Personalized matching method for vehicle-mounted network crowd sensing tasks

Abstract

The invention provides a personalized matching method for vehicle-mounted network crowd sensing tasks, and aims to protect privacy and meet personalized requirements of vehicles. A task publisher encrypts a task file by using a public key, generates a keyword index table at the same time, and uploads the keyword index table to a cloud server. The vehicle generates a retrieval trap door based on the interest words, and the cloud server matches tasks and returns the tasks to a task table. After the vehicle selects the sensing task according to the income, the number of the selected task is sent to the cloud server, the cloud server re-encrypts the selected task, and the vehicle decrypts and executes the sensing task. And after completion, the sensing data and the reputation report are uploaded to the block chain, and the smart contract updates the reputation value to ensure that the data is credible and cannot be tampered. And the cloud server sends an encryption aggregation result to the publisher, and the publisher decrypts the obtained data. According to the method, the enthusiasm of the vehicle participating in the sensing task is improved, and the security of task execution and data privacy are guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of Internet of Vehicles crowd intelligence perception, and relates to a privacy-protected personalized matching method for vehicle network crowd intelligence perception tasks. In particular, a perception vehicle can select crowdsourcing tasks based on the vehicle's own attributes and personal interests while protecting privacy, and improves the fairness of task matching and the security of task results. Background Art

[0002] At present, vehicle networks are regarded as innovative technologies for information collection due to their powerful transmission capabilities and excellent mobility, and have become an important platform for geographic crowdsourcing services. In terms of environmental data collection, the application of vehicle networks is particularly prominent. By carrying sensors and communication equipment, vehicle networks can collect various environmental data in real time, such as traffic conditions, air quality, climate information, etc., providing great convenience and support for urban planning, traffic management, and environmental monitoring. Its mobility enables it to cover a wide geographical area and achieve rapid data collection and transmission, bringing new possibilities and opportunities for the development of geographic crowdsourcing services. However, in dealing with the challenges of privacy and personalization in matching crowdsourcing tasks in vehicle networks, modern intelligent vehicle systems face the dual pressures of growing data exchange needs and user privacy protection. With the rapid development of vehicle networking technology, vehicle networks not only need to effectively match personalized crowdsourcing tasks to improve user experience, but also must ensure that users' privacy in multiple aspects such as location, identity, perception data, and reputation value is properly protected. This challenge requires innovative methods to balance the relationship between the personalized needs of task matching and the protection of user privacy.

[0003] In order to solve the privacy issues and personalized task selection problems of in-vehicle network crowdsourcing task matching, this study proposes a novel method that combines blockchain technology with personalized crowdsourcing task matching in in-vehicle networks, and uses searchable encryption, k-anonymity, homomorphic encryption, and proxy re-encryption to protect different aspects of privacy. Through the comprehensive use of these technologies, the location privacy, identity privacy, perceived data privacy, and reputation value privacy of in-vehicle network users can be effectively protected, providing users with a more secure and privacy-protected personalized service experience. Summary of the invention

[0004] The method takes the cloud server as the core, acts as the relay of the task publisher and the vehicle, generates the necessary public-private key pairs to ensure the security of the whole process of intelligent group perception, enhances the traceability of the operation and promotes the enthusiasm of the vehicle to participate in the perception task. The task publisher uses the public key to encrypt the task file and uploads the encrypted data and keyword index table to the cloud server. The vehicle interacts with the cloud server, uses the vehicle attribute keywords or interest words and the private key to generate a trapdoor, and submits it to the cloud server. The cloud server matches the keywords through the query function, returns the relevant task table to the vehicle, enables the vehicle to select the perception task more profitably, and sends the selected task number to the cloud server. The cloud server performs proxy re-encryption on the task ciphertext, and the vehicle decrypts the task with its own private key and obtains the data required for the task. The cloud server homomorphically aggregates the perception results after verification, and uploads the reputation report to the blockchain. After the vehicle consensus, the preset smart contract calculates and updates the vehicle reputation value. Finally, the cloud server aggregates the perception results and transmits them to the task publisher. The publisher uses the private key to decrypt the ciphertext and obtain the necessary data.

