Privacy protection keyword searching method and system for intelligent high-speed rescue

Abstract

The invention discloses a privacy protection keyword search method and system for intelligent high-speed rescue. The method comprises the steps that 1, a data owner generates an order data file and encrypts keywords of the order data file; step 2, a terminal user obtains a search keyword in the help-seeking information according to the help-seeking information, constructs a randomized ciphertext, and obtains a trap door according to the randomized ciphertext and the search keyword; step 3, the cloud server A processes the ciphertext data and the trap door by using a security unequal test, and constructs a random number file to disturb the successfully matched ciphertext data; 4, the cloud server B decrypts the disturbed ciphertext data to obtain a plaintext search result and sends the plaintext search result to the terminal user; and 5, the terminal user processes the plaintext search result according to the random number file to obtain a final plaintext result corresponding to the search keyword. According to the method, the calculation overhead and the communication cost are reduced, and the privacy, the real-time performance and the expansibility are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of data privacy protection and information retrieval, and particularly to a privacy protection keyword search method and system for intelligent highway rescue. Background Art

[0002] As an important part of intelligent transportation, intelligent highways significantly improve the efficient operation efficiency and emergency management capabilities of highways by introducing intelligent technologies. In the intelligent highway rescue scenario, real-time and efficient keyword search services are particularly important, supporting rescue vehicles to quickly locate accident locations, query rescue orders, extract road condition information, etc. However, rescue services need to process large-scale and real-time dynamic data, which poses stringent requirements on storage and computing resources. The traditional local data processing mode is difficult to meet the needs, and outsourcing data and computing tasks to the cloud has become the mainstream choice, effectively alleviating the load pressure of local devices and improving the response timeliness and accuracy. However, at the same time, data outsourcing brings serious privacy leakage risks. Intelligent highway rescue data contains a large amount of sensitive information, such as accident locations, license plate numbers, and personnel identities. Outsourcing to the cloud is vulnerable to speculative analysis and reconstruction attacks, resulting in privacy leakage. Malicious attackers may infer user intentions by analyzing query patterns or data access frequencies. Therefore, how to protect data privacy while achieving efficient search has become a problem that needs to be solved urgently.

[0003] To address the data leakage problem in the high-efficiency search scenario, existing privacy protection technologies include anonymization, homomorphic encryption, and searchable encryption. First, although anonymization technology reduces the privacy leakage risk in the search process in the high-speed accident rescue scenario by obfuscating data, its ability to resist re-identification attacks is insufficient. Second, homomorphic encryption has received wide attention because it supports encrypted computing, but fully homomorphic encryption has high computational overhead and is not suitable for the real-time requirements of high-speed rescue. Partially homomorphic encryption can achieve a balance between efficiency and privacy protection, but the computational and communication costs are relatively high. Finally, searchable encryption technology supports efficient encrypted search, but the storage and communication overheads increase significantly in large-scale data scenarios, and there are deficiencies in protecting against search pattern leakage and access pattern leakage. Although existing anonymization, homomorphic encryption, and searchable encryption methods can solve the data leakage problem in the high-efficiency search scenario to a certain extent, in the intelligent highway rescue scenario, the above methods still face the following challenges:

[0004] First, the challenge of balancing computational and communication overheads. Intelligent highway rescue needs to process large-scale data in a short time. Traditional homomorphic encryption schemes are difficult to meet the real-time search requirements due to high computational complexity and large encrypted data transmission volume. Frequent encrypted data interaction between local devices and servers further increases the communication burden and reduces the efficiency of the keyword search system.

[0005] Secondly, there is the problem of search and access pattern leakage. Attackers can speculate on data content or user intentions by analyzing user query patterns or data access frequencies. For example, frequently querying accident records may disclose sensitive information such as accident locations and involved personnel. The existing technologies still lack sufficient protection against such attacks.

[0006] Finally, there are problems with real-time performance and scalability. Smart highway rescue has extremely high requirements for the timeliness of search services. However, in the existing methods, when the data volume increases, the computing efficiency drops significantly, making it difficult to respond quickly. In addition, as the system scale expands, the bottlenecks of storage and computing resources become more prominent, resulting in deficiencies in the system's support for large-scale complex scenarios. Summary of the Invention

