A sample-based connectable query method for privacy-preserving spatial datasets

Abstract

This invention belongs to the field of data retrieval and discloses a sample-based method for privacy-preserving joinable queries on spatial datasets. In the data preprocessing stage, the spatial dataset in the data warehouse is first processed by vectorizing it and encrypting it with a key. The encrypted spatial dataset is then stored in a hierarchical vector index and sent to a cloud server. In the data query stage, a given sample spatial dataset is vectorized and encrypted to form a search trapdoor, which is sent to the cloud server as a search instruction. The cloud server uses the trapdoor to perform a privacy-preserving joinable query operation on the encrypted data index, achieving a joinable query search while ensuring the privacy of the spatial dataset data, and returning the top k datasets as the final result set. This method utilizes the latest indexes and new query methods, resulting in high search efficiency.

Description

Technical Field

[0001] This invention belongs to the field of data retrieval, specifically relating to a sample-based method for connecting and querying privacy-preserving spatial datasets. Background Art

[0002] We are currently in an era of data explosion, and with the development of technology, more and more data is being collected to serve society. As a crucial component of real-world datasets, spatial datasets are finding increasingly widespread applications. Therefore, spatial dataset search technology, which meets user needs, has become an important research direction, and research into spatial dataset search methods is urgently needed. Sample-based spatial dataset joinable queries are a novel query method. Unlike density-based spatial dataset queries, this method finds spatially connected datasets with high spatial dissimilarity from a data warehouse based on a given sample dataset. This technology has significant application value in fields such as geographic systems, transportation and rail systems, urban management systems, and municipal planning.

[0003] Connectivity queries for spatial datasets require similarity calculations between datasets. Existing literature has proposed numerous methods for this purpose. For example, in connectivity searches for tabular data, the difference calculation method compares the number of different columns while ensuring one or more columns are connected to determine the difference. In spatial data searches, similarity calculations use normalized Jaccard similarity, cosine similarity, or distance similarity as the overall similarity for comparison. In trajectory data searches, density is often used as a similarity metric. In dataset search research, various methods exist: grid overlap (GBO) filtering, which performs coarse-grained filtering based on the overlap of grid indices; intersecting distance (IA) filtering, which performs coarse-grained filtering based on the overlap or intersection relationship of the MBRs between two datasets; and Hausdorff distance, which determines the maximum and minimum distances between two datasets for finer-grained filtering. Therefore, spatial dataset connectable query, as a new query method, can achieve efficient spatial dataset connectable query by drawing analogies to new similarity judgment methods from the search of other datasets and by leveraging filtering and pruning strategies in dataset search.

[0004] However, performing connectable queries on spatial datasets can lead to the leakage of spatial data. Privacy-preserving connectable queries on spatial datasets can retrieve the datasets that users need while ensuring that the privacy of the datasets is not compromised. Summary of the Invention

[0005] To address the shortcomings of existing technologies and resolve privacy concerns in the process of querying connectable spatial datasets, this invention provides a novel and efficient example-based method for privacy-preserving connectable queries of spatial datasets. This method vectorizes the spatial dataset, encrypts it, and then converts it into a ciphertext vector form. This ciphertext vector enables privacy-preserving connectable queries of spatial datasets without revealing the plaintext data. A method for calculating the spatial dissimilarity of spatial datasets under ciphertext is proposed. Furthermore, a vector hierarchical index is used to improve search efficiency.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0007] This invention is a sample-based method for connecting and querying privacy-preserving spatial datasets, applicable to an application scenario consisting of a data owner, a cloud server, and a querying user. The query method includes a data preprocessing stage and a data query stage.

[0008] The first stage, the data preprocessing stage, specifically includes the following steps:

[0009] Step 1-1: Based on the set connectability parameter threshold ξ, perform the following steps on the spatial data warehouse. Perform global mesh generation and set the connectability parameter threshold. The spatial data warehouse is divided into grids of equal size θ×θ, forming a grid set. Each grid cell is assigned a unique cell ID (e.g., ID0 for g0), ranging from [0, θ]. 2 -1];

[0010] Steps 1-2: The data owner generates a key K = {s, M1, M2}, where s is a randomly generated θ. 2 2D vectors, M1, M2 are (θ) 2 )×(θ 2 A 1 / 2 invertible matrix, where the key K is shared by the data owner and the querying user;

