CN113254743A - Secure semantic perception search method for dynamic spatial data in Internet of vehicles - Google Patents

Secure semantic perception search method for dynamic spatial data in Internet of vehicles Download PDF

Info

Publication number
CN113254743A
CN113254743A CN202110602922.XA CN202110602922A CN113254743A CN 113254743 A CN113254743 A CN 113254743A CN 202110602922 A CN202110602922 A CN 202110602922A CN 113254743 A CN113254743 A CN 113254743A
Authority
CN
China
Prior art keywords
vehicle
search
theme
data
vector
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110602922.XA
Other languages
Chinese (zh)
Other versions
CN113254743B (en
Inventor
马建峰
杨帆
李佳忆
苗银宾
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xidian University
Original Assignee
Xidian University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xidian University filed Critical Xidian University
Priority to CN202110602922.XA priority Critical patent/CN113254743B/en
Publication of CN113254743A publication Critical patent/CN113254743A/en
Application granted granted Critical
Publication of CN113254743B publication Critical patent/CN113254743B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/907—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/909—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using geographical or spatial information, e.g. location
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/22—Indexing; Data structures therefor; Storage structures
    • G06F16/2228—Indexing structures
    • G06F16/2246—Trees, e.g. B+trees
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/29—Geographical information databases
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/3331—Query processing
    • G06F16/334—Query execution
    • G06F16/3344—Query execution using natural language analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/602—Providing cryptographic facilities or services

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Bioethics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Hardware Design (AREA)
  • Computer Security & Cryptography (AREA)
  • Library & Information Science (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Remote Sensing (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention discloses a security semantic perception search method of dynamic spatial data in a car networking, which comprises the steps of establishing an encryption position index structure for a standard data set, encrypting the data set, sending the encryption index structure and a ciphertext data set to a public cloud server, carrying out main body training on a legal car user given query keyword set w by a vehicle by adopting a theme model to obtain a query theme vector, generating a position search token and a theme probability search token by the vehicle by utilizing an encryption key and query information, searching the encryption position index structure by the position search token and the theme probability search token, carrying out matching calculation on search results to obtain a matching set, obtaining a result set after sequencing, and sending plaintext information of the result set to the vehicle by a private cloud server, wherein the method accelerates the search efficiency of an address position without influencing a comparison result after vector calculation, the data security on the public cloud is ensured, and the privacy information inquired by the user is also ensured.

