Multidimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing

Abstract

The invention relates to a multidimensional task matching method based on a homomorphic encryption technology in spatial crowdsourcing. A system applied by the method comprises a service provider, a cloud server, a plurality of workers and at least one data requester, wherein the trust center and the service provider are in communication connection with the data requester and the workers; and the cloud server is in communication connection with the service provider and the trust center. According to the method, a multi-dimensional privacy matching protocol based on homomorphic encryption is applied, the homomorphic encryption technology and multiple protocol designs are organically combined, privacy matching of multi-dimensional attributes of tasks and workers in a spatial crowdsourcing system is achieved, the safety of privacy information of the tasks and the workers is greatly improved, privacy risks caused by data decryption are avoided, and the privacy matching efficiency is improved. And the accuracy and privacy of the matching process are ensured.

Description

Technical Field

[0001] The present invention relates to the field of spatial crowdsourcing and privacy protection, and in particular to a multidimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing. Background Art

[0002] Crowdsourcing has given rise to spatial crowdsourcing, which has been widely applied in a variety of fields, including air quality data collection, smart city development, and natural disaster and traffic reporting. It breaks down complex problems and outsources them to numerous non-expert participants, leveraging the collective wisdom of the crowd. The spatial crowdsourcing ecosystem comprises three entities: service providers, requesters, and workers. Service providers build the crowdsourcing platform and play a central role in the entire process. They receive tasks from requesters, allocate tasks based on location information provided by workers, and collect worker submissions to implement incentive mechanisms. Requesters submit tasks to the platform within a specific area, each requiring a specific location. Workers voluntarily join the system to take on tasks. They share their location information with the service provider, enabling the system to rationally allocate tasks. Upon completing a task, workers provide feedback to the service provider.

[0003] Spatial crowdsourcing faces numerous challenges in balancing matching accuracy and privacy. First, workers must disclose their real-time location information to complete tasks at specific locations, raising the risk of location privacy breaches and exposing sensitive information such as medical records and home addresses. Second, striking a balance between accuracy and privacy protection in multi-dimensional feature matching that considers factors such as spatial location, skills, and reward range remains extremely difficult. Existing technologies use location obfuscation and perturbation techniques to achieve location privacy. For example, k-anonymity allows users to hide their location within a group of k-1 other users. However, this implementation requires centralizing entities such as workers and requesters. In high-dimensional data, generalizing a large number of attributes to achieve k-anonymity requires over-ambiguity, resulting in overly ambiguous data and impacting matching accuracy. In real-world scenarios, the lack of trust between requesters and workers limits the application of this technology. Differential privacy utilizes a noise function, where users generate and report perturbed location information, achieving privacy protection. However, this also presents drawbacks. Location perturbations can interfere with the task allocation algorithm and impact the overall allocation of the system. Encryption technology relies on encryption methods to protect user locations and can achieve the highest level of privacy assurance. However, most of them are designed based on static data and face a trade-off between security and efficiency when supporting dynamic updates. Some methods sacrifice privacy protection strength or matching accuracy to achieve dynamic updates, resulting in a reduction in the overall allocation effect of the system.

[0004] Therefore, considering the frequent dynamic changes in task and worker information in spatial crowdsourcing systems, existing methods are difficult to adapt to this feature. There is a lack of efficient protocols for joint privacy matching of multi-dimensional attributes such as spatial location, skill characteristics, and reward ranges. When the system needs to support dynamic updates of data, existing methods need to make a difficult trade-off between the security of searchable encryption structures and operational efficiency. Summary of the Invention

[0005] To address the above issues, the present invention provides a multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing. The system used in this method includes a service provider, a cloud server, several workers, and at least one data requester. The trust center and the service provider are all in communication with the data requester and the worker; the cloud server is in communication with the service provider and the trust center.