[0011] Steps 1-3: For spatial data warehouses Each spatial dataset D in i , spatial dataset D i Each spatial data point in the dataset is mapped to a grid, and each of the spatial datasets D i Form a grid data set and grid data set Mapped onto a Bloom filter, based on each grid data set Form a vector BF i and for vector BF iAfter vectorization, encryption is performed to form an encrypted vector. Let the decision vector The decision vector v is then vectorized and encrypted to form an encrypted decision vector. Finally, the encryption vector and encryption decision vector Store in a high-dimensional vector index;

[0012] The second stage, the data query stage, specifically includes the following steps:

[0013] Step 2-1: For the user-provided sample space dataset D e The sample space dataset D e Mapping the spatial grid to the spatial data warehouse, the sample spatial dataset D e Corresponding grid data set Each ID in the dataset is mapped to a Bloom filter, forming a vector set. Then for the vector set After vectorization and encryption, it is passed to the cloud server as a search trapdoor Tr.

[0014] Step 2-2: After receiving the search trapdoor Tr, the cloud server performs a joinable dataset query on the high-dimensional vector index to obtain the candidate set. Candidate set The corresponding set of encryption vectors is

[0015] Steps 2-3: Calculate the candidate set The corresponding set of encryption vectors Encryption vectors in With encryption decision vector The vector dot product is used to find the k spatial datasets with the largest spatial differences from the sample spatial dataset, which are then used as the query results R. Finally, the query results R are returned to the user.

[0016] A further improvement of the present invention is that: step 1-1 is based on the connectability parameter threshold ξ for the spatial data warehouse. Perform global mesh partitioning to form a mesh set. The specific process is as follows: Define the connectivity threshold ξ, that is... Time represents the sample space dataset D e and spatial dataset D i Connectivity is achieved by transforming this problem into a mesh problem at this stage and setting a connectivity threshold. Using θ as the dividing force, the entire data warehouse data space is divided into θ×θ grids, forming a grid set. Each grid cell is assigned a unique cell ID.

[0017] A further improvement of the present invention is that: steps 1-3 involve the spatial dataset D in the spatial data warehouse. i The specific process of vectorization encryption is as follows:

[0018] Step 1-3-1, For spatial data warehouse Each spatial dataset D in i The resulting grid data set Perform Bloom filter mapping to form a vector BF i ;

[0019] Step 1-3-2: For each spatial dataset, the vector BF i Using the random vector s in the key K to transform the vector BF i Split into BF i ′ and BF i ", denoted as BF i =(BF i ′,BF i The splitting rules are as follows:

[0020]

[0021] Using the invertible matrices M1 and M2 in key K to pair vector BF i Encryption is performed to generate the corresponding encryption vector. Let θ×θ-dimensional all-1 vector be the decision vector. The decision vector v is segmented and encrypted using a random vector s and invertible matrices M1 and M2 to generate an encrypted decision vector.

[0022] Step 1-3-3: Use a high-dimensional vector index (such as Faiss) to store the encrypted vector. With encryption decision vector This index has a multi-level structure, which can meet the needs of fast querying of multi-dimensional vectors.

[0023] A further improvement of the present invention is that: step 2-1 is based on the sample space dataset D e The specific process of constructing a trapdoor is as follows:

[0024] Step 2-1-1: Transfer the sample space dataset D e Mapped into the spatial grid of the spatial data warehouse to form a grid data set grid data set Each ID is mapped to a Bloom filter, forming a vector set.

[0025] Step 2-1-2: Using the random vector s and invertible matrices M1 and M2 in the key K to plot the vector set Each vector in the dataset is split and encrypted to form a set of encrypted vectors. For each item in the set, specifically... The set of encryption vectors As a search trapdoor Tr, the search trapdoor Tr is sent to the cloud server to perform the search operation.

[0026] A further improvement of the present invention is that in step 2-2, the cloud server performs the search operation through the search trapdoor as follows: after receiving the search trapdoor Tr, the cloud server determines the sample space dataset D in the high-dimensional vector index through the vector inner product. e and spatial dataset D i Whether it is connectable, when the sample space dataset D e and spatial dataset D i A space is considered connectable when the number of overlapping grids is greater than the connectable threshold τ. If it is connectable, the spatial dataset D is... i Add to candidate set The calculation and judgment method is as follows:

[0027] In high-dimensional vector indexing, encrypted vector sets are used. Each item in the set is associated with the encryption vector. Perform a dot product operation. If the dot product result is 0, it indicates that the sample space dataset D... e and spatial dataset D i There is no connectable region; otherwise, it indicates that the sample space dataset D does not exist. e and spatial dataset D i There are connectable regions. The number of overlapping grid cells, Sum, is recorded. Given a connectability threshold τ, a connection is considered complete when Sum > τ, indicating a valid sample space dataset D. e and spatial dataset D i To make it connectable, the spatial dataset D i Add to candidate set