[0005] In order to achieve the above object, the technical solution of the present invention is:

[0006] A privacy-preserving personalized matching method for vehicle-mounted network crowd intelligence perception tasks is characterized by adopting the following steps:

[0007] Step 1: The data requester registers and assigns the sensing task, the location and reputation requirements of the sensing task, and the task keywords to the cloud server;

[0008] Step 2: Each sensing vehicle willing to sense data sends its location and reputation value to the cloud server through the roadside unit RSU to search for trapdoors;

[0009] Step 3: The cloud server determines whether its location and reputation value meet the location and reputation requirements of the sensing task, and then sorts them and issues a task list;

[0010] Step 4: The perception vehicle selects a task based on the perception task list and sends the task number to the cloud server;

[0011] Step 5: The cloud server agent re-encrypts and sends the task ciphertext;

[0012] Step 6: The vehicle decrypts the task ciphertext, obtains the perception data and encrypts the uploaded data;

[0013] Step 7: The cloud server verifies and aggregates the perception results and returns them to the data requester. At the same time, the cloud server uploads the reputation feedback report to the blockchain, which will notify the new block.

[0014] Step 8: The vehicle reviews the results on the blockchain and reports to the cloud platform. The vehicle performs the consensus process and the smart contract updates the reputation value of each sensed vehicle.

[0015] 2. The method according to claim 1 is characterized in that the step 1 specifically comprises the following steps: Step 1.1, the task publisher initializes two key pairs: text key and homomorphic key pairs and a searchable encryption key pair (pk, sk), where is an asymmetric key pair that supports proxy re-encryption, and is a homomorphic key pair generated by the homomorphic Paillier algorithm, that is Among them, n, g a , δ represent the product of large prime numbers p, q, random integers, and the least common multiple of p-1 and q-1, respectively. Then define the L function Calculate the module inverse element η = (L (g a δ mod n 2 )) -1 mod n; the task publisher will send the real identity Vid, registered address address and public key Send to the audit node on the blockchain service platform to apply for registration of information in the blockchain; in this process, the audit node verifies the identity information of the task publisher. After successful identity authentication, the contract interface node_register(pk,address,role) is called to write the task publisher's information into the blockchain, where address represents the user's unique identifier, role represents the user type, and the mapping address_pk[address] between the user address and its corresponding public key is updated. At the same time, the information is updated to the audit node variable audit_node; after the task publisher successfully registers, it can download the reputation opinion of the data owner;

[0016] Step 1.2, the task publisher uses Encrypt the outsourced storage task document set F to generate the ciphertext set CT, and use Generate valid range ciphertext C Δd The scope of the effective perception data;

[0017] Step 1.3: The task publisher calls the index building algorithm BuildIndex(W, params, pk) → (S n ,Inv,cnt1 W ), generate the encrypted forward index S corresponding to the task document n and the inverted index Inv, and the keyword ciphertext cnt1 encrypted with the public key pk W,W, params, and pk are respectively the keyword set, the system parameters, and the searchable encryption public key;

[0018] Step 1.4, the task publisher first generates a location matrix Md1 for the crowdsourcing task, and then uses the K-anonymity technology to divide each location into k location points;

[0019] Step 1.5: The task publisher sends the previous information into Send to the cloud server.

[0020] 3. The method according to claim 1, characterized in that the step 2 adopts the following steps:

[0021] Step 2.1, each sensor vehicle is registered when entering the Internet of Vehicles, and the vehicle's real identity information vid and reputation value are stored on the blockchain, where vid is the real identity of the sensor vehicle;

[0022] Step 2.2: The vehicle constructs the Trapdoor algorithm based on the interest tag (params, sk, Q, cnt Q ) Generate a search trap, where params, sk, Q, cnt Q They are system parameters, which can search for encryption private keys, query keyword sets, and counters for each keyword;

[0023] Step 2.3, each vehicle that wants to participate in the task generates a random position matrix M1′, uses the K-anonymity technique to divide each position into k position points, and replaces each point with 1;

[0024] Step 2.4, the vehicle sends information I s1 ={vid, Mv1, T} is uploaded to the cloud server, where vid, Mv1, and T are identity information, vehicle location information, and retrieval trap.