[0006] The service provider pre-deploys a framework with a dynamic matching structure;

[0007] The cloud server includes a cloud server , cloud server The cloud server is pre-deployed with a privacy feature matching protocol, a privacy reward matching protocol, and a privacy comparison protocol; the method comprises the following steps:

[0008] Step 1: The trust center generates a homomorphic encryption public key, outputs a master key set, and initializes the system. The trust center sends the private key to the registered data requester and worker, and sends the private key to the cloud server. Send the homomorphic encryption public key to the cloud server Send the private key. The data requester and worker use the homomorphic encryption public key to encrypt the task information and private information respectively, and send them to the service provider.

[0009] Step 2: The service provider sends the dataset consisting of task information to the cloud server , cloud server 、 Run the privacy feature matching protocol and the privacy reward matching protocol respectively to find eligible tasks for each worker, and store the task matching results in the dynamic matching structure that is synchronously received when the task information is first received;

[0010] Step 3: The worker who needs the task uses the homomorphic encryption public key to generate the task request ciphertext and sends the token corresponding to the task request ciphertext to the service provider;

[0011] Step 4: After receiving the token, the service provider sends it to the cloud server , use the token to match and find tasks that meet the worker's conditions in the dynamic matching structure, and run the privacy comparison protocol to find the nearest task to complete the task assignment.

[0012] By constructing an encrypted matching structure and combining it with a privacy matching and comparison protocol, this paper achieves multi-dimensional attribute privacy matching and efficient task matching based on homomorphic encryption technology, with the following beneficial effects:

[0013] 1. This paper uses a multi-dimensional privacy matching protocol based on homomorphic encryption, organically combining homomorphic encryption technology with multiple protocol designs, to achieve privacy matching of multi-dimensional attributes of tasks and workers in the spatial crowdsourcing system, greatly improving the security of task and worker privacy information, avoiding the privacy risks brought by data decryption, and ensuring the accuracy and privacy of the matching process.

[0014] 2. This invention constructs a complete privacy protection and task matching system, innovatively designs a dynamic searchable encryption structure, and implements efficient task matching operations based on encrypted message exchange while ensuring data privacy, thus ensuring data security without affecting task matching efficiency.

[0015] 3. This invention possesses excellent practicality and adaptability. The proposed method can strictly safeguard user privacy. It demonstrates excellent efficiency in terms of matching time and structure construction time. It is suitable for real-world scenarios in spatial crowdsourcing systems where tasks and worker information frequently and dynamically change, and has significant application value for large-scale data processing and task matching. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a schematic diagram of the system model used in the present invention and the interaction between its parts. DETAILED DESCRIPTION

[0017] like Figure 1 As shown in the figure, the system model architecture used in this system covers multiple parties: service provider; cloud server includes two cloud servers, namely cloud server (hereinafter referred to as ), cloud server (hereinafter referred to as ); several workers and at least one data requester. The trust center and service provider are both in communication with the data requester and the workers; and the cloud server is in communication with the service provider and the trust center.

[0018] The functions of the trust center are: generating and distributing keys.

[0019] The function of a worker is to undertake and complete crowdsourcing tasks based on his or her own position.

[0020] The function of the data requester is to initiate a task request to the service provider.

[0021] The functions of the service provider are: receiving an encrypted task request of a data requester, returning a request result; receiving an encrypted privacy information of a worker, returning a task allocation result; outsourcing a task to a cloud server, and receiving a task allocation result returned by the cloud server. The service provider sends a dynamic matching structure framework to the cloud server when sending a task for the first time, which is used for subsequent filling of data.

[0022] The functions of the cloud server are: pre-deploying the privacy feature matching protocol, the privacy reward matching protocol, and the privacy comparison protocol (collectively referred to as a security protocol) invented by the application. With the powerful computing and storage resources of the cloud, the cloud server provides multi-dimensional task matching services for users and fills the matching results into the dynamic matching structure. The cloud server has ample storage space and powerful computing speed, is responsible for storing encrypted information, and uses the matching algorithm proposed in the application to comprehensively analyze the location distribution of the workers and the task query features of the requesters, and selects suitable task execution objects.

