [8]

[10]

[11] , and the efficiency of literature [9] is slightly higher than that of this scheme.

[0140] Therefore, the inner product encryption scheme designed by the present invention is more suitable for the data processing stage of the neural network prediction model.

[0141] The inner product encryption scheme of this embodiment is compared with the radar chart: Figure 11 As shown, the inner product encryption scheme designed by the present invention achieves long-term security and has higher efficiency, which is suitable for application scenarios of data prediction.

[0142] Based on the potential space and computational advantages of the inner product encryption scheme, the efficiency of the designed traffic situation awareness system is significantly improved, as shown in Table 5. Since this system is based on the prediction model proposed in reference [1] and has improved data processing, the data prediction accuracy of this system is the same as that of reference [1].

[0143] Table 5: Comparison of storage space and computing efficiency of traffic situation awareness systems plan Running memory Storage Runtime [1] 162.00KB 64.00KB 76.23ms This system 37.63KB 36.01KB 49.31ms

[0144] Among them, running memory refers to the space required by the data prediction model to store data during operation; storage space refers to the space required by the data prediction model to permanently store data; and running time refers to the time required to run the data prediction model.

Claims

1. The inner product encryption scheme supporting privacy protection and the extended application of traffic situation prediction are characterized by: The following steps are involved: Step 1: Taking the cloud computing environment as the research background, an inner product encryption model in cloud computing is constructed, which includes cloud servers, key generation centers and users; Step 2: Inner product encryption. System establishment: The key generation center runs the algorithm to generate system public parameters, family public key and family private key; Step 3: Inner product encryption. Key extraction: After receiving the registration request from the user and the cloud server, the key generation center runs the algorithm to generate the decryption key and returns it to the cloud server; Step 4: Inner product encryption. Data encryption: The user uses the family private key and the algorithm to generate the corresponding data ciphertext and upload it to the cloud server; Step 5: Inner product encryption. Data decryption: The user sends a decryption request, and the cloud server uses the decryption key to run the algorithm to decrypt the data and obtain the inner product of the relevant data.

2. The inner product encryption scheme supporting privacy protection and the extended application of traffic situation prediction are characterized by: The following steps are involved: Step 1: Taking the cloud computing environment as the research background, the proposed inner product encryption is applied to complete data processing to build a traffic situation awareness system, which includes a cloud server and a traffic management department; Step 2: System initialization: The traffic management department runs the inner product encryption. The system establishes an algorithm to generate the corresponding family public key and family private key; then runs the inner product encryption. The key extraction algorithm generates the data decryption key and sends it to the cloud server; Step 3: Traffic data encryption: The traffic management department uses inner product encryption. Data encryption algorithm generates data ciphertext and sends it to the cloud server; Step 4: Traffic data security prediction: Based on the received data ciphertext, the cloud server uses the decryption key to run the inner product encryption. The data decryption algorithm generates the corresponding inner product decryption result; then, the cloud server sends this result data to the trained neural network prediction model to complete the security prediction.

3. The inner product encryption scheme supporting privacy protection and the extended application of traffic situation prediction as claimed in claim 1, characterized in that: The inner product encryption system establishes the algorithm in the following specific steps: The key generation center randomly selects a uniform vector , ,Label and a Gaussian distribution ; run Algorithm output family private key and the family public key ; Public family key And store the family private key secretly ; in, Can be seen as , in which ; is the input vector and tags Algorithm for generating system family public keys and family private keys.

4. The inner product encryption scheme supporting privacy protection and the extended application of traffic situation prediction as claimed in claim 1, characterized in that: The inner product encryption key extraction algorithm specifically includes the following steps: Set up two secure hash functions with fixed output lengths as well as ; for ,calculate And combine all Get attribute strategy ; Using logical functions ,calculate ; Based on the above results, run Get the decryption key And send it to the cloud server; Among them, secure hash function and : Map the input string of arbitrary length to a string of fixed length and range; Is the system identifier used for attribute strategy generation; logical function , which is stated as well as Represented as function vector and user attributes respectively; The algorithm is a pre-image sampling algorithm, which is mainly used to extract private keys.

5. The inner product encryption scheme supporting privacy protection and the extended application of traffic situation prediction as claimed in claim 1, characterized in that: The inner product encryption scheme and data encryption algorithm specifically include the following steps: Based on user attributes and equation , calculate its attribute hash value and The value of , and combine ; Select Vector ,use Combined Message And calculate the ciphertext , upload the encrypted data to the cloud server; Among them, for the attribute , Through the hash function And the equation Calculated; It is a remainder operation, which is mainly to fix the calculated value at between.

6. The inner product encryption scheme supporting privacy protection and the extended application of traffic situation prediction as claimed in claim 1, characterized in that: The inner product encryption scheme and data decryption algorithm specifically include the following steps: The cloud server receives the ciphertext , function vector And decryption key calculation ； calculate , and the decrypted result Return to the user.

7. The inner product encryption scheme supporting privacy protection and the extended application of traffic situation prediction as claimed in claim 2, characterized in that: The specific implementation of the traffic situation prediction system includes: the traffic situation prediction system relies on the neural network prediction model proposed by Zhang et al. to realize data prediction. The present invention mainly improves the data processing in the system to achieve higher computing and storage efficiency in data prediction.

8. The inner product encryption scheme supporting privacy protection and the extended application of traffic situation prediction as claimed in claim 2, characterized in that: The modifications related to the application of inner product encryption to the traffic situation awareness system include: In the inner product encryption and key extraction phase, Modified to ; In the inner product encryption and data encryption stage, The generation of ; In the inner product encryption and data decryption phase, the ciphertext is Finish; in, is the weight vector in the neural network prediction model, which represents the importance of each data feature in the prediction. ,use Encrypt the inner product in the function vector The reason for the replacement is The data range cannot effectively represent the weight information of the data features of the prediction model.

Images