[0028] A further improvement of the present invention lies in the cloud server computing the sample space dataset D in steps 2-3. e and spatial dataset D i Spatial dissimilarity operation is as follows: using candidate sets The set of encrypted vectors in Each item in the encryption decision vector Spatial dissimilarity is calculated using vector dot product, and the top k spatial datasets are returned as the final result set R. The specific calculation and judgment method is as follows:

[0029] In candidate datasets In the process, encrypted decision vectors are used. and candidate set Chinese encrypted vector set After performing a dot product operation on each item, the number of connectable regions (Sum) of each spatial dataset is subtracted to obtain the spatial dissimilarity. Specifically, this is calculated as follows: The results show that the sample space dataset D e and spatial dataset D i The spatial dissimilarity is calculated, and the results are sorted and the top k spatial datasets are returned to the user as the result set R.

[0030] The beneficial effects of the present invention are:

[0031] This invention utilizes a grid inverted index to determine the connectivity of spatial datasets, which can effectively reduce the computational load and complexity of queries and shorten query time.

[0032] This invention transforms the calculation of spatial dissimilarity in non-overlapping regions of candidate sets connected to sample datasets into the calculation of the number of unconnected grids, greatly reducing the computational load of spatial dissimilarity.

[0033] This invention allows for the specification of specific connectability precision by adjusting the parameter ξ, enabling finer-grained connectability queries on datasets.

[0034] This invention employs vector dot product operations throughout the query process, perfectly matching the above design and balancing security and query efficiency. Attached Figure Description

[0035] Figure 1 This is a flowchart of the present invention.

[0036] Figure 2 This is a description map of the connectable regions of the spatial dataset of this invention.

[0037] Figure 3 This is a spatial difference description diagram of the spatial dataset of this invention.

[0038] Figure 4 This is the system architecture diagram of the present invention.

[0039] Figure 5 This is a schematic diagram comparing the encryption method without an index and the encryption method of this application in terms of search time.

[0040] Figure 6 This is a schematic diagram comparing the search time of the encryption method without an index and the encryption method of this application with a fixed grid division strength. Detailed Implementation

[0041] The embodiments of the present invention will be disclosed below with illustrations. For clarity, many practical details will also be described in the following description. However, it should be understood that these practical details should not be used to limit the present invention. That is, in some embodiments of the present invention, some practical details are not essential.

[0042] For ease of description, the relevant symbols are defined as follows:

[0043] Spatial Data Warehouse It contains m spatial datasets, Given a spatial dataset containing n two-dimensional spatial location points, D i ={p i,1 ,p i,2 ,…,p i,n The sample space dataset is D. e The grid set of the entire data space is

[0044] Figure 1 As shown, the present invention is a sample-based method for connecting and querying privacy-preserving spatial datasets. The method includes the following two stages: the first stage is the data preprocessing stage; the second stage is the data query stage.

[0045] The first stage, the data preprocessing stage, specifically includes the following steps:

[0046] Step 1-1: Based on the set connectability parameter threshold ξ, perform the following steps on the spatial data warehouse. Perform global mesh generation and set the connectability parameter threshold. The spatial data warehouse is divided into grids of equal size θ×θ, forming a grid set. Each grid cell is assigned a unique cell ID (e.g., ID0 for f0), ranging from [0, θ]. 2 -1].

[0047] The specific process is as follows: Define the connectivity threshold ξ, that is... Time represents the sample space dataset D e and spatial dataset D i Connectivity is achieved by transforming this problem into a mesh problem at this stage and setting a connectivity threshold. Using θ as the dividing force, the entire data warehouse data space is divided into θ×θ grids, forming a grid set. Each grid cell is assigned a unique cell ID.