[0023] The working logic of the above system is: the data requester and the worker are pre-registered at the trust center, the trust center generates a private key and sends it to them; the trust center sends a public key to the cloud server , and sends a private key to the cloud server . The data requester and the worker send encrypted data requests and encrypted privacy information to the service provider, respectively. The service provider sends task information to the cloud server, and the cloud server executes the privacy feature matching protocol and the privacy reward matching protocol to generate a dynamic matching structure. When the worker needs a task, a matching token Token is generated and sent to the service provider. After receiving the matching token Token, the service provider sends it to the cloud server , uses Token to find a task that meets the worker's conditions, and runs the privacy comparison protocol to find the nearest task to the worker, and completes the task allocation.

[0024] The above method will be described in the following steps:

[0025] Step 1: The trust center generates a key, the system is initially initialized, the worker uploads encrypted privacy information, and the data requester uploads encrypted task information, which forms a data set .

[0026] In the key generation phase, the trust center completes the generation of the master key, which is denoted as , and constitutes a master key set. The trust center also generates a homomorphic encryption public key , and outputs the master key set. Then it enters the initialization phase, which mainly sets the global state of the system, including the permutation sequence, the data pool, and the dictionary.

[0027] Specifically, the permutation sequence Initialized to an empty set, the permutation sequence is used to confuse the data position; the data pool Initialize to an empty set; Initialized to an empty set. Workers can use homomorphic encryption public keys Encrypt and upload private information, including worker characteristics (such as skills, experience, etc.), acceptable incentives .in, Indicates the Workers. Data requesters can use homomorphic encryption public key The task information (location , Feature requirements ,excitation ) is encrypted and uploaded to the service provider to form a data set .in, Indicates the Data requester.

[0028] Step 2: The service provider sends the dataset And dynamically match the structure framework to the cloud server , cloud server and Run the privacy feature matching protocol and the privacy reward matching protocol separately to find eligible tasks for each worker and store the results in a dynamic matching structure.

[0029] Dataset Each element in . is a location identifier (i represents the index number, and the subscript id refers to the identifier). Information (subscript inf refers to information). Cloud server Holds homomorphic private keys .

[0030] Dynamic matching structure: In the process of data encryption and dynamic update, the current state and new data As input, after a series of processing operations, the updated encrypted database is finally output This process includes maintaining the state of existing encrypted data, processing and integrating newly added data, and implementing dynamic updates and encryption protection of the database. The newly added data is newly added task information and / or privacy information.

[0031] In the above expression, i is the element number in the dataset M, and t represents the stage of the current operation, such as the tth update; represents the replacement rule of the t-1th stage, n is the maximum index of all types of data in the current stage (t), is the maximum index of the data in the previous stage (t-1), The subscript value range is Greater than or equal to , An encrypted data structure representing the position of a task at stage t.

[0032] The following describes how to update the dictionary.

[0033] The specific process of dictionary update is: for each newly added encrypted data ( Add an index for newly encrypted data) and add its content to the global pool ( For data pool Extract the location identifier from the dynamically updated version of , used for subsequent location encryption . (tm stands for location mapping) Update the global encrypted data pool, ; Update the location identifier set, Generate an incremental arrangement of the new data, and then merge it with the original data. After the merge is completed, the new and old elements are arranged in order, and the associated index of the element keyword is updated.

[0034] Dictionary updates adopt a local update strategy, without regenerating all data. This has the technical effect of preventing attackers from inferring the position of old data from the new permutation, even without considering the historical permutation, thus enhancing confidentiality.

[0035] The incremental permutation method is: using pseudo-random permutation Perform pseudo-random permutation on the new data. . Represents the local arrangement of the newly added data, Represents incremental, indicating replacement of only newly added data. represents the key of G, represents the total index of stage t, ≥ .