[0025] 4. The method according to claim 1, characterized in that the step 3 adopts the following steps:

[0026] Step 3.1, the cloud server calls the matching search algorithm Search (S n , lnv, T) adds the matching encrypted task document to the retrieval result set CT m Among them, S n ,1nv,T are the encrypted forward index, inverted index, and retrieval trapdoor;

[0027] Step 3.2, the cloud server checks whether the vehicle position meets the requirements and calculates the label Mr = Md1·Mv1, where Md1 and Mv1 are the position matrix generated by the crowdsourcing task and the vehicle position information; if Mr = 0, the vehicle does not meet the position requirements of any task; if Mr is not equal to 0, the vehicle meets the position requirements of a task; the cloud server removes the tasks that do not meet the position requirements from the task table;

[0028] Step 3.3, the cloud server queries the blockchain through the smart contract whether the vehicle meets the task reputation requirements, and removes the tasks whose vehicle reputation does not meet the reputation requirements from the task table;

[0029] Step 3.4, using the vector space model to calculate the relevance ranking between the document and the query, and generating an encrypted document list that meets the retrieval requirements;

[0030] In step 3.5, a list of perception tasks that meet the credibility requirements of the location requirements is sent to the perception vehicle.

[0031] 5. The method according to claim 1 is characterized in that in step 5, the cloud server performs proxy re-encryption on the ciphertext Cm of the sensing task content m to obtain Cm′; the cloud server sends I to the vehicle via the roadside unit RSU c1 ={Ack,Cm′,Pk dh}, where the confirmation mark g is a randomly selected integer, p is a large prime number, α i Is a random number.

[0032] 6. The method according to claim 1, characterized in that in step 6, receiving I c1 After that, the vehicle decrypts Cm′ to obtain the plaintext m of the perception task, executes the task and encrypts and uploads the perception data, and generates the ciphertext C after homomorphic encryption. d =g a d* ε d n mod n 2 , ε d is a random number, n is the homomorphic encryption parameter mentioned above, and then the roadside unit RSU sends I to the cloud server s2 ={C d ,A vid}, where A vid =(Ack,vid),C d ,Ack,vid represent the ciphertext of the sensing data, the confirmation character, and the vehicle identification, respectively.

[0033] 7. The method according to claim 1, characterized in that the step 7 adopts the following steps:

[0034] Step 7.1, the cloud server first verifies A vid Is it effective? If it is effective, use the homomorphic computing of the privacy protection comparison protocol to perceive whether the ciphertext is within a reasonable range, that is, calculate d′=dd under the ciphertext 0 The ciphertext value of : where ε d , ε d0 is a random number, d 0 represents the benchmark, d represents the vehicle's perception value, and d′ represents the amount by which the vehicle's perception value exceeds the benchmark. Finally, the cloud server compares C d ' and C Δd , if the result is d'≤Δd, then the sensed data is valid;

[0035] Step 7.2: The cloud server uses the data aggregation algorithm to perform homomorphic aggregation on the data to obtain aggregated data I ca =Agree returns it to the task publisher, and the task publisher can decrypt the perception data with his own private key; Step 7.3, based on the data verification results, the cloud server generates a reputation feedback report F for each vehicle and sends the corresponding vehicle reputation report I c2 = {vid, F} to the blockchain, where the reputation feedback report F is 1 if the perceived ciphertext is within a reasonable range, otherwise it is 0, i.e. The cloud platform transmits the entire data set of the slots to the blockchain, notifying the generation of new blocks and nodes.