[0007] In order to at least partially solve the problems of high computing overhead, large communication costs, privacy leakage risks, and deficiencies in real-time performance and scalability in the existing technologies, the present invention provides a privacy-preserving keyword search method and system for smart highway rescue. In the smart highway rescue scenario, the data owner encrypts each line of keywords in the order data file using an optimized Paillier encryption algorithm, generates ciphertext data, and uploads it to cloud server A, reducing the computing and communication costs of the data owner. Secondly, the end user searches for keywords according to the rescue requirements, obtains a trapdoor based on the randomized ciphertext and the search keyword, adds random perturbation information (randomized ciphertext) to the trapdoor to enhance security, and then securely sends the trapdoor to server A for matching and search operations. Thirdly, cloud server A and server B cooperate to run a secure inequality test algorithm based on homomorphic encryption, screen and match the rescue order results, and generate perturbed ciphertext data. Finally, the end user combines and processes the received perturbed ciphertext data with the plaintext search results, thereby successfully recovering the final plaintext results corresponding to the search keywords, ensuring the accuracy of the search and the privacy of the data, while meeting the real-time rescue requirements.

[0008] To achieve the above object, the technical solution of the present invention is:

[0009] The first aspect of the present invention proposes a privacy-preserving keyword search method for smart highway rescue, including:

[0010] Step 1: The data owner receives smart highway information and information of the user in need of rescue, generates an order data file, encrypts the keywords in the order data file, obtains ciphertext data, and uploads the ciphertext data to cloud server A; the keywords in the order data file include the longitude and latitude, mobile phone number, accident date, and number of accident victims of the user in need of rescue; this is convenient for protecting data privacy;

[0011] Step 2: The end user obtains the search keywords in the distress message according to the distress message and constructs a randomized ciphertext. A trapdoor is obtained based on the randomized ciphertext and the search keywords, and the randomized ciphertext and the trapdoor are sent to Cloud Server A to enhance the privacy protection ability;

[0012] Step 3: Cloud Server A processes the ciphertext data and the trapdoor using a secure inequality test to obtain the matching result between the keyword in the ciphertext data and the trapdoor, and constructs a random number file to perturb the ciphertext data with a successful match to obtain the perturbed ciphertext data. The random number file is sent to the end user, and the perturbed ciphertext data is sent to Cloud Server B to improve the keyword matching efficiency;

[0013] Step 4: Cloud Server B decrypts the perturbed ciphertext data to obtain the plaintext search result and sends the plaintext search result to the end user;

[0014] Step 5: The end user processes the plaintext search result according to the random number file to obtain the final plaintext result corresponding to the search keyword.

[0015] Further, the encryption of the keywords in the order data file specifically includes:

[0016] Two mutually independent random prime numbers p and q are generated according to the preset key size Ks, and then a key pair (pk, sk) is generated;

[0017] The key pair (pk, sk) is sent to Cloud Server B, and the public key pk is distributed to the data owner, Cloud Server A, and the end user to facilitate the decryption of the keywords in the encrypted order data file;

[0018] The keywords in the order data file are encrypted one by one using an optimized Paillier algorithm.

[0019] Further, Step 2 specifically includes:

[0020] The end user obtains the search keywords in the distress message according to the distress message;

[0021] The search keyword is multiplied by -1 and then homomorphically encrypted to obtain the ciphertext of the search keyword;

[0022] A random number is generated and homomorphically encrypted to obtain a randomized ciphertext. The randomized ciphertext and the ciphertext of the search keyword are added together to obtain a trapdoor;

[0023] The end user sends the randomized ciphertext and the trapdoor to Cloud Server A.

[0024] Further, the cloud server A processes the ciphertext data and the trapdoor using a secure inequality test to obtain the matching result between the encrypted keyword in the ciphertext data and the search keyword, specifically including:

[0025] The cloud server A reads the encrypted keyword in the ciphertext data and processes the trapdoor and the encrypted keyword using the homomorphic addition property of Paillier encryption to obtain a ciphertext value;

[0026] The cloud server A sends the ciphertext value and the randomized ciphertext to the cloud server B. The cloud server A and the cloud server B cooperate to run a secure inequality test algorithm based on the homomorphic encryption addition property to obtain the matching result between the keyword in the ciphertext data and the trapdoor, and further obtain the matching result between the keyword in the ciphertext data and the search keyword.

[0027] Further, the cloud server A and the cloud server B cooperate to run a secure inequality test algorithm based on the homomorphic encryption addition property, specifically including:

[0028] The server A calculates the encrypted difference using the Paillierr homomorphic encryption addition property;

[0029] The server A sends the encrypted difference to the cloud server B, and the cloud server B decrypts the encrypted difference to obtain the plaintext difference;

[0030] Judge whether the plaintext difference is 0. If the plaintext difference is 0, output that the matching result is incorrect and send the result back to the cloud server A. The ciphertext value and the randomized ciphertext match successfully, and further the matching between the encrypted keyword in the ciphertext value and the search keyword in the trapdoor is successful. If the plaintext difference is not 0, output that the matching result is correct, and return to execute step two for the matching between the encrypted keyword in the next ciphertext value and the search keyword in the trapdoor.