[0048] Steps 1-2: The data owner generates a key K = {s, M1, M2}, where s is a randomly generated θ. 22D vectors, M1, M2 are (θ) 2 )×(θ 2 A 1 / 2 invertible matrix, where the key K is shared by the data owner and the querying user;

[0049] Steps 1-3: For spatial data warehouses Each spatial dataset D in i , spatial dataset D i Each spatial data point in the dataset is mapped to a grid, and each of the spatial datasets D i Form a grid data set and grid data set Mapped onto a Bloom filter, based on each grid data set Form a vector BF i and for vector BF i After vectorization, encryption is performed to form an encrypted vector. Let the decision vector The decision vector v is then vectorized and encrypted to form an encrypted decision vector. Finally, the encryption vector and encryption decision vector Store it in a high-dimensional vector index. The specific steps are as follows:

[0050] The specific process is as follows:

[0051] Step 1-3-1, For spatial data warehouse Each spatial dataset D in i The resulting grid data set Perform Bloom filter mapping to form a vector BF i ;

[0052] Step 1-3-2: For each spatial dataset, the vector BF i Using the random vector s in the key K to transform the vector BF i Split into BF i ′ and BF i ", denoted as BF i =(BF i ′,BF i The splitting rules are as follows:

[0053]

[0054] Using the invertible matrices M1 and M2 in key K to pair vector BF i Encryption is performed to generate the corresponding encryption vector. Let θ×θ-dimensional all-1 vector be the decision vector. The decision vector v is segmented and encrypted using a random vector s and invertible matrices M1 and M2 to generate an encrypted decision vector.

[0055] Step 1-3-3: Use high-dimensional vector indexes such as Faiss to store the encrypted vectors. With encryption decision vector This index has a multi-level structure, which can meet the needs of fast querying of multi-dimensional vectors.

[0056] The data query stage includes the following steps:

[0057] Step 2-1: For the user-provided sample space dataset D e The sample space dataset D e Mapping the spatial grid to the spatial data warehouse, the sample spatial dataset D e Corresponding grid data set Each ID in the dataset is mapped to a Bloom filter, forming a vector set. Then for the vector set After vectorization and encryption, it is passed to the cloud server as a search trapdoor Tr.

[0058] Based on the sample space dataset D e The specific process of constructing a trapdoor is as follows:

[0059] Step 2-1-1: Transfer the sample space dataset D e Mapped into the spatial grid of the spatial data warehouse to form a grid data set grid data set Each ID is mapped to a Bloom filter, forming a vector set.

[0060] Step 2-1-2: Using the random vector s and invertible matrices M1 and M2 in the key k, the vector set... Each vector in the dataset is split and encrypted to form a set of encrypted vectors. For each item in the set, specifically... The set of encryption vectors As a search trapdoor Tr, the search trapdoor Tr is sent to the cloud server to perform the search operation.

[0061] Step 2-2: After receiving the search trapdoor Tr, the cloud server performs a joinable dataset query on the high-dimensional vector index to obtain the candidate set. Candidate set The corresponding set of encryption vectors is The cloud server performs a search operation through a search trapdoor as follows: After receiving the search trapdoor Tr, the cloud server determines the sample space dataset D by using the vector inner product in the high-dimensional vector index. e and spatial dataset D i Whether it is connectable, when the sample space dataset D e and spatial dataset D i A space is considered connectable when the number of overlapping grids is greater than the connectable threshold τ. If it is connectable, the spatial dataset D is... i Add to candidate set The calculation and judgment method is as follows:

[0062] In high-dimensional vector indexing, encrypted vector sets are used. Each item in the set is associated with the encryption vector. Perform a dot product operation. If the dot product result is 0, it indicates that the sample space dataset D... e and spatial dataset D i There is no connectable region; otherwise, it indicates that the sample space dataset D does not exist. e and spatial dataset D i There are connectable regions. The number of overlapping grid cells, Sum, is recorded. Given a connectability threshold τ, a connection is considered complete when Sum > τ, indicating a valid sample space dataset D. e and spatial dataset D i To make it connectable, the spatial dataset D i Add to candidate set

[0063] Steps 2-3: Calculate the candidate set The corresponding set of encryption vectors Encryption vectors in With encryption decision vector The vector dot product is used to find the k spatial datasets with the largest spatial differences from the sample spatial dataset, which are then used as the query results R. Finally, the query results R are returned to the user.

[0064] Cloud server computing sample space dataset D e and spatial dataset D i Spatial dissimilarity operation is as follows: using candidate sets The set of encrypted vectors in Each item in the encryption decision vector Spatial dissimilarity is calculated using vector dot product, and the top k spatial datasets are returned as the final result set R. The specific calculation and judgment method is as follows:

[0065] In candidate datasets In the process, encrypted decision vectors are used. and candidate set Chinese encrypted vector set After performing a dot product operation on each item, the number of connectable regions (Sum) of each spatial dataset is subtracted to obtain the spatial dissimilarity. Specifically, this is calculated as follows: The results show that the sample space dataset D e and spatial dataset D i The spatial dissimilarity is calculated, and the results are sorted and the top k spatial datasets are returned to the user as the result set R.