[0036] 8. The method according to claim 1, characterized in that said step 8 adopts the following steps:

[0037] Step 8.1, the vehicle executes the consensus algorithm to verify the validity of the reputation opinion on the blockchain;

[0038] Step 8.2, check whether the number of reputation feedback reports reaches the group size G n Or whether the time reaches the time threshold Δt; if it is satisfied, the smart contract will calculate a new reputation value according to the preset algorithm, that is, calculate a new reputation value t for the vehicle; if it is not satisfied, the reputation value of the vehicle will not be updated this time;

[0039] Step 8.3, the reputation value calculation algorithm dynamically updates the reputation value according to the number and quality of the perception tasks in which the perception vehicle participates, as well as the historical reputation value. It considers three factors, among which the specific method for calculating the reputation value t is as follows; where α Rn represents the positive reputation parameter of the nth round, β Rnrepresents the negative reputation parameter of the nth round, t is the reputation value, Rf represents the forgetting factor, ranging from 0 to 1, Rg represents the group weight, Rn represents the nth round, Gn reputation feedback reports constitute a group, Gn represents the size of the group, Tn represents the number of perception tasks, and there should be Gn <Tn,s Rn represents the number of positive feedback in the nth round, u Rn Indicates the number of negative feedback in the nth round;

[0040] Step 8.4, after the new reputation value is calculated, the smart contract will update the reputation value of each sensing vehicle on the blockchain.

[0041] The beneficial effects of the present invention are:

[0042] (1) A method for matching personalized crowdsourcing tasks for intelligent group perception is provided to address the challenges faced by vehicle networks in terms of privacy and personalization. The personalized matching of intelligent group perception tasks in vehicle networks is realized. Task publishers can limit the credibility level of vehicles participating in perception tasks, and vehicles can obtain perception tasks that are more in line with their own needs and capabilities, thereby improving the enthusiasm and satisfaction of users in participating in perception tasks.

[0043] (2) This method uses a variety of privacy protection technologies, such as searchable encryption, k-anonymity, proxy re-encryption, and homomorphic encryption to effectively protect the user's location, identity, and perception data privacy, and uses proxy re-encryption to effectively protect task privacy, providing in-vehicle network users with a safer, more privacy-protected personalized service experience. (3) By establishing a fair and just blockchain reputation mechanism, without the need for a third party to serve as a reputation center, it incentivizes vehicles to effectively complete perception tasks, thereby improving the efficiency and quality of perception tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is the work flow chart of the present invention.

[0045] Figure 2 This is the multi-keyword dynamic searchable encryption system model of the present invention. DETAILED DESCRIPTION

[0046] The present invention will be described in detail below in conjunction with the various embodiments shown in the accompanying drawings. However, these embodiments do not limit the present invention, and any structural, methodological, or functional changes made by a person skilled in the art based on these embodiments are all within the scope of protection of the present invention.

[0047] like Figure 1As shown, a privacy-preserving personalized matching method for in-vehicle network crowd perception tasks is presented, which includes: the task publisher encrypts data and data keywords and uploads them to the cloud server, the vehicle generates a search trap and interacts with the cloud server, the cloud server queries the keyword matching tasks, the cloud server performs proxy re-encryption on the task ciphertext, the vehicle decrypts the task ciphertext with its own private key and obtains the perception data encrypted and uploaded to the cloud server, the cloud performs homomorphic aggregation on the data, uses a consensus mechanism to generate blocks, the smart contract updates the vehicle's reputation, and the task publisher obtains the perception result.

[0048] A privacy-preserving personalized matching method for vehicle-mounted network crowd intelligence perception tasks is characterized by adopting the following steps:

[0049] Step 1: The data requester registers and assigns the sensing task, the location and reputation requirements of the sensing task, and the task keywords to the cloud server;

[0050] Step 2: Each sensing vehicle willing to sense data sends its location and reputation value to the cloud server through the roadside unit RSU to search for trapdoors;

[0051] Step 3: The cloud server determines whether its location and reputation value meet the location and reputation requirements of the sensing task, and then sorts them and issues a task list;

[0052] Step 4: The perception vehicle selects a task based on the perception task list and sends the task number to the cloud server;

[0053] Step 5: The cloud server agent re-encrypts and sends the task ciphertext;

[0054] Step 6: The vehicle decrypts the task ciphertext, obtains the perception data and encrypts the uploaded data;

[0055] Step 7: The cloud server verifies and aggregates the perception results and returns them to the data requester. At the same time, the cloud server uploads the reputation feedback report to the blockchain, which will notify the new block.

[0056] Step 8: The vehicle reviews the results on the blockchain and reports to the cloud platform. The vehicle performs the consensus process and the smart contract updates the reputation value of each sensed vehicle.