[0031] Further, the encrypted difference is expressed by the following formula:

[0032] Enc(Diff) = Add_Num · (E_R) -1

[0033] where Enc(Diff) is the encrypted difference, Add_Num is the ciphertext value, and E_R is the randomized ciphertext.

[0034] Further, constructing a random number file to perturb the ciphertext data to obtain the perturbed ciphertext data, specifically including:

[0035] Cloud server A generates a random number file with the same dimension size as the ciphertext data that matches the search keyword, homomorphically encrypts each random number, and uses the homomorphic addition property of Paillier encryption to process the encrypted random number file and the ciphertext data that matches the search keyword, and stores the processed result in the ciphertext data to obtain the perturbed ciphertext data, which is convenient for enhancing privacy.

[0036] Further, cloud server B decrypts the perturbed ciphertext data, which specifically includes:

[0037] Cloud server B decrypts the perturbed ciphertext data according to the private key sk in the key pair (pk, sk).

[0038] Further, the specific steps of step five include: The end user processes the plaintext search result according to the random number file to obtain the final plaintext result corresponding to the search keyword;

[0039] Subtract the value of each keyword in the plaintext search result from the corresponding random number in the random number file. If the result is not 0, then store the matching items included in the row where the keyword in the plaintext search result is located into the result file to obtain the final plaintext result. If the result is 0, there is no result of the keyword in the order data file.

[0040] The second aspect of the present invention proposes a privacy-preserving keyword search system for intelligent highway rescue, including:

[0041] An encryption module, which is used to enable the data owner to receive intelligent highway information and the information of the user to be rescued to generate an order data file and encrypt the keywords of the order data file to obtain ciphertext data and upload the ciphertext data to cloud server A; the keywords of the order data file include the longitude and latitude, mobile phone number, accident date, and number of accident victims of the user to be rescued; it is convenient for protecting data privacy;

[0042] A terminal encryption module, which is used to enable the end user to obtain the search keyword in the distress message and construct a randomized ciphertext, obtain a trapdoor according to the randomized ciphertext and the search keyword, and send the randomized ciphertext and the trapdoor to cloud server A, which is convenient for enhancing the privacy protection ability;

[0043] A secure inequality test module, which is used to enable cloud server A to process the ciphertext data and the trapdoor using a secure inequality test to obtain the matching result of the keyword in the ciphertext data and the trapdoor, and construct a random number file to perturb the ciphertext data with successful matching to obtain the perturbed ciphertext data, send the random number file to the end user, and send the perturbed ciphertext data to cloud server B, which is convenient for improving the keyword matching efficiency;

[0044] A decryption module, which is used to enable cloud server B to decrypt the perturbed ciphertext data, obtain the plaintext search result and send the plaintext search result to the end user;

[0045] A matching module, which is used to enable the end user to process the plaintext search result according to the random number file to obtain the final plaintext result corresponding to the search keyword.

[0046] Advantages of the present invention:

[0047] The present invention has significant advantages in terms of computing efficiency, communication cost, and privacy protection ability. By optimizing the probabilistic encryption of homomorphic encryption into deterministic encryption, the computing overhead of data owners is significantly reduced and the amount of ciphertext data is reduced, thereby effectively reducing the communication cost; a secure inequality test algorithm is designed to achieve keyword search in the encrypted state, and the privacy of search results is enhanced by introducing random numbers; the decryption operation is migrated to the cloud to reduce the computing and communication burdens of end users while ensuring the privacy of the system; a dual-cloud server architecture is adopted to effectively prevent a single cloud server from inferring sensitive information by analyzing multiple search requests, greatly improving the anti-analysis ability of the system. Theoretical analysis and experimental verification show that the present invention exhibits excellent performance and practical application value in large-scale data processing scenarios. Description of the drawings

[0048] Figure 1 It is a flowchart of a privacy-preserving keyword search method for intelligent highway rescue provided by an embodiment of the present invention.

[0049] Figure 2 It is a schematic diagram of privacy-preserving keyword search in outsourced data provided by an embodiment of the present invention.

[0050] Figure 3 It is a schematic diagram of the interaction of each entity in privacy-preserving keyword search in outsourced data provided by an embodiment of the present invention.

[0051] Figure 4 It is a schematic diagram of the computing cost of each entity at different security levels provided by an embodiment of the present invention.