Description

Secure semantic perception search method for dynamic spatial data in Internet of vehicles
Technical Field
The invention relates to the field of information security, in particular to a secure semantic perception searching method for dynamic spatial data in a vehicle networking.
Background
Location-based services, such as Spatial Key Search (SKS), are becoming increasingly common in many different applications, from social media to vehicle ad hoc networks (VANET). In VANET, communication between high speed vehicles relies on board units (OBUs; complex modules including wireless communication, embedded systems, sensors and positioning systems) and roadside units (RSUs). The dynamic nature of VANET (e.g., fast-moving vehicles) exacerbates the challenges of designing and implementing a secure and efficient SKS scheme, e.g., due to frequent communications, data explosion, and insufficient coverage between high-speed moving vehicles and other devices/systems in the infrastructure, and one widely used approach to addressing constraints in cellular networks is to send search queries and retrieve content from RSUs. Although the network link between the RSU and the vehicle has a relatively high bandwidth and low latency, it requires the vehicle to continuously transmit messages over the RSU. Since RSUs are physically exposed, they are vulnerable to a wide range of attacks.
The RSU and SKS service providers may not be fully trusted by the vehicle user. Since the search query from the vehicle user contains private information such as real-time coordinates and search preferences, any privacy disclosure of the query may have undesirable consequences. For example, a user is searching for a place that offers accommodation, rather than a restaurant, and the leakage of such a search may affect the privacy of the user. Forward privacy is also a basic requirement in SKS schemes that support dynamic updates. For example, a file injection attack is a typical attack that is applicable to dynamic schemes. In such an attack, the adversary may forge some data and send the forged data to the data owner. After being encrypted by the data owner, the ciphertext of the forged data is outsourced to a service provider for storage. By analyzing previous queries in conjunction with newly inserted data, an adversary may be able to recover some of the outsourced data and queries. This requires that the SKS scheme for VANET must be designed with both search efficiency and privacy protection in mind.
Searchable Encryption (SE) may be used to implement a privacy preserving spatial key search in VANET. However, many existing SE schemes focus only on security key searches on outsourced documents, without regard to scope limitations. In addition, the mobility of the vehicle user requires a faster and more efficient search process because the vehicle may be out of coverage in minutes when the vehicle is traveling at high speeds (e.g., 80 miles per hour). The public key cryptosystem scheme is too time consuming to provide real-time search results. Therefore, it remains challenging how to return top-level results that satisfy conditions such as the vehicle user's search intent and query scope constraints in a persistent, private manner.
Disclosure of Invention
Aiming at the defects of the searchable encryption technology in the vehicle-mounted ad hoc network, the invention provides a secure semantic perception searching method for dynamic spatial data in the vehicle-mounted ad hoc network, which solves the contradiction between semantic perception searching efficiency and privacy protection in the prior art and increases the limit of considering the geographical position range.
The invention is realized by the following technical scheme:
a secure semantic perception search method for dynamic spatial data in the Internet of vehicles comprises the following steps:
step 1, a private cloud server generates an encryption key for a data set D;
step 2, performing theme training on the data set D0 by using the trained LDA theme model to obtain a theme deviation vector of a store corresponding to each record, and converting merchant descriptions in the data set D0 into theme deviation vectors to obtain a standard data set D;
step 3, the private cloud server establishes an encrypted position index structure for the standard data set D, encrypts the data set D at the same time, and sends the encrypted position index structure and the ciphertext data set to the public cloud server;
step 4, the vehicle adopts an LDA topic model to carry out main body training on a given query keyword set w of a legal vehicle user to obtain a query topic vector, and the vehicle generates a position search token and a topic probability search token by using an encryption key and query information;
step 5, the public cloud server searches the encrypted position index structure through the position search token to obtain a suspected area set of suspected search results, then screens the suspected area set through the position search token and the theme probability search token, calculates the matching degree of the screened results to obtain a matching set P, sorts the matching set P according to the matching degree scores, and forms the first k results into a result set;
and 6, the private cloud server sends the plaintext information of the result set to the vehicle.
Preferably, in step 2, the private cloud server performs topic training on the keyword array od in each record o 'in the D0 set according to the trained LDA topic model to obtain a topic bias vector of the store corresponding to o', and updates the topic bias vector for the keyword array in each record in the data set D0 to obtain an updated data set D.
Preferably, the private cloud server in step 3 adopts an R tree structure, and establishes a position index structure for the data set D according to the geographic information ol
Figure BDA0003093299830000031
Position indexing structure
Figure BDA0003093299830000032
The node in (1) is encrypted to obtain an encrypted position index structure
Figure BDA0003093299830000033
Preferably, the encryption method of the position index structure is as follows:
the private cloud server vectorizes each node to obtain a vector set, and encrypts the vector set by adopting an asymmetric scalar product order-preserving encryption algorithm to obtain an encrypted position index structure
Figure BDA0003093299830000034
Preferably, in step 4, legal vehicle users give a query keyword set w and a query range vlThe vehicle requests the trained LDA theme model and the encryption key K from the private server;
the vehicle carries out theme training on the query keyword set w by utilizing the LDA theme model to obtain a query theme vector vψVehicle using key K, inquiry range vlAnd query topic vector vψGenerating a vehicle to location lookup token TLAnd topic probability lookup token TNAnd sending the data to a public cloud server for searching.