[0057] 2. The method according to claim 1 is characterized in that the step 1 specifically comprises the following steps: Step 1.1, the task publisher initializes two key pairs: text key and homomorphic key pairs and a searchable encryption key pair (pk, sk), where is an asymmetric key pair that supports proxy re-encryption, and is a homomorphic key pair generated by the homomorphic Paillier algorithm, that is Among them, n, g a , δ represent the product of large prime numbers p, q, random integers, and the least common multiple of p-1 and q-1, respectively. Then define the L function Calculate the module inverse element η = (L (g a δ mod n 2 )) -1 mod n; the task publisher will send the real identity Vid, registered address address and public key Send to the audit node on the blockchain service platform to apply for registration of information in the blockchain; in this process, the audit node verifies the identity information of the task publisher. After successful identity authentication, the contract interface node_register(pk,address,role) is called to write the task publisher's information into the blockchain, where address represents the user's unique identifier, role represents the user type, and the mapping address_pk[address] between the user address and its corresponding public key is updated. At the same time, the information is updated to the audit node variable audit_node; after the task publisher successfully registers, it can download the reputation opinion of the data owner;

[0058] Step 1.2, the task publisher uses Encrypt the outsourced storage task document set F to generate the ciphertext set CT, and use Generate valid range ciphertext C Δd The scope of the effective perception data;

[0059] Step 1.3: The task publisher calls the index building algorithm BuildIndex(W, params, pk) → (S n ,Inv,cnt1 W ), generate the encrypted forward index S corresponding to the task document n and the inverted index Inv, and the keyword ciphertext cnt1 encrypted with the public key pk W ,W, params, and pk are respectively the keyword set, the system parameters, and the searchable encryption public key;

[0060] Step 1.4, the task publisher first generates a location matrix Md1 for the crowdsourcing task, and then uses the K-anonymity technology to divide each location into k location points;

[0061] Step 1.5: The task publisher sends the previous information into Send to the cloud server.

[0062] 3. The method according to claim 1, characterized in that the step 2 adopts the following steps:

[0063] Step 2.1, each sensor vehicle is registered when entering the Internet of Vehicles, and the vehicle's real identity information vid and reputation value are stored on the blockchain, where vid is the real identity of the sensor vehicle;

[0064] Step 2.2: The vehicle constructs the Trapdoor algorithm based on the interest tag (params, sk, Q, cnt Q ) Generate a search trap, where params, sk, Q, cnt Q They are system parameters, which can search for encryption private keys, query keyword sets, and counters for each keyword;

[0065] Step 2.3, each vehicle that wants to participate in the task generates a random position matrix M1′, uses the K-anonymity technique to divide each position into k position points, and replaces each point with 1;

[0066] Step 2.4, the vehicle sends information I s1 ={vid, Mv1, T} is uploaded to the cloud server, where vid, Mv1, and T are identity information, vehicle location information, and retrieval trap.

[0067] 4. The method according to claim 1, characterized in that the step 3 adopts the following steps:

[0068] Step 3.1, the cloud server calls the matching search algorithm Search (S n , lnv, T) adds the matching encrypted task document to the retrieval result set CT m Among them, S n ,1nv,T are the encrypted forward index, inverted index, and retrieval trapdoor;

[0069] Step 3.2, the cloud server checks whether the vehicle position meets the requirements and calculates the label Mr = Md1·Mv1, where Md1 and Mv1 are the position matrix generated by the crowdsourcing task and the vehicle position information; if Mr = 0, the vehicle does not meet the position requirements of any task; if Mr is not equal to 0, the vehicle meets the position requirements of a task; the cloud server removes the tasks that do not meet the position requirements from the task table;

[0070] Step 3.3, the cloud server queries the blockchain through the smart contract whether the vehicle meets the task reputation requirements, and removes the tasks whose vehicle reputation does not meet the reputation requirements from the task table;

[0071] Step 3.4, using the vector space model to calculate the relevance ranking between the document and the query, and generating an encrypted document list that meets the retrieval requirements;

[0072] In step 3.5, a list of perception tasks that meet the credibility requirements of the location requirements is sent to the perception vehicle.