[0052] Figure 5 It is a schematic diagram of the communication cost of each entity at different security levels provided by an embodiment of the present invention.

[0053] Figure 6 It is a schematic diagram of the comparison of the computing costs of data providers of different methods at different security levels provided by an embodiment of the present invention.

[0054] Figure 7 It is a schematic diagram of the comparison of the communication costs of data providers of different methods at different security levels provided by an embodiment of the present invention.

[0055] Figure 8 Schematic diagram of the comparison of the computing costs of different methods provided by the embodiments of the present invention for the terminal to search for users at different security levels.

[0056] Figure 9 Architecture diagram of a privacy-preserving keyword search system for intelligent highway rescue provided by the embodiments of the present invention. Detailed implementation manners

[0057] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] Embodiment 1

[0059] As Figure 1 、 Figure 2 and Figure 3 shown, a privacy-preserving keyword search method for intelligent highway rescue includes:

[0060] S101: The data owner receives the intelligent highway information and the information of the user to be rescued, generates an order data file, encrypts the keywords of the order data file, obtains ciphertext data, and uploads the ciphertext data to cloud server A; the keywords of the order data file include the longitude and latitude, mobile phone number, accident date, and number of accident victims of the user to be rescued.

[0061] Specifically, the data owner receives the intelligent highway information and the information of the user to be rescued and generates an order data file. The data owner extracts keywords from the order data file D = {d1, d2,..., d n} according to a preset rule, and encrypts them line by line based on an optimized Paillier encryption algorithm to generate a ciphertext data set E D = {E(d1), E(d2),..., E(d n )}.

[0062] The encryption process is represented as follows: The key generation center (KGC) generates two independent random prime numbers p and q according to a preset key size KS, and then calculates n, λ, and μ, and further generates a key pair (pk, sk). Among them, n is the product of p and q, λ is the security parameter, that is, the least common multiple value of (p - 1) and (q - 1), and μ is the inverse element modulo n, that is, L(g λ mod n 2)), where the L function is L(x) = (x - 1) / n. Subsequently, the KGC sends (pk, sk) to cloud server B and distributes the public key pk to the data owner, cloud server A, and all registered end users. The order data file D contains several order data D = {d1, d2, …, d n}, and uses the optimized Paillier encryption to encrypt each keyword in the order data file to generate a set E of ciphertext data E_File D = {E(d1), E(d2), …, E(d n )}, and sends the ciphertext data E_File to cloud server A. To improve efficiency, the present invention fixes the random number r = 1 to remove the randomness in the encryption process, making the encryption operation deterministic, thereby improving the calculation efficiency.

[0063] S102: The end user obtains the search keyword in the distress message according to the distress message and constructs a randomized ciphertext, obtains a trapdoor according to the randomized ciphertext and the search keyword, and sends the randomized ciphertext and the trapdoor to cloud server A.

[0064] Specifically, the client of the end user obtains the search keyword in the distress message according to the distress message. Multiply the search keyword kw by -1 and perform homomorphic encryption to generate the search keyword ciphertext E_Nkw. Then generate a random number and perform homomorphic encryption to obtain the randomized ciphertext E_R. Subsequently, add E_R to the pre-encrypted search keyword ciphertext E_Nkw to obtain the trapdoor value T, that is, T = add(E_R, E_Nkw). Finally, the end user sends E_R and the trapdoor value t to cloud server A for keyword matching.

[0065] S103: Cloud server A processes the ciphertext data and the trapdoor using a secure inequality test to obtain the matching result of the keyword in the ciphertext data and the trapdoor, and constructs a random number file to perturb the ciphertext data with successful matching to obtain the perturbed ciphertext data, sends the random number file to the end user, and sends the perturbed ciphertext data to cloud server B.

[0066] Specifically, after cloud server A receives T and the ciphertext data E_File, it reads the encrypted keywords in E_File line by line, and performs an operation on the trapdoor T and the encrypted keywords using the homomorphic addition property of Paillier encryption to generate the ciphertext value Add_Num.

[0067] Cloud server A sends the ciphertext Add_Num and the randomized ciphertext E_R to cloud server B. Cloud server A and cloud server B collaboratively run a secure inequality test algorithm based on the homomorphic encryption addition property to determine whether the ciphertext keyword matches the trapdoor value. If the algorithm output is False, it means the keyword match is successful; if the plaintext difference is not 0, the output match result is True, indicating that the keyword match fails, and the result is sent back to cloud server A. If the plaintext difference is not 0, the output match result is correct, and S102 is executed to match the encrypted keyword in the next ciphertext value and the search keyword in the trapdoor.