Preferably, the method for generating the location finding token is as follows:
vehicle user pair query range vlAnd disturbing, vectorizing to obtain a vector set, and encrypting the vector set by adopting an ASPE (asymmetric scalar product order preserving encryption algorithm) to obtain the position search token.
Preferably, the method for generating the topic probability lookup token is as follows:
query subject vector v by adopting asymmetric scalar product order-preserving encryption algorithm ASPEψAnd encrypting to obtain the theme probability search token.
Preferably, the method for searching the suspected area set in step 5 is as follows:
and starting searching from a root node of the encryption position index structure, comparing the inner product of the current node and the area search token, determining whether the node comprises a suspected area, and traversing child nodes of the node if the node comprises the suspected area to obtain a suspected area set.
Preferably, the method for obtaining the matching set P is as follows:
and judging the geographic position of each node in the suspected area set by using the position search token, if the node meets the searched geographic requirement, calculating the matching degree, performing matching degree score calculation on the theme deviation vector in the node by using the theme probability search token, then sequencing the id of the nodes according to the matching degree score, selecting the id of k nodes before ranking, forming a result set, and returning the result set to the vehicle.
Preferably, after the vehicle obtains the result set in step 6, the plaintext information of the result set is requested to the private cloud through the encryption channel, and after the private cloud server receives the result set, the plaintext information corresponding to id in the result set is searched from the data set D to form a plaintext result set G which is sent to the vehicle.
Compared with the prior art, the invention has the following beneficial technical effects:
the invention provides a security semantic perception searching method of dynamic spatial data in the internet of vehicles,
according to the method for searching dynamic spatial data in the internet of vehicles through secure semantic perception, the R tree index is established for the data set through the geographic position of the object, and the process of searching suspected objects in the searching process is accelerated. Potential semantic features are extracted from keywords of the object description and the query based on the LDA topic model, and text similarity is measured through the extracted semantic features. Semantic awareness functionality of the query is provided. The nodes and the subject bias vectors in the R-tree are encrypted using the ASPE encryption algorithm. The encryption algorithm ensures that the operation between the encrypted vectors has the same size characteristics as the operation between the plaintext vectors. The encryption algorithm can ensure the security of data on one hand and realize the functions of searching and matching information on the basis of encrypting information on the other hand.
The data index structure is constructed using an R-tree. The R tree structure facilitates the storage of the object index and simultaneously makes the search on the geographic position more efficient; corresponding algorithms are also designed to prune non-leaf nodes and speed up the process of searching potential leaf nodes.
Abandoning the encryption keyword matching method and adopting an LDA model. The textual description of the object and the query keywords are converted into topic bias vectors using an LDA model. The similarity between objects and queries can be measured by calculating the similarity of their topic bias vectors. The semantic search capability is fully improved, the search process is changed from a text matching process to a vector calculation process, and the search speed is greatly increased.
And encrypting the geographic position vector and the theme deviation vector by adopting an ASPE (asymmetric scalar product order preserving encryption algorithm) method, so that the characteristics of a calculation result after the vector is encrypted are not influenced. On one hand, confidentiality of plaintext data is guaranteed through encryption, and on the other hand, operational characteristics of the encrypted vector are effectively maintained, so that effective searching can be conducted on encrypted data. Therefore, the data security of the index structure information and the data set is effectively protected, privacy disclosure brought by vehicle search is reduced to the maximum extent, and the data security under the outsourcing data search scene is effectively solved.
Drawings
FIG. 1 is a schematic diagram of a vehicle find merchant system for use with the present invention;
FIG. 2 is a flowchart of the secure semantic perception search method of dynamic spatial data in the Internet of vehicles according to the present invention.
Detailed Description
The present invention will now be described in further detail with reference to the attached drawings, which are illustrative, but not limiting, of the present invention.
Referring to fig. 1, the system of the invention takes a vehicle lookup merchant as an example, and comprises three entities, a trusted private cloud server, a public cloud server and a vehicle user.
The private cloud server trains a model, generates a theme vector and an auxiliary vector, and establishes an R tree index, encrypted data and an index structure; the public cloud server provides storage and retrieval service of the ciphertext data; the vehicle user is responsible for initiating the query request and generating the query token.
Referring to fig. 2, a method for secure semantic perception search of dynamic spatial data in the internet of vehicles includes the following steps:
step 1: the private cloud server generates an encryption key K for the data set D by adopting an ASPE encryption algorithm;
wherein, K ═ { M1, M2}, M1, M2 are random n × n reversible matrices, and n is the topic number of LDA topic model training.
And 2, performing theme training on the data set D0 by using the trained LDA theme model to obtain a theme deviation vector of a store corresponding to each record, and converting merchant descriptions in the data set D0 into the theme deviation vector to obtain a standard data set D.
The specific implementation of this step is as follows:
2.1) given a data set D0, the data set contains a plurality of records o', each record o ═ id, ol,odIs composed of 3 dimensions, id is the identity number of the store corresponding to each record, olGeographical location information, o, indicating the store to which the record correspondsl＝{ox，oyIn which o isxIndicates the longitude, o, of the storeyRepresenting the latitude; odIs a set of arrays describing the store keywords, for example: fruits, food, etc., of variable length.
2.2) the private cloud server pairs the keyword array o in each record o' in the D0 set according to the trained LDA topic modeldSubject training is carried out to obtain a subject deviation vector o of a shop corresponding to the subject deviation vector oψ＝{p1，…，pnN is the number of topics trained by the LDA model, e.g. 10, which is a fixed value.
O' in each record in D0 is set to { id, o ═ Dl,odUpdating to o ═ id, ol,oψ}. Record o in all ddAfter the replacement, the new data set is obtained as D.
Step 3, the private cloud server establishes an encrypted index structure for the data set D
Figure BDA0003093299830000071
And encrypts the data set and constructs an encryption index
Figure BDA0003093299830000072
And sending the ciphertext data set C to a public cloud server;
the steps are specifically realized as follows:
3.1) the private cloud server establishes a position index structure for the data set D, adopts an R tree structure, and obtains the geographic information olBuilding an R-Tree index for dataset D
Figure BDA0003093299830000073
Namely a position index structure, and each leaf node stores all store information id contained in the area;
3.2) the private cloud server encrypts the nodes in the R tree index to obtain the encrypted R tree index
Figure BDA0003093299830000074
The specific steps for encrypting the R tree index are as follows:
the encryption of the R tree structure is to encrypt each node information mi in the R tree,
Figure BDA0003093299830000075
Figure BDA0003093299830000076
Figure BDA0003093299830000077
is the range of mi in longitude,
Figure BDA0003093299830000078
is the range of mi in latitude, i is more than or equal to 1 and less than or equal to R, and R is the number of nodes in the whole R tree. The mi is encrypted as follows:
3.2.1.1) carrying out vectorization processing on mi by the private cloud server to obtain a vector set
Figure BDA0003093299830000079
Figure BDA00030932998300000710
Wherein
Figure BDA00030932998300000711
Figure BDA00030932998300000712
Where w1, …, wj, (j ═ n-4), is a random number between (0,1), and is given a vector length of n for the extended portion of the vector.
3.2.1.2) pair of ASPEs (asymmetric scalar product order preserving encryption algorithms)
Figure BDA00030932998300000713
And (3) encryption:
Figure BDA00030932998300000714
obtaining an encrypted vector
Figure BDA0003093299830000081
The vector length is n. Encrypting mi as
Figure BDA0003093299830000082
3.2.2) encryption of all nodes of the R tree is completed to obtain the index of the encrypted R tree
Figure BDA0003093299830000083
I.e. the encryption position index structure.
3.3) the private cloud server encrypts the data in the data set D to obtain an encrypted data set C;
the specific steps for encrypting the data set D are as follows:
3.3.1) taking out each record in the data set, and carrying out the operation of setting a single record o as { id, o ═ idl,oψAnd (5) encrypting. The specific steps for encrypting a single record are as follows:
3.3.1.1) to o in a single record ol＝{ox，oyEncrypting, specifically comprising the following steps:
3.3.1.1.1) private cloud server pair olVectorizing to obtain a vector set
Figure BDA0003093299830000084
Figure BDA0003093299830000085
3.3.1.1.2) adopt asymmetric scalar product order-preserving encryption algorithm ASPE pair
Figure BDA0003093299830000086
And (3) encryption:
Figure BDA0003093299830000087
obtaining an encrypted vector
Figure BDA0003093299830000088
The vector length is n. Encryption olIs composed of
Figure BDA0003093299830000089
3.3.1.2) o to a single record oψ＝{p1，…，pnCarry on encryption. ASPE (asynchronous serial encryption processor) pair o adopting asymmetric scalar product order-preserving encryption algorithmψAnd (3) encryption:
Figure BDA00030932998300000810
obtaining an encrypted vector
Figure BDA00030932998300000811
The vector length is n.
3.3.2) Total o in each record for dataset Dl,oψAfter being encrypted, obtain
Figure BDA00030932998300000812
And
Figure BDA00030932998300000813
replacing o in data set Dl,oψObtaining an encrypted data set C, wherein the recorded data in C is
Figure BDA00030932998300000814
3.4) private cloud Server encryption index
Figure BDA00030932998300000815
And sending the ciphertext data set C to a public cloud server;
and 4, step 4: legal vehicle user gives query keyword set w and query range vlThe vehicle requests the trained LDA topic model and the encryption key K from the private server, and the vehicle utilizes the keyK and query information { vl，vψGenerate a vehicle to location lookup token TLAnd topic probability lookup token TNVehicle location finding token TLAnd topic probability lookup token TNAnd sending the data to a public cloud server for searching.
The specific implementation of this step is as follows:
4.1) legitimate vehicle user gives a set of query keywords w ═ { q1, q 2.. qt } and a query range vl＝{[xl，xr]，[yl，yr]}. Wherein q1, q 2.. qt are keywords such as: spicy soup, cate food, night, etc. [ x ] ofl，xr]Is the range of the longitude of the looking merchant, [ yl，yr]Is to find a range of the merchant's latitude.
4.2) vehicle requests trained LDA topic model and encryption key K ═ M from private server1,M2}；
4.3) the vehicle carries out topic training on the query keyword set w by utilizing the LDA topic model to obtain a query topic vector vψ＝{p1，…,pn}；
4.4) vehicle utilization Key K and query information { vl，vψGenerate a location lookup token TLAnd topic probability lookup token TN。
The method comprises the following specific steps:
4.2.1) generating a location lookup token TL：
The method comprises the following specific steps:
4.2.1.1) vehicle user to query range vlDisturbing and vectorizing to obtain a vector set
Figure BDA0003093299830000091
Figure BDA0003093299830000092
Where δ and δ' are large positive numbers of randomly generated perturbations.
4.2.1.2) pair of ASPEs (asymmetric scalar product order preserving encryption algorithms)
Figure BDA0003093299830000093
And (3) encryption:
Figure BDA0003093299830000094
obtaining an encrypted vector
Figure BDA0003093299830000095
Having a vector length of n, wherein
Figure BDA0003093299830000096
Is M1，M2The inverse matrix of (c). Obtaining location finding tokens
Figure BDA0003093299830000097
4.2.2) generating a topic probability lookup token TN. ASPE (asynchronous sequence-preserving encryption) pair v adopting asymmetric scalar product order-preserving encryption algorithmψAnd (3) encryption:
Figure BDA0003093299830000101
obtaining an encrypted vector
Figure BDA0003093299830000102
The vector length is n. Finding a topic probability lookup token
Figure BDA0003093299830000103
4.2.3) vehicle to find the location token TLAnd topic probability lookup token TNAnd sending the data to a public cloud server for searching.
And 5: public cloud server searches token T through positionLSlave encryption index structure
Figure BDA0003093299830000104
The suspected area set e _ M of the possible suspected result is obtained through searching, and the public cloud server searches the token T through the positionLAnd topic probability lookup token TNScreening the merchants contained in the suspected area set e _ M, and matchingAnd calculating to obtain a matching set P. And sequencing the P by the public cloud server according to the matching degree score to obtain a TopK (first K) result set Pk and sending the TopK result set Pk to the vehicle.
The method comprises the following specific steps:
5.1) the public cloud Server looks up the token T by locationLSlave encryption index structure
Figure BDA0003093299830000105
The steps of searching to obtain a suspected area set e _ M are as follows, 5.1.1) -5.1.4):
5.1.1) encryption index Structure
Figure BDA0003093299830000106
The root node of (a) starts the lookup. Enqueue the root node into compare queue B.
5.1.2) taking out the head node e _ m from the comparison queue B, and calculating the current node
Figure BDA0003093299830000107
Figure BDA0003093299830000108
And a region search token
Figure BDA0003093299830000109
The inner products of the corresponding vectors of (a) and (b) are compared,
inner product of corresponding vectors:
Figure BDA00030932998300001010
5.1.3) if rx＜0and ry< 0: and (4) proving that the current node area contains the suspected area, and if the current node is a leaf node, adding the leaf node into the suspected area set. Otherwise, the child nodes of the current node are enqueued in the comparison queue B.
If r is not satisfiedx＜0and ryIf < 0, the node is discarded.
Execution continues at step 5.1.2. Until the comparison queue B is empty
5.1.4) the comparison queue B is empty, and all the area searching is finished. The set of areas containing suspect results, e _ M ═ { e _ mj, … }, where j represents the index of the suspect area.
5.2) the public cloud Server looks up the token T by the locationLAnd topic probability lookup token TNAnd screening from the suspected area set e _ M, and calculating the matching degree to obtain a matching set P.
The method comprises the following specific steps:
5.2.1) extracting the id of all the shops contained in the suspected area in the set e _ M, and forming a suspected target set e _ ids ═ { i, j, … }, wherein i, j … is the id value of a possible result.
5.2.2) extracting the shop information corresponding to the id in the e _ ids from the encrypted data set
Figure BDA0003093299830000111
Carrying out geographic position judgment and matching degree score calculation on the obtained data;
the method comprises the following specific steps:
5.2.2.1) location determination, using location finding tokens
Figure BDA0003093299830000112
For shop
Figure BDA0003093299830000113
Performing a geographic location
Figure BDA0003093299830000114
And (6) judging. Calculating the inner product of the corresponding vectors:
Figure BDA0003093299830000115
if r isx＜0and ryIf the number is less than 0, the node is proved to meet the geographic requirement of searching, and the matching degree is calculated. If not, discarding and continuing to search the next.
5.2.2.2) matching score calculation. Finding tokens using topic probabilities
Figure BDA0003093299830000116
To store
Figure BDA0003093299830000117
Subject probability vector in (1)
Figure BDA0003093299830000118
And (3) carrying out matching degree score calculation:
Figure BDA0003093299830000119
5.2.3) screening and calculating the matching degree score of the object o in the suspected target set e _ ids, and then sorting the id according to the matching degree score.
5.2.4) performing TopK interception on the sorted results, selecting id of k before score ranking to form a query result set RES, and returning the query result set RES to the vehicle.
Step 6: after the vehicle obtains the TopK result set RES, plaintext information of the result set RES is requested to the private cloud server through the encryption channel, and after the private cloud server receives the result set, the plaintext information of each element id of the set is searched from the data set D to form a plaintext result set which is sent to the vehicle.
The method not only considers the text matching degree between the vehicle query and the merchant description, but also adds the limitation to the geographical range. Efficient and safe searching of merchants is realized from the aspects of geographic position and semantic matching.
Second, a data index structure is constructed using the R-tree. The R tree structure facilitates the storage of the object index and simultaneously makes the search on the geographic position more efficient; corresponding algorithms are designed to prune non-leaf nodes and accelerate the process of searching potential leaf nodes.
In addition, the method of matching the encryption keywords is abandoned, and text description and query keywords of the object are converted into topic deviation vectors by using an LDA model. The similarity between objects and queries can be measured by calculating the similarity of their topic bias vectors. The semantic search capability is fully improved, the search process is changed from a text matching process to a vector calculation process, and the search speed is greatly increased.
And finally, encrypting the geographic position vector and the theme deviation vector by adopting an ASPE (asymmetric scalar product order preserving encryption algorithm) method, so that the characteristics of a calculation result after the vector is encrypted are not influenced. On one hand, confidentiality of plaintext data is guaranteed through encryption, and on the other hand, operational characteristics of the encrypted vector are effectively maintained, so that effective searching can be conducted on encrypted data. Therefore, the data security of the index structure information and the data set is effectively protected, privacy disclosure brought by vehicle search is reduced to the maximum extent, and the data security under the outsourcing data search scene is effectively solved.
The above-mentioned contents are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and any modification made on the basis of the technical idea of the present invention falls within the protection scope of the claims of the present invention.