[0073] 5. The method according to claim 1 is characterized in that in step 5, the cloud server performs proxy re-encryption on the ciphertext Cm of the sensing task content m to obtain Cm′; the cloud server sends I to the vehicle via the roadside unit RSU c1 ={Ack,Cm′,Pk dh}, where the confirmation mark g is a randomly selected integer, p is a large prime number, α i Is a random number.

[0074] 6. The method according to claim 1, characterized in that in step 6, receiving I c1 After that, the vehicle decrypts Cm′ to obtain the plaintext m of the perception task, executes the task and encrypts and uploads the perception data, and generates the ciphertext C after homomorphic encryption. d =g a d* ε d n mod n 2 , ε d is a random number, n is the homomorphic encryption parameter mentioned above, and then the roadside unit RSU sends I to the cloud server s2 ={C d ,A vid}, where A vid =(Ack,vid),C d ,Ack,vid represent the ciphertext of the sensing data, the confirmation character, and the vehicle identification, respectively.

[0075] 7. The method according to claim 1, characterized in that the step 7 adopts the following steps:

[0076] Step 7.1, the cloud server first verifies A vid Is it effective? If it is effective, use the homomorphic computing of the privacy protection comparison protocol to perceive whether the ciphertext is within a reasonable range, that is, calculate d′=dd under the ciphertext 0 The ciphertext value of : where ε d , is a random number, d 0 represents the benchmark, d represents the vehicle's perception value, and d′ represents the amount by which the vehicle's perception value exceeds the benchmark. Finally, the cloud server compares C d ' and C Δd , if the result is d'≤Δd, then the sensed data is valid;

[0077] Step 7.2: The cloud server uses the data aggregation algorithm to perform homomorphic aggregation on the data to obtain aggregated data I ca =Agree returns it to the task publisher, and the task publisher can decrypt the perception data with his own private key; Step 7.3, based on the data verification results, the cloud server generates a reputation feedback report F for each vehicle and sends the corresponding vehicle reputation report I c2 = {vid, F} to the blockchain, where the reputation feedback report F is 1 if the perceived ciphertext is within a reasonable range, otherwise it is 0, i.e. The cloud platform transmits the entire data set of the slots to the blockchain, notifying the generation of new blocks and nodes.

[0078] 8. The method according to claim 1, characterized in that said step 8 adopts the following steps:

[0079] Step 8.1, the vehicle executes the consensus algorithm to verify the validity of the reputation opinion on the blockchain;

[0080] Step 8.2, check whether the number of reputation feedback reports reaches the group size G n Or whether the time reaches the time threshold Δt; if it is satisfied, the smart contract will calculate a new reputation value according to the preset algorithm, that is, calculate a new reputation value t for the vehicle; if it is not satisfied, the reputation value of the vehicle will not be updated this time;

[0081] Step 8.3, the reputation value calculation algorithm dynamically updates the reputation value according to the number and quality of the perception tasks in which the perception vehicle participates, as well as the historical reputation value. It considers three factors, among which the specific method for calculating the reputation value t is as follows; where α Rn represents the positive reputation parameter of the nth round, β Rn represents the negative reputation parameter of the nth round, t is the reputation value, Rf represents the forgetting factor, ranging from 0 to 1, Rg represents the group weight, Rn represents the nth round, Gn reputation feedback reports constitute a group, Gn represents the size of the group, Tn represents the number of perception tasks, and there should be Gn <Tn,s Rn represents the number of positive feedback in the nth round, u Rn Indicates the number of negative feedback in the nth round;

[0082] Step 8.4, after the new reputation value is calculated, the smart contract will update the reputation value of each sensing vehicle on the blockchain.