[0068] Cloud server A generates a random number file random with the same dimension size as the ciphertext data that successfully matches the search keyword, homomorphically encrypts each random number, and performs operations item by item with the entries in the ciphertext data that successfully matches the search keyword using the homomorphic addition property of Paillier encryption. The result is saved in the ciphertext data to obtain the perturbed ciphertext data. Finally, the random number file random is sent to the end user, while the perturbed ciphertext data is sent to cloud server B.

[0069] S104: Cloud server B decrypts the perturbed ciphertext data to obtain the plaintext search result and sends the plaintext search result to the end user.

[0070] Specifically, cloud server B receives the perturbed ciphertext data sent by cloud server A, decrypts it using the private key sk, saves the obtained plaintext as the plaintext search result D_File, and then cloud server B sends the plaintext search result D_File to the end user.

[0071] S105: The end user processes the plaintext search result according to the random number file to obtain the final plaintext result corresponding to the search keyword.

[0072] Specifically, the end user receives the random number file and the plaintext search result from cloud server A and cloud server B respectively. According to the random number file and the plaintext search result, calculate the difference of each keyword (the value of each keyword minus the corresponding random number in the random number file, D_File_Value - random_Value). If the result is not 0, mark that row as containing a matching item and store the entry in the result file result. If the result file result is not empty, it means there is a search match result, and the final plaintext result is obtained; otherwise, there is no matching item, that is, there is no data corresponding to the search keyword in the order data file.

[0073] In the present invention, the data owner generates an order data file and encrypts the keywords therein. The end user obtains search keywords based on the distress information, constructs a randomized ciphertext and obtains a trapdoor, uses cloud server A and cloud server B to perform a secure inequality test, realizes the matching of the keywords in the ciphertext data and the trapdoor, encrypts the ciphertext data to obtain the perturbed ciphertext data, cloud server B decrypts the perturbed ciphertext data, obtains the plaintext search result and sends it to the end user, and finally the end user processes the plaintext search result to obtain the final plaintext result corresponding to the search keywords.

[0074] Embodiment 3

[0075] Based on the above embodiments, the present invention proposes a secure inequality test algorithm, which specifically includes:

[0076] Server A calculates the encrypted difference using the Paillierr homomorphic encryption addition property. The encrypted difference is expressed by the following formula:

[0077] Enc(Diff) = Add_Num·(E_R) -1

[0078] where Enc(Diff) is the encrypted difference, Add_Num is the ciphertext value, and E_R is the randomized ciphertext.

[0079] Server A sends the encrypted difference to cloud server B, and cloud server B decrypts the encrypted difference to obtain the plaintext difference.

[0080] Judge whether the plaintext difference is 0. If the plaintext difference is 0, output a wrong matching result and send the result back to cloud server A. The ciphertext value and the randomized ciphertext match successfully, and thus the encrypted keyword in the ciphertext value and the search keyword in the trapdoor match successfully. If the plaintext difference is not 0, output a correct matching result and return to execute S102 for matching the encrypted keyword in the next ciphertext value and the search keyword in the trapdoor.

[0081] Embodiment 3

[0082] Based on the above embodiments, the present invention proposes an evaluation process for a privacy-preserving keyword search method for intelligent highway rescue, which is specifically as follows:

[0083] Experimental parameters and environment: The experiment used the python-paillier library developed by CSIRO Data61. This library is known for its optimized and rich functions and relies on the "gmpy2" library for fast multi-precision arithmetic. The experiment was based on the Paillier cryptosystem, relying on the difficulty of factoring large integer N to ensure security. To meet the NIST recommended standards for encryption key lengths, security parameters λ of 80, 112, and 128 bits were selected, corresponding to key lengths of 1024, 2048, and 3072 bits. The experiment was conducted on the Windows 10 operating system with a hardware configuration of an Intel i5 processor, 16GB of memory, and a 512GB solid-state drive.

[0084] The experimental dataset was from open data. It covered accident information that occurred on Road A between 2007 and 2022. The data included the accident date, time, cause, type, and the climate conditions on that day, etc. The data was divided by year and stored in different CSV files.