[0036] The method to merge the arrangements is: .in Represents a function composite operation, that is, applying the previous stage permutation first, and then applying the new permutation. After the merge and permutation are completed, the new and old elements are sorted in order.

[0037] Each element in the dictionary has a keyword , used to index the matching task features or worker features. ( is a collection of all keywords in the system), update the keyword association index ,if , then calculate the parameters used to search for keyword tokens and calculate the timestamped verification tag , .in, is the keyword The associated index set at stage t, () is the key A pseudo-random function with parameters , Is the key A pseudorandom function with .

[0038] Apply a global permutation function Confusing data order: , calculate the maximum filling length , then fill evenly , last updated dictionary Only update keyword entries related to newly added data and verify with timestamp Bound tags To prevent the reuse of old tokens. All entries are padded to the current maximum length , to ensure search efficiency. It is a global permutation function used to confuse the data order. In the position information update phase, the position identifier Rearrange according to global arrangement For all In the case of will be ciphertext with the previously existing encrypted position Merge, form This method implements the location identifier Sort by global Encrypt after reordering, and only encrypt the newly added positions.

[0039] Privacy Feature Matching Protocol: Cloud Server Confirm whether worker characteristics match task characteristics under privacy protection.

[0040] Execution Participants:

[0041] Cloud Server : Holds encrypted task feature data , and the characteristic data of the worker's Token . stands for encryption, Representative Task No. Features, total Features, Representative The worker's Features, total Features.

[0042] Cloud Server : holds the key used for decryption .

[0043] Protocol Process: Cloud Server Computing using homomorphic encryption properties , the sum of squares of the worker's characteristics and the characteristic sum of squares of the task ,calculate , the matching threshold is , cloud server Requires verification: The symbol represents equivalence.

[0044] Cloud Server Generate verification items and select Random number , , calculate , , , , then and Send to , Decryption and , after decryption and Determined based on the following conditions:

[0045] Will and encryption, , Send back , calculate ,like , it means that the worker characteristics match the task characteristics.

[0046] Privacy Reward Matching Protocol: Used to confirm whether the rewards of workers and tasks match under privacy protection.

[0047] Execution Participants

[0048] Cloud Server :Hold the encrypted task reward , and the workers’ token rewards = . represents the minimum pay that worker j can accept, Indicates the highest pay that worker j can accept.

[0049] Cloud Server : holds the key used for decryption .

[0050] Protocol Process: Cloud Server Computing using homomorphic encryption properties , , , cloud server Select Satisfaction Random number 、 , and select a flag , takes value 1 or -1, calculate , then Send to cloud server Cloud Server Decrypt using the private key = ,according to Symbol generation .

[0051]

[0052] Will Return to the cloud server Cloud Server Received After that, calculate = * ,if =1, then it satisfies , the match is successful.

[0053] After the matching results are filled into the dynamic matching structure, the subsequent matching of workers and tasks does not need to recalculate the characteristics and whether the rewards match. The system only needs to compare and find the task closest to the worker and assign it. It can then find the nearest task that meets the worker's characteristics, greatly improving the efficiency of subsequent matching searches.

[0054] Step 3: When a worker needs a task, use the homomorphic encryption public key According to the self needs, the task request is encrypted into ciphertext, and the matching token is Token. The worker sends the token corresponding to the task request ciphertext to the service provider, which is used to search the encrypted dynamic encryption matching structure. The information in the task request ciphertext includes location , distance threshold , and acceptable task number .

[0055] The generation method of Token is: given input , generate wherein: . Since the label contains a timestamp , the token is time effective and prevents replay attacks. In this process, is used to protect the integrity of data, supports operations related to search keys.

[0056] Step 4: After receiving the token sent by the worker, the service provider sends it to the cloud server , uses the token to search the dynamic matching structure for tasks that meet the conditions contained in the worker's task request, and runs a privacy comparison protocol to find the nearest task to the worker, completing task allocation.

[0057] The privacy comparison protocol is used to compare the nearest distance between the worker and the task and the maximum number of tasks that the worker can accept under privacy protection.

[0058] Its execution participants include the cloud server , the cloud server .

[0059] The cloud server holds the encrypted dictionary , the location of the task , the location of the worker's task request Token , the allocation distance threshold of the task , and the number of tasks that the worker can accept , the current number of tasks

[0060] The cloud server holds the key for decryption

[0061] Protocol flow: assuming =( ), =( ), the elements in the parentheses are longitude and latitude, calculate , , calculate the distance difference ,like , the distance meets the conditions. Calculate the difference in the number of tasks ,like , the number of tasks has not reached the upper limit. Construct a joint polynomial , cloud server Select Satisfaction Random number , Construct the perturbation value, , , then Send to .

[0062] Cloud Server Decrypt using the private key = , = , calculate the disturbance difference .if >0, then set up , which means the worker is successfully matched with the current task, otherwise, set up ,Will Return to .

[0063] Cloud Server The specific algorithm for matching tasks that meet the worker's conditions using tokens in the dynamic matching structure is as follows: and , the output is The specific implementation process is as follows: Check Does it exist If it does not exist, an empty set is returned, indicating that the search keyword is invalid; if it exists, , (S is a set containing data indices) and then verify the label. Not equal to ( Stored in the dynamic matching structure ), an empty set is returned, indicating that the tag verification failed. If they are equal, Remove the filler items and get the collection ,Finally, extract data from the global pool and get And output. The algorithm can be directly The index from the data pool Get data from .

[0064] at last, The final matching results are obtained and returned to the service provider, and the query results are returned to the worker.

Claims

1. A multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing, characterized by: The system used by the method includes a service provider, a cloud server, several workers, and at least one data requester, wherein the trust center and the service provider are in communication with the data requester and the worker; Communication connection between cloud server, service provider and trust center; The service provider pre-deploys a framework with a dynamic matching structure; The cloud server includes a cloud server , cloud server The cloud server is pre-deployed with a privacy feature matching protocol, a privacy reward matching protocol, and a privacy comparison protocol; the method comprises the following steps: Step 1: The trust center generates a homomorphic encryption public key, outputs a master key set, and initializes the system. The trust center sends the private key to the registered data requester and worker, and sends the private key to the cloud server. Send homomorphic encryption public key to the cloud server Send the private key. The data requester and worker use the homomorphic encryption public key to encrypt the task information and private information respectively, and send them to the service provider. Step 2: The service provider sends the dataset consisting of task information to the cloud server , cloud server 、 Run the privacy feature matching protocol and the privacy reward matching protocol respectively to find eligible tasks for each worker, and store the task matching results in the dynamic matching structure that is synchronously received when the task information is first received; Step 3: The worker who needs the task uses the homomorphic encryption public key to generate the task request ciphertext and sends the token corresponding to the task request ciphertext to the service provider; Step 4: After receiving the token, the service provider sends it to the cloud server , use the token to match and find tasks that meet the worker's conditions in the dynamic matching structure, and run the privacy comparison protocol to find the nearest task to complete the task assignment.

2. The multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing according to claim 1 is characterized in that: In step 1, the master key set contains three keys, which are respectively ; The system initialization process will replace the series , data pool ,dictionary Initialized to an empty set; private information includes the worker's characteristics , acceptable incentives , task information includes location , Feature requirements ,excitation ;in, Indicates the A worker, Indicates the data requesters, and the superscript 0 indicates the initial stage.

3. The multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing according to claim 1 is characterized in that: The dynamic matching structure is: in the process of data encryption and dynamic update, the current state and new data As input, after processing, the final output is the updated encrypted database ; Where t represents the sequence number of the current operation phase, t-1 represents the sequence number of the t-1th operation phase, The encrypted data structure representing the task location, n is the maximum index of the newly added data in the current operation phase, is the maximum index of the data in stage t-1.

4. The multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing according to claim 3 is characterized in that: There is also a dictionary update process in step 1, specifically: for each newly added encrypted data , add its content to the data pool, extract the location identifier , used for subsequent location encryption ; Update the global encryption data pool, ; Update the location identifier set, Generate an incremental arrangement of the newly added data, then merge it with the original data in the order of old and new, and update the associated index of the element keyword; the subscript tm represents the position mapping, Index for newly added encrypted data.

5. The multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing according to claim 4 is characterized in that: The incremental permutation method is: using pseudo-random permutation Perform pseudo-random permutation on the new data. ; Represents the local arrangement of the newly added data, represents the increment, represents the total index of stage t, ≥ ; The method of merging and permuting is: ;in Represents function composition operation.

6. The multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing according to claim 4 is characterized in that: Privacy feature matching protocol for cloud servers Confirm whether the worker characteristics match the task characteristics under privacy protection, cloud server Holds encrypted task feature data , and the characteristic data of the worker's Token ; stands for encryption, Representative Task No. Features, total Features, Representative The worker's Features, total Features, Cloud Server Holds the key used for decryption ; Cloud Server Computing using homomorphic encryption properties , the sum of squares of the worker's characteristics and the characteristic sum of squares of the task ,calculate , the matching threshold is , cloud server Requires verification: Symbols represent equivalence; Cloud Server Generate verification items and select random numbers greater than 0 , calculate , , , , then and Send to cloud server , cloud server Decryption and , after decryption and Determined based on the following conditions: , Using homomorphic encryption public key and Then send it back to the cloud server ,calculate ,like , it means that the worker characteristics match the task characteristics.

7. The multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing according to claim 4 is characterized in that: The privacy reward matching protocol is used to confirm whether the rewards of workers and tasks match under privacy protection; cloud server Task rewards after holding encryption , and the incentives workers can accept = ; represents the minimum pay that worker j can accept, Indicates the highest remuneration that worker j can accept; Protocol Process: Cloud Server Computation using homomorphic encryption keys , , , cloud server Select a random number greater than 0 、 ,in , and select a flag , takes value 1 or -1, calculate , then Send to cloud server , cloud server Decrypt using the private key = ,according to Symbol generation , Cloud Server Will Return to the cloud server , cloud server Received After that, calculate = * ,if =1, then it satisfies , the match is successful.

8. The multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing according to claim 4 is characterized in that: In step 3, the worker uses the public key Encrypt the task request into ciphertext according to your own needs. The information in the ciphertext includes the location , distance threshold , the number of acceptable tasks .

9. The multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing according to claim 8, characterized in that: In step 3, the token is generated by: given input ,generate in: .

10. A multi-dimensional task matching method based on homomorphic encryption technology in spatial crowdsourcing according to claim 9, characterized in that: The privacy comparison protocol is used to compare the closest distance between workers and tasks and the maximum number of tasks that workers can accept under privacy protection. Its execution participants include cloud servers , cloud server ; Cloud Server Holds the encrypted dictionary , the location of the task , the position of the token , the task allocation distance threshold , the number of tasks that workers can accept , Current number of tasks ; Cloud Server Holds the key used for decryption ; Protocol Flow: Assumptions =( ), =( ), the elements in brackets are longitude and latitude values, cloud server calculate , , calculate the distance difference ,like , the distance meets the conditions; calculate the difference in the number of tasks ,like , the number of tasks has not reached the upper limit; construct a joint polynomial , cloud server Select Satisfaction Random number , Construct the perturbation value, , , then Send to cloud server ; Cloud Server Decrypt using the private key 、 , calculate the disturbance difference ,if >0, then the cloud server set up , indicating that the worker successfully matches the current task, otherwise the cloud server set up ,Will Return to the cloud server .

Images