[0066] Figure 2 The diagram shown is a description of the connectable regions of the spatial dataset of this invention, where the shaded area represents the sample spatial dataset D. e and spatial dataset D i The spatially connectable area is composed of Figure 2 It can be seen that the sample space dataset is D. e and spatial dataset D i There are two connectable regions.

[0067] Figure 3 The figure shown is a spatial difference description diagram of the spatial dataset of this invention. The yellow area represents the spatial difference between the query dataset and the sample spatial dataset. As can be seen from the figure, the spatial dataset D... i With the sample space dataset being D e The spatial variability is 5.

[0068] Figure 4 The system framework diagram of this invention includes a data owner, a cloud server, and a query user. The data owner possesses plaintext data in the data warehouse and is responsible for vectorizing and encrypting the spatial dataset within the data warehouse. The encrypted data is then outsourced and stored on the cloud server, with the encryption key shared with the query user. When performing a spatial dataset sample query, the query user constructs a search trapdoor based on the spatial distribution characteristics of the sample dataset and sends it to the cloud server. The cloud server uses the search trapdoor and the encrypted spatial dataset to perform a privacy-preserving joinable query on the spatial dataset and returns the query results to the query user.

[0069] To verify this application, a search time comparison was made between the encryption method without an index and the encryption method of this application when the dimension of the vector generated by the fixed Bloom filter was 1000. Figure 5 As shown, the search time for both methods fluctuates around a certain value as the grid division intensity changes, but the query method in this application is two orders of magnitude faster than the ordinary query method.

[0070] At the same time, such as Figure 6As shown, this application compares the search time of the encryption method without an index and the method of this application when the mesh partitioning intensity is fixed. Both methods show an increase in query time with increasing vector dimension, but the query time of the method without an index increases more significantly with increasing vector dimension, while the query time of the encryption method of this application increases less significantly with increasing vector dimension. The search time of the encryption method of this application is three orders of magnitude faster than that of the method without an index, a substantial difference. Figure 6 The time data is represented using logarithmic scale data of log2.

[0071] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A sample-based method for connecting and querying privacy-preserving spatial datasets, applicable to an application scenario consisting of a data owner, a cloud server, and a querying user, characterized in that: Let the spatial data warehouse be The spatial data warehouse contains m datasets, and each spatial dataset D... i ={p i,1 ,p i,2 ,…,p i,n }, which contains n spatial data points, and the sample spatial dataset is D. e Based on the example-based privacy-preserving spatial dataset, join queries can be performed in the encrypted state within the spatial data warehouse. Obtaining the sample space dataset D from the sample space dataset e Connectable candidate datasets, and finding the original sample space dataset D within the candidate datasets. e The query method for the top k spatially disparate datasets involves two stages: The first stage, the data preprocessing stage, specifically includes the following steps: Step 1-1: Based on the set connectability parameter threshold ξ, perform the following steps on the spatial data warehouse. Perform global mesh generation and set the connectability parameter threshold. The spatial data warehouse is divided into grids of equal size θ×θ, forming a grid set. Each grid cell is assigned a unique cell ID, ranging from [0, θ]. 2 -1]; Steps 1-2: The data owner generates a key K = {s, M1, M2}, where s is a randomly generated θ. 2 2D vectors, M1, M2 are (θ) 2 )×(θ 2 A 1 / 2 invertible matrix, where the key K is shared by the data owner and the querying user; Steps 1-3: For spatial data warehouses Each spatial dataset D in i , spatial dataset D i Each spatial data point in the dataset is mapped to a grid, and each of the spatial datasets D i Form a grid data set and grid data set Mapped onto a Bloom filter, based on each grid data set Form a vector BF i and for vector BF i After vectorization, encryption is performed to form an encrypted vector. Let the decision vector The decision vector v is then vectorized and encrypted to form an encrypted decision vector. Finally, the encryption vector and encryption decision vector Store in a high-dimensional vector index; The second stage, the data query stage, specifically includes the following steps: Step 2-1: For the user-provided sample space dataset D e The sample space dataset D e Mapping the spatial grid to the spatial data warehouse, the sample spatial dataset D e Corresponding grid data set Each ID in the dataset is mapped to a Bloom filter, forming a vector set. Then for the vector set After vectorization and encryption, it is passed to the cloud server as a search trapdoor Tr. Step 2-2: After receiving the search trapdoor Tr, the cloud server performs a joinable dataset query on the high-dimensional vector index to obtain the candidate set. Candidate set The corresponding set of encryption vectors is Steps 2-3: Calculate the candidate set The corresponding set of encryption vectors Encryption vectors in With encryption decision vector The vector dot product is used to find the k spatial datasets with the largest spatial differences from the sample spatial dataset, which are then used as the query results R. Finally, the query results R are returned to the user.

2. The example-based method for connecting and querying privacy-preserving spatial datasets according to claim 1, characterized in that: Step 1-1 is based on the connectability parameter threshold ξ for the spatial data warehouse. Perform global mesh partitioning to form a mesh set. The specific process is as follows: Define the connectivity threshold ξ, that is... Time represents the sample space dataset D e and spatial dataset D i Connectivity is achieved by transforming this problem into a mesh problem at this stage and setting a connectivity threshold.

3. The example-based method for connecting and querying privacy-preserving spatial datasets according to claim 1, characterized in that: Steps 1-3 involve the spatial dataset D in the spatial data warehouse. i The specific process of vectorization encryption is as follows: Step 1-3-1, For spatial data warehouse Each spatial dataset D in i The resulting grid data set Perform Bloom filter mapping to form a vector BF i ; Step 1-3-2: For each spatial dataset, the vector BF i Using the random vector s in the key K to transform the vector BF i Split into BF i ′ and BF i ", denoted as BF i =(BF i ′,BF i Using the invertible matrices M1 and M2 in the key K to manipulate vector BF i Encryption is performed to generate the corresponding encryption vector. Let θ×θ-dimensional all-1 vector be the decision vector. The decision vector v is segmented and encrypted using a random vector s and invertible matrices M1 and M2 to generate an encrypted decision vector. Step 1-3-3: Store the encrypted vector using a high-dimensional vector index. With encryption decision vector 4. The example-based method for connecting and querying privacy-preserving spatial datasets according to claim 1, characterized in that: Step 2-1 is based on the sample space dataset D. e The specific process of constructing a trapdoor is as follows: Step 2-1-1: Transfer the sample space dataset D e Mapped into the spatial grid of the spatial data warehouse to form a grid data set grid data set Each ID is mapped to a Bloom filter, forming a vector set. Step 2-1-2: Using the random vector s and invertible matrices M1 and M2 in the key K to plot the vector set Each vector in the dataset is split and encrypted to form a set of encrypted vectors. For each item in the set, specifically... The set of encryption vectors As a search trapdoor Tr, the search trapdoor Tr is sent to the cloud server to perform the search operation.

5. The example-based method for connecting and querying privacy-preserving spatial datasets according to claim 4, characterized in that: In step 2-2, the cloud server performs a search operation through the search trapdoor as follows: After receiving the search trapdoor Tr, the cloud server determines the sample space dataset D in the high-dimensional vector index through the vector inner product. e and spatial dataset D i Whether it is connectable, when the sample space dataset D e and spatial dataset D i A space is considered connectable when the number of overlapping grids is greater than the connectable threshold τ. If it is connectable, the spatial dataset D is... i Add to candidate set The calculation and judgment method is as follows: In high-dimensional vector indexing, encrypted vector sets are used. Each item in the set is associated with the encryption vector. Perform a dot product operation. If the dot product result is 0, it indicates that the sample space dataset D... e and spatial dataset D i There is no connectable region; otherwise, it indicates that the sample space dataset D does not exist. e and spatial dataset D i There are connectable regions. The number of overlapping grid cells, Sum, is recorded. Given a connectability threshold τ, a connection is considered complete when Sum > τ, indicating a valid sample space dataset D. e and spatial dataset D i To make it connectable, the spatial dataset D i Add to candidate set 6. A sample-based privacy-preserving space dataset joinable query method according to claim 1: In steps 2-3, the cloud server calculates the sample space dataset D. e and spatial dataset D i Spatial dissimilarity operation is as follows: using candidate sets The set of encrypted vectors in Each item in the encryption decision vector Spatial dissimilarity is calculated using vector dot product, and the top k spatial datasets are returned as the final result set R. The specific calculation and judgment method is as follows: In candidate datasets In the process, encrypted decision vectors are used. and candidate set Chinese encrypted vector set After performing a dot product operation on each item, the number of connectable regions (Sum) of each spatial dataset is subtracted to obtain the spatial dissimilarity. Specifically, this is calculated as follows: The results show that the sample space dataset D e and spatial dataset D i The spatial dissimilarity is calculated, and the results are sorted and the top k spatial datasets are returned to the user as the result set R.

Images