[0085] To verify the efficiency and feasibility in the scenario of outsourcing data in intelligent highways, the present invention evaluated the performance of all parties (data owners, cloud servers, end-users) from two aspects: computational overhead and communication overhead. Since the performance of Paillier encryption is independent of data distribution and is mainly affected by the data volume, different scales of data entries were selected for the experiment to evaluate. The experiment counted the computational and communication overheads of all parties during encryption, decryption, homomorphic operations, data transmission, and search processes. The computational overhead included the time taken for encryption, decryption, and homomorphic operations, and the communication overhead was measured by the file size. The overhead data of end-users was obtained by performing 100 independent experiments. Finally, the present invention compared the experimental results with the computational and communication overheads of the data owners in the Lulu Han scheme (fully privacy-preserving location recommendation in an outsourcing environment) and the computational overhead of end-users in the Malik scheme (a homomorphic method for intelligent airport security and privacy protection). The method has little impact on the communication cost of end-users, so this part of the comparison was omitted.

[0086] Figure 4Shows the relationship between the computational overhead of each entity and the data scale under different key lengths. The results show that as the data volume and key length increase, the system computational overhead gradually rises, especially with a relatively large increase on the cloud server side. The computational overhead of the data owner mainly comes from data encryption. As the data volume increases, the overhead when the key lengths are 1024, 2048, and 3072 bits increases from 0.15 seconds to 0.71 seconds, from 1.18 seconds to 4.11 seconds, and from 3.26 seconds to 12.52 seconds respectively. The increase in key length significantly increases the computational overhead, especially growing rapidly at 3072 bits. The computational overhead of the end user is related to trapdoor generation and ciphertext operations, and the impact of increasing key length is relatively small, increasing from 1.09 seconds at 1024 bits to 1.80 seconds, and being 3.10 seconds to 3.71 seconds and 8.48 seconds to 9.86 seconds at 2048 bits and 3072 bits respectively. The computational overhead of the cloud server mainly comes from encrypted data processing, increasing from 0.78 seconds at 1024 bits to 3.10 seconds, being 5.23 seconds to 22.49 seconds at 2048 bits, and 16.76 seconds to 76.46 seconds at 3072 bits. The key length has the most significant impact on the cloud server.

[0087] Figure 5 Shows the communication overhead under different data volumes and key lengths. The communication overhead of the data owner grows linearly as the data volume increases, and the overhead is significantly higher when the key length is longer. Under the 1024-bit key, the communication overhead increases from 61.08 KB to 244.22 KB; it is 121.27 KB to 485.06 KB at 2048 bits; and 181.43 KB to 725.96 KB at 3072 bits, mainly due to the encrypted data expansion effect. The communication overhead of the end user is relatively stable, fluctuating between 7.02 KB and 10.17 KB, and the key length has little impact on it, mainly for transmitting trapdoor information. The communication overhead of Cloud Server A and Cloud Server B increases significantly with the data volume and key length, increasing from 388.81 KB to 1557.41 KB at 1024 bits; being 716.75 KB to 2872.52 KB at 2048 bits; and 1041.18 KB to 4173.60 KB at 3072 bits, becoming the system performance bottleneck.

[0088] Figure 6 and Figure 7 Compares the computational and communication costs of the present invention and the Lulu Han scheme for the data owner entity. The results show that the present invention is significantly superior to the Lulu Han scheme in terms of computational overhead, especially being more prominent at high key lengths; in terms of communication overhead, the difference between the two is small at medium and high key lengths, and the present invention is slightly superior only at low key lengths. Therefore, in scenarios with large data volumes and high security requirements, the present invention can effectively reduce the computational burden and improve system performance.

[0089] Figure 8 It shows the differences in the computing overhead of the present invention and the Malik scheme for end - user computing. Regardless of the key length (1024, 2048, 3072 bits) or the data volume condition, the present invention shows significant advantages. Specifically, the computing overhead of the present invention increases linearly with the data volume but always remains at a low level. For example, with a 1024 - bit key, the overhead of the present invention increases from 1.09 seconds to 1.80 seconds, while the Malik scheme increases from 9.88 seconds to 27.44 seconds, with a maximum gap of up to 6.5 times; with a 2048 - bit key, the present invention increases from 3.10 seconds to 4.91 seconds, and the Malik scheme increases from 46.31 seconds to 216.15 seconds, with a gap of more than 95%; with a 3072 - bit key, the overhead of the present invention increases from 8.48 seconds to 9.86 seconds, while the Malik scheme increases from 142.72 seconds to 610.87 seconds, with a maximum gap of up to 98.4%. These results indicate that as the data volume increases, the computing overhead of the Malik scheme increases significantly, while the computing efficiency of the present invention remains stable, especially at high security levels, and the gap between the two is more significant.

[0090] Example 4

[0091] Based on the above - mentioned embodiments, as Figure 9 shown, the present invention proposes a privacy - protected keyword search system for intelligent high - speed rescue, including:

[0092] An encryption module, which is used to enable the data owner to receive intelligent high - speed information and the information of the user in need of rescue to generate an order data file and encrypt the keywords of the order data file to obtain ciphertext data and upload the ciphertext data to cloud server A; the keywords of the order data file include the longitude and latitude, mobile phone number, accident date, and number of accident victims of the user in need of rescue.

[0093] A terminal encryption module, which is used to enable the terminal user to obtain the search keywords in the distress message according to the distress message and construct a randomized ciphertext, obtain a trapdoor based on the randomized ciphertext and the search keywords, and send the randomized ciphertext and the trapdoor to cloud server A.

[0094] A secure inequality test module, which is used to enable cloud server A to process the ciphertext data and the trapdoor using a secure inequality test to obtain the matching result of the keywords in the ciphertext data and the trapdoor, and construct a random number file to perturb the ciphertext data with a successful match to obtain the perturbed ciphertext data, send the random number file to the terminal user, and send the perturbed ciphertext data to cloud server B.

[0095] A decryption module, which is used to enable cloud server B to decrypt the perturbed ciphertext data to obtain the plaintext search result and send the plaintext search result to the terminal user.

[0096] A matching module for enabling an end user to process the plaintext search results according to a random number file to obtain the final plaintext result corresponding to the search keyword.

[0097] It should be noted that a privacy - protected keyword search system for intelligent highway rescue provided by an embodiment of the present invention is to implement the above - mentioned privacy - protected keyword search method for intelligent highway rescue. Its functions can be specifically referred to the above - mentioned method embodiments and will not be elaborated here.

[0098] In summary, the present invention has significant advantages in terms of computational efficiency, communication cost, and privacy protection ability. By optimizing the probabilistic encryption of homomorphic encryption into deterministic encryption, it significantly reduces the computational overhead of the data owner and the amount of ciphertext data, thus effectively reducing the communication cost; designs a secure inequality test algorithm to implement keyword search in the encrypted state and enhances the privacy of search results by introducing random numbers; migrates the decryption operation to the cloud to relieve the computational and communication burdens of end users while ensuring the privacy of the system; adopts a dual - cloud server architecture to effectively prevent a single cloud server from inferring sensitive information by analyzing multiple search requests, greatly improving the anti - analysis ability of the system. Theoretical analysis and experimental verification show that the present invention exhibits excellent performance and practical application value in large - scale data processing scenarios.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

"1. A privacy-preserving keyword search method for intelligent highway rescue, characterized in that," Including: Step 1: The data owner receives the intelligent highway information and the information of the user in need of rescue, generates an order data file, encrypts the keywords of the order data file to obtain ciphertext data, and uploads the ciphertext data to cloud server A; the keywords of the order data file include the longitude and latitude, mobile phone number, accident date, and number of accident victims of the user in need of rescue. Step 2: The end user obtains the search keywords in the distress message according to the distress message, constructs a randomized ciphertext, obtains a trapdoor based on the randomized ciphertext and the search keywords, and sends the randomized ciphertext and the trapdoor to cloud server A. Step 3: Cloud server A processes the ciphertext data and the trapdoor using a secure inequality test to obtain the matching result of the keywords in the ciphertext data and the trapdoor, constructs a random number file to perturb the ciphertext data with successful matching, obtains the perturbed ciphertext data, sends the random number file to the end user, and sends the perturbed ciphertext data to cloud server B. Step 4: Cloud server B decrypts the perturbed ciphertext data to obtain the plaintext search result and sends the plaintext search result to the end user. Step 5: The end user processes the plaintext search result according to the random number file to obtain the final plaintext result corresponding to the search keywords.

2. The privacy - protected keyword search method for intelligent highway rescue according to claim 1, wherein, The encryption of the keywords of the order data file specifically includes: Generating two mutually independent random prime numbers q and q according to the preset key size KS, and then generating a key pair (pk, sk). Sending the key pair (pk, sk) to cloud server B, and distributing the public key pk to the data owner, cloud server A, and end user. Using the optimized Paillier algorithm to encrypt the keywords of the order data file one by one.

3. A privacy - protected keyword search method for intelligent highway rescue according to claim 1, characterized in that, The specific content of step 2 includes: The end user obtains the search keywords in the distress message according to the distress message. Multiplying the search keywords by -1 and then performing homomorphic encryption to obtain the ciphertext of the search keywords. Generating a random number and performing homomorphic encryption to obtain a randomized ciphertext, adding the randomized ciphertext and the ciphertext of the search keywords to obtain a trapdoor. The end user sends the randomized ciphertext and the trapdoor to cloud server A.