Claims (10)

1. A secure semantic perception search method for dynamic spatial data in the Internet of vehicles is characterized by comprising the following steps:
step 1, a private cloud server generates an encryption key for a data set D;
step 2, performing theme training on the data set D0 by using the trained LDA theme model to obtain a theme deviation vector of a store corresponding to each record, and converting merchant descriptions in the data set D0 into theme deviation vectors to obtain a standard data set D;
step 3, the private cloud server establishes an encrypted position index structure for the standard data set D, encrypts the data set D at the same time, and sends the encrypted position index structure and the ciphertext data set to the public cloud server;
step 4, the vehicle adopts an LDA topic model to carry out main body training on a given query keyword set w of a legal vehicle user to obtain a query topic vector, and the vehicle generates a position search token and a topic probability search token by using an encryption key and query information;
step 5, the public cloud server searches the encrypted position index structure through the position search token to obtain a suspected area set of suspected search results, then screens the suspected area set through the position search token and the theme probability search token, calculates the matching degree of the screened results to obtain a matching set P, sorts the matching set P according to the matching degree scores, and forms the first k results into a result set;
and 6, the private cloud server sends the plaintext information of the result set to the vehicle.
2. The method for secure semantic-aware searching of dynamic spatial data in the internet of vehicles according to claim 1, wherein in step 2, the private cloud server performs keyword array o in each record o' in the D0 set according to the trained LDA topic modeldAnd performing theme training to obtain a theme deviation vector of the shop corresponding to the o', and updating the theme deviation vector by the keyword group in each record in the data set D0 to obtain an updated data set D.
3. The method for secure semantic perception search of dynamic spatial data in the internet of vehicles according to claim 1, wherein the private cloud server in step 3 adopts an R tree structure and searches according to geographic information olBuilding a position index structure for a data set D
Figure FDA0003093299820000021
Position indexing structure
Figure FDA0003093299820000022
The node in (1) is encrypted to obtain an encrypted position index structure
Figure FDA0003093299820000023
4. The method for searching dynamic space data in the internet of vehicles according to claim 3, wherein the encryption method of the position index structure is as follows:
vectorizing each node by the private cloud serverObtaining a vector set, and encrypting the vector set by adopting an asymmetric scalar product order-preserving encryption algorithm to obtain an encrypted position index structure
Figure FDA0003093299820000024
5. The method for secure semantic aware search of dynamic spatial data in internet of vehicles according to claim 1, wherein legal vehicle users in step 4 give a query keyword set w and a query range vlThe vehicle requests the trained LDA theme model and the encryption key K from the private server;
the vehicle carries out theme training on the query keyword set w by utilizing the LDA theme model to obtain a query theme vector vψVehicle using key K, inquiry range vlAnd query topic vector vψGenerating a vehicle to location lookup token TLAnd topic probability lookup token TNAnd sending the data to a public cloud server for searching.
6. The method for secure semantic aware search of dynamic spatial data in a vehicle networking system according to claim 5, wherein the method for generating the location finding token is as follows:
vehicle user pair query range vlAnd disturbing, vectorizing to obtain a vector set, and encrypting the vector set by adopting an ASPE (asymmetric scalar product order preserving encryption algorithm) to obtain the position search token.
7. The method for secure semantic aware search of dynamic spatial data in a vehicle networking system according to claim 5, wherein the method for generating topic probability lookup tokens is as follows:
query subject vector v by adopting asymmetric scalar product order-preserving encryption algorithm ASPEψAnd encrypting to obtain the theme probability search token.
8. The method for searching for dynamic spatial data in the internet of vehicles according to claim 1, wherein the method for searching for the suspected region set in step 5 is as follows:
and starting searching from a root node of the encryption position index structure, comparing the inner product of the current node and the area search token, determining whether the node comprises a suspected area, and traversing child nodes of the node if the node comprises the suspected area to obtain a suspected area set.
9. The method for searching dynamic spatial data in the internet of vehicles according to claim 1, wherein the method for obtaining the matching set P is as follows:
and judging the geographic position of each node in the suspected area set by using the position search token, if the node meets the searched geographic requirement, calculating the matching degree, performing matching degree score calculation on the theme deviation vector in the node by using the theme probability search token, then sequencing the id of the nodes according to the matching degree score, selecting the id of k nodes before ranking, forming a result set, and returning the result set to the vehicle.
10. The method for secure semantic perception search of dynamic spatial data in the internet of vehicles according to claim 9, wherein in step 6, after the vehicle obtains the result set, clear text information of the result set is requested to the private cloud through an encryption channel, and after the private cloud server receives the result set, clear text information corresponding to id in the result set is searched from the data set D to form a clear text result set G which is sent to the vehicle.
CN202110602922.XA 2021-05-31 2021-05-31 Security semantic perception searching method for dynamic spatial data in Internet of vehicles Active CN113254743B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110602922.XA CN113254743B (en) 2021-05-31 2021-05-31 Security semantic perception searching method for dynamic spatial data in Internet of vehicles