Claims

1. A privacy-preserving personalized matching method for in-vehicle network crowd intelligence perception tasks, characterized in that: Use the following steps, Step 1: The data requester registers and assigns the sensing task, the location and reputation requirements of the sensing task, and the task keywords to the cloud server; Step 2: Each sensing vehicle willing to sense data sends its location and reputation value to the cloud server through the roadside unit RSU to search for trapdoors; Step 3: The cloud server determines whether its location and reputation value meet the location and reputation requirements of the sensing task, and then sorts them and issues a task list; Step 4: The perception vehicle selects a task based on the perception task list and sends the task number to the cloud server; Step 5: The cloud server agent re-encrypts and sends the task ciphertext; Step 6: The vehicle decrypts the task ciphertext, obtains the perception data and encrypts the uploaded data; Step 7: The cloud server verifies and aggregates the perception results and returns them to the data requester. At the same time, the cloud server uploads the reputation feedback report to the blockchain, which will notify the new block. Step 8: The vehicle reviews the results on the blockchain and reports to the cloud platform. The vehicle performs the consensus process and the smart contract updates the reputation value of each sensed vehicle.

2. The method according to claim 1, characterized in that The step 1 specifically comprises the following steps: Step 1.1: The task publisher initializes two key pairs: text key and homomorphic key pairs and a searchable encryption key pair (pk, sk), where is an asymmetric key pair that supports proxy re-encryption, and is a homomorphic key pair generated by the homomorphic Paillier algorithm, that is Among them, n, g a , δ represent the product of large prime numbers p, q, random integers, and the least common multiple of p-1 and q-1, respectively. Then define the L function Calculate the module inverse element η = (L (g a δ mod n 2 )) -1 mod n; the task publisher will send the real identity Vid, registered address address and public key ( pk) is sent to the audit node on the blockchain service platform to apply for registration of the information in the blockchain; in this process, the audit node verifies the identity information of the task publisher. After successful identity authentication, the contract interface node_register(pk,address,role) is called to write the task publisher's information into the blockchain, where address represents the user's unique identifier, role represents the user type, and the mapping address_pk[address] between the user address and its corresponding public key is updated. At the same time, the information is updated to the audit node variable audit_node; after the task publisher successfully registers, it can download the reputation opinion of the data owner; Step 1.2, the task publisher uses Encrypt the outsourced storage task document set F to generate the ciphertext set CT, and use Generate valid range ciphertext C Δd The scope of the effective perception data; Step 1.3: The task publisher calls the index building algorithm BuildIndex(W, params, pk) → (S n ,Inv,cnt1 W ), generate the encrypted forward index S corresponding to the task document n and the inverted index Inv, and the keyword ciphertext cnt1 encrypted with the public key pk W ,W, params, and pk are respectively the keyword set, the system parameters, and the searchable encryption public key; Step 1.4, the task publisher first generates a location matrix Md1 for the crowdsourcing task, and then uses the K-anonymity technology to divide each location into k location points; Step 1.5: The task publisher sends the previous information into Send to the cloud server.

3. The method according to claim 1, characterized in that The step 2 adopts the following steps: Step 2.1, each sensor vehicle is registered when entering the Internet of Vehicles, and the vehicle's real identity information vid and reputation value are stored on the blockchain, where vid is the real identity of the sensor vehicle; Step 2.2: The vehicle constructs the Trapdoor algorithm based on the interest tag (params, sk, Q, cnt Q ) Generate a search trap, where params, sk, Q, cnt Q They are system parameters, which can search for encryption private keys, query keyword sets, and counters for each keyword; Step 2.3, each vehicle that wants to participate in the task generates a random position matrix M1′, uses the K-anonymity technique to divide each position into k position points, and replaces each point with 1; Step 2.4, the vehicle sends information I s1 ={vid, Mv1, T} is uploaded to the cloud server, where vid, Mv1, and T are identity information, vehicle location information, and retrieval trap.

4. The method according to claim 1, characterized in that: The step 3 adopts the following steps: Step 3.1, the cloud server calls the matching search algorithm Search (S n , lnv, T) adds the matching encrypted task document to the retrieval result set CT m Among them, S n ,1nv,T are the encrypted forward index, inverted index, and retrieval trapdoor; Step 3.2, the cloud server checks whether the vehicle position meets the requirements and calculates the label Mr = Md1·Mv1, where Md1 and Mv1 are the position matrix generated by the crowdsourcing task and the vehicle position information; if Mr = 0, the vehicle does not meet the position requirements of any task; if Mr is not equal to 0, the vehicle meets the position requirements of a task; the cloud server removes the tasks that do not meet the position requirements from the task table; Step 3.3, the cloud server queries the blockchain through the smart contract whether the vehicle meets the task reputation requirements, and removes the tasks whose vehicle reputation does not meet the reputation requirements from the task table; Step 3.4, using the vector space model to calculate the relevance ranking between the document and the query, and generating an encrypted document list that meets the retrieval requirements; In step 3.5, a list of perception tasks that meet the credibility requirements of the location requirements is sent to the perception vehicle.

5. The method according to claim 1, characterized in that In step 5, the cloud server performs proxy re-encryption on the ciphertext Cm of the sensing task content m to obtain Cm′; the cloud server sends I to the vehicle via the roadside unit RSU. c1 ={Ack,Cm′,Pk dh }, where the confirmation mark g is a randomly selected integer, p is a large prime number, α i Is a random number.

6. The method according to claim 1, characterized in that In step 6, I c1 After that, the vehicle decrypts Cm′ to obtain the plaintext m of the perception task, executes the task and encrypts and uploads the perception data, and generates the ciphertext C after homomorphic encryption. d =g a d* ε d n mod n 2 , ε d is a random number, n is the homomorphic encryption parameter mentioned above, and then the roadside unit RSU sends I to the cloud server s2 ={C d ,A vid }, where A vid =(Ack,vid),C d ,Ack,vid represent the ciphertext of the sensing data, the confirmation character, and the vehicle identification, respectively.

7. The method according to claim 1, characterized in that The step 7 adopts the following steps: Step 7.1, the cloud server first verifies A vid Is it effective? If it is effective, use the homomorphic computing of the privacy protection comparison protocol to perceive whether the ciphertext is within a reasonable range, that is, calculate the ciphertext value of d′=d-d0 under the ciphertext: where ε d , is a random number, d0 represents the benchmark, d represents the vehicle's perception value, and d′ represents the amount by which the vehicle's perception value exceeds the benchmark. Finally, the cloud server compares C d ' and C Δd , if the result is d'≤Δd, then the sensed data is valid; Step 7.2: The cloud server uses the data aggregation algorithm to perform homomorphic aggregation on the data to obtain aggregated data I ca =Agree returns it to the task publisher, and the task publisher can decrypt the perception data with his own private key; Step 7.3: Based on the data verification results, the cloud server generates a reputation feedback report F for each vehicle and sends the corresponding vehicle reputation report I c2 = {vid, F} to the blockchain, where the reputation feedback report F is 1 if the perceived ciphertext is within a reasonable range, otherwise it is 0, i.e. The cloud platform transmits the entire data set of the slots to the blockchain, notifying the generation of new blocks and nodes.

8. The method according to claim 1, characterized in that The step 8 adopts the following steps: Step 8.1, the vehicle executes the consensus algorithm to verify the validity of the reputation opinion on the blockchain; Step 8.2, check whether the number of reputation feedback reports reaches the group size G n Or whether the time reaches the time threshold Δt; if it is satisfied, the smart contract will calculate a new reputation value according to the preset algorithm, that is, calculate a new reputation value t for the vehicle; if it is not satisfied, the reputation value of the vehicle will not be updated this time; Step 8.3, the reputation value calculation algorithm dynamically updates the reputation value according to the number and quality of the perception tasks in which the perception vehicle participates, as well as the historical reputation value. It considers three factors, among which the specific method for calculating the reputation value t is as follows; where α Rn represents the positive reputation parameter of the nth round, β Rn represents the negative reputation parameter of the nth round, t is the reputation value, Rf represents the forgetting factor, ranging from 0 to 1, Rg represents the group weight, Rn represents the nth round, Gn reputation feedback reports constitute a group, Gn represents the size of the group, Tn represents the number of perception tasks, and there should be Gn <Tn,s Rn represents the number of positive feedback in the nth round, u Rn Indicates the number of negative feedback in the nth round; Step 8.4, after the new reputation value is calculated, the smart contract will update the reputation value of each sensing vehicle on the blockchain.

Images