4. A privacy-preserving keyword search method for intelligent highway rescue according to claim 1, characterized in that, The processing of the ciphertext data and the trapdoor by cloud server A using a secure inequality test to obtain the matching result of the encrypted keywords and the search keywords in the ciphertext data specifically includes: Cloud server A reads the encrypted keywords in the ciphertext data, processes the trapdoor and the encrypted keywords using the homomorphic addition property of Paillier encryption to obtain a ciphertext value. Cloud server A sends the ciphertext value and the randomized ciphertext to cloud server B. Cloud server A and cloud server B cooperate to run a secure inequality test algorithm based on the homomorphic encryption addition property to obtain the matching result of the keywords and the trapdoor in the ciphertext data, and further obtain the matching result of the keywords and the search keywords in the ciphertext data.

5. A privacy - protected keyword search method for intelligent highway rescue according to claim 4, characterized in that, The cooperation between cloud server A and cloud server B to run a secure inequality test algorithm based on the homomorphic encryption addition property specifically includes: Server A calculates the encrypted difference using the homomorphic addition property of Paillierr. Server A sends the encrypted difference to cloud server B, and cloud server B decrypts the encrypted difference to obtain the plaintext difference. Determine whether the plaintext difference is 0. If the plaintext difference is 0, output an error in the matching result and send the result back to Cloud Server A. The ciphertext value and the randomized ciphertext match successfully, and thus the matching between the encrypted keyword in the ciphertext value and the search keyword in the trapdoor is successful. If the plaintext difference is not 0, output a correct matching result and return to execute Step 2.

6. The privacy - protected keyword search method for intelligent highway rescue according to claim 5, wherein, The encryption difference is represented by the following formula: Enc(Diff) = Add_Num·(E_R) -1 where Enc(Diff) is the encryption difference, Add_Num is the ciphertext value, and E_R is the randomized ciphertext.

7. A privacy - protected keyword search method for intelligent highway rescue according to claim 1, characterized in that, Constructing a random number file to perturb the ciphertext data to obtain the perturbed ciphertext data specifically includes: Cloud Server A generates a random number file with the same dimension size as the ciphertext data that matches the search keyword, homomorphically encrypts each random number, and uses the homomorphic addition property of Paillier encryption to process the encrypted random number file and the ciphertext data that matches the search keyword, and stores the processed result in the ciphertext data to obtain the perturbed ciphertext data.

8. A privacy - protected keyword search method for intelligent highway rescue according to claim 2, characterized in that, Cloud Server B decrypts the perturbed ciphertext data, specifically including: Cloud Server B decrypts the perturbed ciphertext data according to the private key sk in the key pair (pk, sk).

9. A privacy - protected keyword search method for intelligent highway rescue according to claim 1, characterized in that, Step 5 specifically includes: The end user processes the plaintext search result according to the random number file to obtain the final plaintext result corresponding to the search keyword; Subtract the value of each keyword in the plaintext search result from the corresponding random number in the random number file. If the result is not 0, store the matching items included in the row where the keyword in the plaintext search result is located in the result file to obtain the final plaintext result. If the result is 0, there is no result of the keyword in the order data file.

10. A privacy - protected keyword search system for intelligent highway rescue, characterized in that, It includes: An encryption module, which is used to enable the data owner to receive the intelligent highway information and the information of the user in need of rescue, generate an order data file, encrypt the keywords of the order data file to obtain ciphertext data, and upload the ciphertext data to Cloud Server A; The keywords of the order data file include the longitude and latitude, mobile phone number, accident date, and number of accident victims of the user in need of rescue; A terminal encryption module, which is used to enable the end user to obtain the search keyword in the distress message according to the distress message, construct a randomized ciphertext, obtain a trapdoor according to the randomized ciphertext and the search keyword, and send the randomized ciphertext and the trapdoor to Cloud Server A; A secure inequality test module, which is used to enable Cloud Server A to process the ciphertext data and the trapdoor using a secure inequality test to obtain the matching result between the keyword in the ciphertext data and the trapdoor, construct a random number file to perturb the ciphertext data with a successful match to obtain the perturbed ciphertext data, send the random number file to the end user, and send the perturbed ciphertext data to Cloud Server B; A decryption module, which is used to enable Cloud Server B to decrypt the perturbed ciphertext data to obtain the plaintext search result and send the plaintext search result to the end user; A matching module, which is used to enable the end user to process the plaintext search result according to the random number file to obtain the final plaintext result corresponding to the search keyword.

Images