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110602922.XA CN113254743B (en) 2021-05-31 2021-05-31 Security semantic perception searching method for dynamic spatial data in Internet of vehicles

Publications (2)

Publication Number Publication Date
CN113254743A true CN113254743A (en) 2021-08-13
CN113254743B CN113254743B (en) 2022-12-09

Family

ID=77185541

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110602922.XA Active CN113254743B (en) 2021-05-31 2021-05-31 Security semantic perception searching method for dynamic spatial data in Internet of vehicles

Country Status (1)

Country Link
CN (1) CN113254743B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113630250A (en) * 2021-08-23 2021-11-09 济南大学 Model training method and system based on data encryption
CN113962391A (en) * 2021-12-22 2022-01-21 鹏城实验室 Neural network architecture searching method, device, equipment and storage medium

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150262069A1 (en) * 2014-03-11 2015-09-17 Delvv, Inc. Automatic topic and interest based content recommendation system for mobile devices
CN106559422A (en) * 2016-11-10 2017-04-05 西安电子科技大学 Multidimensional ciphertext interval query method based on key agreement
CN108763293A (en) * 2018-04-17 2018-11-06 平安科技（深圳）有限公司 Point of interest querying method, device and computer equipment based on semantic understanding
US20190121873A1 (en) * 2017-10-24 2019-04-25 University Of Louisiana At Lafayette Architecture for semantic search over encrypted data in the cloud
CN111556495A (en) * 2020-03-19 2020-08-18 西安电子科技大学 Multi-user searchable encryption method and encryption system in Internet of vehicles environment
CN112257455A (en) * 2020-10-21 2021-01-22 西安电子科技大学 Semantic-understanding ciphertext space keyword retrieval method and system

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150262069A1 (en) * 2014-03-11 2015-09-17 Delvv, Inc. Automatic topic and interest based content recommendation system for mobile devices
CN106559422A (en) * 2016-11-10 2017-04-05 西安电子科技大学 Multidimensional ciphertext interval query method based on key agreement
US20190121873A1 (en) * 2017-10-24 2019-04-25 University Of Louisiana At Lafayette Architecture for semantic search over encrypted data in the cloud
CN108763293A (en) * 2018-04-17 2018-11-06 平安科技（深圳）有限公司 Point of interest querying method, device and computer equipment based on semantic understanding
CN111556495A (en) * 2020-03-19 2020-08-18 西安电子科技大学 Multi-user searchable encryption method and encryption system in Internet of vehicles environment
CN112257455A (en) * 2020-10-21 2021-01-22 西安电子科技大学 Semantic-understanding ciphertext space keyword retrieval method and system

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
YINBIN MIAO 等: "《privacy-preserving attribute-based keyword search in shared multi-owner setting》", 《IEEE》 *
向广利等: "基于向量空间模型的短文本密文检索方法", 《计算机工程与设计》 *
徐艺丹等: "基于希尔伯特R树和LDA的混合机制研究", 《计算机仿真》 *
戴雪龙: "《面向云环境的语义感知多关键词密文排序检索方法》", 《中国优秀硕士学位论文全文数据库 信息科技辑》 *
鲜学丰等: "基于主题模型的位置感知订阅发布系统", 《计算机科学》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113630250A (en) * 2021-08-23 2021-11-09 济南大学 Model training method and system based on data encryption
CN113630250B (en) * 2021-08-23 2023-04-25 济南大学 Model training method and system based on data encryption
CN113962391A (en) * 2021-12-22 2022-01-21 鹏城实验室 Neural network architecture searching method, device, equipment and storage medium
CN113962391B (en) * 2021-12-22 2022-03-01 鹏城实验室 Neural network architecture searching method, device, equipment and storage medium

Also Published As

Publication number Publication date
CN113254743B (en) 2022-12-09

Similar Documents

Publication Publication Date Title
US20210385236A1 (en) System and method for the automated detection and prediction of online threats
CN106815350B (en) Dynamic ciphertext multi-keyword fuzzy search method in cloud environment
Mouratidis et al. Anonymous query processing in road networks
US11341128B2 (en) Poly-logarithmic range queries on encrypted data
CN113254743B (en) Security semantic perception searching method for dynamic spatial data in Internet of vehicles
Fan et al. Secure and efficient privacy-preserving ciphertext retrieval in connected vehicular cloud computing
US20140090023A1 (en) Method and Apparatus for Authenticating Location-based Services without Compromising Location Privacy
US20070271372A1 (en) Locational tagging in a capture system
US20110197284A1 (en) Attributes of captured objects in a capture system
CN108345802B (en) Safe and efficient ciphertext retrieval method based on vehicle-connected cloud system and vehicle-mounted system
CN109992995B (en) Searchable encryption method supporting location protection and privacy inquiry
CN109063509A (en) It is a kind of that encryption method can search for based on keywords semantics sequence
CN106059988B (en) Method for protecting track privacy based on location-based service
CN103873602A (en) Network resource naming method and generating device
Zhang et al. You are where you have been: Sybil detection via geo-location analysis in OSNs
CN110113338A (en) A kind of encryption traffic characteristic extracting method based on Fusion Features
Zhang et al. An efficient privacy-preserving multi-keyword query scheme in location based services
Alsaawy et al. Triple cache approach for preserving privacy and enhancing performance of LBS
Li et al. Secure semantic-aware search over dynamic spatial data in VANETs
CN107086917A (en) The encryption method that a kind of parallelization and structuring public key can search for
Bai et al. Application behavior identification in DNS tunnels based on spatial-temporal information
CN116069957A (en) Information retrieval method, device and equipment
CN106961386B (en) Location privacy protection method in sign-in service
CN110290149B (en) Location privacy protection method based on C/S multi-level cache
CN110489669B (en) Information pushing method and device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant