CN114944914B - Internet of vehicles data security sharing and privacy protection method based on secret sharing - Google Patents
Internet of vehicles data security sharing and privacy protection method based on secret sharing Download PDFInfo
- Publication number
- CN114944914B CN114944914B CN202210616532.2A CN202210616532A CN114944914B CN 114944914 B CN114944914 B CN 114944914B CN 202210616532 A CN202210616532 A CN 202210616532A CN 114944914 B CN114944914 B CN 114944914B
- Authority
- CN
- China
- Prior art keywords
- service provider
- internet
- shared
- vehicles
- vehicles service
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L9/00—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols
- H04L9/08—Key distribution or management, e.g. generation, sharing or updating, of cryptographic keys or passwords
- H04L9/0816—Key establishment, i.e. cryptographic processes or cryptographic protocols whereby a shared secret becomes available to two or more parties, for subsequent use
- H04L9/085—Secret sharing or secret splitting, e.g. threshold schemes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
-
- 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
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/04—Network architectures or network communication protocols for network security for providing a confidential data exchange among entities communicating through data packet networks
- H04L63/0428—Network architectures or network communication protocols for network security for providing a confidential data exchange among entities communicating through data packet networks wherein the data content is protected, e.g. by encrypting or encapsulating the payload
- H04L63/0435—Network architectures or network communication protocols for network security for providing a confidential data exchange among entities communicating through data packet networks wherein the data content is protected, e.g. by encrypting or encapsulating the payload wherein the sending and receiving network entities apply symmetric encryption, i.e. same key used for encryption and decryption
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Security & Cryptography (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Bioethics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a secret sharing-based internet-of-vehicles data security sharing and privacy protection method, wherein in an internet-of-vehicles system, malicious vehicle users can cause local model parameters to change, so that toxic local model parameters also participate in global model training, and the accuracy of a global model is reduced. To detect and screen out these vehicle users, the present invention calculates a similarity score for each vehicle user, and vehicle users with similarity scores below a set threshold will be rejected to improve the accuracy of the global model. During training, the local model trained each time is divided into two parts and respectively sent to different Internet of vehicles service providers, the two Internet of vehicles service providers do not directly share local model parameters, and privacy of a vehicle user is prevented from being revealed.
Description
Technical Field
The invention relates to the technical field of internet of vehicles data transmission, in particular to an internet of vehicles data security sharing and privacy protection method based on secret sharing.
Background
In privacy preserving research of distributed machine learning, privacy preserving machine learning algorithms based on secure multiparty computing have been available to securely calculate the sum of user parameter updates, which is commonly referred to as secure syndication. Where the desired aggregate result is calculated without displaying any user's parameter updates in plain text. The non-plaintext form can effectively reduce the risk of privacy disclosure caused by parameter interception, but has no resistance to poisoning attack. Malicious users maliciously influence the global model by poisoning the locally trained local model. Common poisoning attacks include tag rollover attacks, back door attacks, and the like. In both attacks, the attacker first participates in learning training, and then deliberately tampers with his own training data to generate a model that incorrectly classifies certain classes or certain inputs into classes selected by the attacker. Several defense mechanisms have been proposed in the prior art to mitigate poisoning attacks. For example, multi-krum can identify users whose parameter updates are greatly different from other users, and screen out the users as malicious users, so that the malicious model accounts for a certain ratio in the models participating in aggregation, and the defending effect is achieved. In the prior art, a privacy-enhanced federal learning scheme based on an addition homomorphic cryptosystem is proposed, which can apply a function to encrypted data without revealing the data value. And a distributed selective SGD method is adopted to realize distributed encryption and reduce communication cost. An authentication mechanism is also incorporated to authenticate the client. The prior art proposes to enhance the security of existing joint transfer learning models in the case of malicious settings, where some participants may deviate arbitrarily from the predefined protocol. They use variants of multiparty computing (MPC) to improve the usability of the system in the presence of malicious clients. In the prior art, the influence of poisoning data and the number of attackers on the distributed poisoning attack performance is analyzed, and a scheme for discarding the poisoning local model in the global model training process is provided. However, all the proposed defensive methods assume that the server can clearly observe the update of the parameters, and therefore this does not take into account the privacy disclosure risk of the participating users.
Disclosure of Invention
Aiming at the defects in the prior art, the method for safely sharing and protecting the data of the Internet of vehicles based on secret sharing solves the problems that privacy of participating users is easy to leak in the existing defense method.
In order to achieve the aim of the invention, the invention adopts the following technical scheme: a method for safely sharing and protecting privacy of internet of vehicles data based on secret sharing comprises the following steps:
s1, acquiring local data of a vehicle user, and constructing a local training data set;
s2, carrying out privacy training on the local model of the vehicle user by adopting a local training data set and global model parameters to obtain a local model for each training;
s3, performing fuzzy processing on the local model trained each time, and dividing the local model into two parts to obtain a shared local model parameter secret of the two parts;
s4, sending a part of the shared local model parameter secret to the Internet of vehicles service provider S1, and sending another part of the shared local model parameter secret to the Internet of vehicles service provider S2;
s5, calculating the similarity according to the secret sharing of local model parameters of the Internet of vehicles service provider S1 and the Internet of vehicles service provider S2;
s6, judging whether the similarity is larger than a threshold value, if so, jumping to the step S7, otherwise, discarding the shared local model parameter secret, and jumping to the step S1;
s7, in the Internet of vehicles service provider S1 and the Internet of vehicles service provider S2, weighting and aggregating the shared local model parameter secrets to obtain a global model;
s8, judging whether the global model converges, if so, obtaining the trained global model, and if not, jumping to the step S2;
s9, sharing the trained global model among the vehicle users, and realizing safe sharing and privacy protection of the Internet of vehicles data.
In summary, the invention has the following beneficial effects:
1. a malicious attacker may organize multiple vehicle users to participate in the same training task in order to break the global model. For example, in the training task of image recognition, certain objects are incorrectly marked as other tags. Thus, these parameter updates may deviate more significantly from those of honest vehicle users. To detect and screen out these vehicle users, the present invention calculates a similarity score for each vehicle user, and vehicle users with similarity scores below a set threshold will be rejected to improve the accuracy of the global model.
2. During training, the local model trained each time is divided into two parts and respectively sent to different Internet of vehicles service providers, the two Internet of vehicles service providers do not directly share local model parameters, and privacy of a vehicle user is prevented from being revealed.
Further, the step S3 includes the following sub-steps:
s31, at 2 l Is assigned to the first part shared local model parameter secret u by taking a random value in the range of (a) 1 Where l is the number of bits of the data of the local model per training;
s32, sharing the local model parameter secret u according to the first portion 1 Through u 2 ＝u-u 1 mod 2 l Where u is the data of the local model for each training, mod is the modulo operation, u 2 The local model parameter secret is shared for the second part.
Further, the step S5 includes the following sub-steps:
s51a, constructing a shared value of a vehicle user p and a vehicle user q according to the shared local model parameter secret, v pq ＝u p ·u q Wherein u is p Shared local model parameter secret for vehicle user p, u q Shared local model parameter secret for vehicle user q, v pq Sharing a value for vehicle user p and vehicle user q;
s52a, calculating a first similar parameter on the Internet of vehicles service provider Si<e> i And a second similar parameter<f> i ；
<e> i ＝<O p > i -<u p > i
<f> i ＝<O q > i -<u q > i
Wherein, the liquid crystal display device comprises a liquid crystal display device,<u p > i 、<u q > i is multiplication triplet v pq ＝u p ·u q Shared secret in internet of vehicles service provider Si, 1 or 2,u p ＝<u p > 1 +<u p > 2 ，u q ＝<u q > 1 +<u q > 2 ，<O p > i For the shared local model parameter vector of the vehicle user p of the set internet of vehicles service provider Si,<O q > i a shared local model parameter vector for a vehicle user q of a set internet of vehicles service provider Si;
s53a, for the first similar parameters<e> i And a second similar parameter<f> i Respectively carrying out reconstruction to obtain a first reconstruction similar parameter e and a second reconstruction similar parameter f;
s54a, calculating an inner product of the vehicle user p and the vehicle user q on the internet of vehicles service provider Si according to the first reconstruction similar parameter e and the second reconstruction similar parameter f:
<O p ·O q > 1 ＝f·<u p > 1 +e·<u q > 1 +<v pq > 1
<O p ·O q > 2 ＝e·f+f·<u p > 2 +e·<u q > 2 +<v pq > 2
wherein, the liquid crystal display device comprises a liquid crystal display device,<O p ·O q > 1 for the inner product of vehicle user p and vehicle user q calculated on the internet of vehicles service provider S1,<O p ·O q > 2 for the inner product of vehicle user p and vehicle user q calculated on the internet of vehicles service provider S2,<u p > 1 to multiply u in triples on a car networking service provider S1 p Is used to determine the shared secret of (1),<u q > 1 to multiply u in triples on a car networking service provider S1 q Is used to determine the shared secret of (1),<v pq > 1 to multiply v in triples on a car networking service provider S1 pq Is used to determine the shared secret of (1),<u p > 2 to multiply u in triples on a car networking service provider S2 p Is used to determine the shared secret of (1),<u q > 2 to multiply triplets on a car networking service provider S2Middle u q Is used to determine the shared secret of (1),<v pq > 2 to multiply v in triples on a car networking service provider S2 pq V pq ＝<v pq > 1 +<v pq > 2 ；
S55a, calculating inner products between the vehicle user p and all other vehicle users on the same Internet of vehicles service provider Si through the methods of the steps S51 to S53, and superposing all inner products to obtain first shared content on the Internet of vehicles service provider S1<ρ p > 1 Second shared content on a car networking service provider S2<ρ p > 2 ；
S56a, the first shared content<ρ p > 1 And a second shared content<ρ p > 2 Reconstructing to obtain a similarity value rho of the vehicle user p p 。
The beneficial effects of the above-mentioned further scheme are: and calculating a similarity value for subsequent investigation of the attacker.
Further, the method for constructing the sharing value of the vehicle user p and the vehicle user q in the step S51a includes the following sub-steps:
a1a, will v pq ＝u p ·u q The construction is as follows:
u p ·u q ＝(<u p > 1 +<u p > 2 )·(<u q > 1 +<u q > 2 )＝<u p > 1 ·<u q > 1 +<u p > 1 ·<u q > 2 +<u p > 2 ·<u q > 1 +<u p > 2 ·<u q > 2
wherein, the liquid crystal display device comprises a liquid crystal display device,<u p > 1 to multiply u in triples on a car networking service provider S1 p Is used to determine the shared secret of (1),<u q > 1 to multiply u in triples on a car networking service provider S1 q Is used to determine the shared secret of (1),<u p > 2 is in the carU in multiplication triplets on networking service provider S2 p Is used to determine the shared secret of (1),<u q > 2 to multiply u in triples on a car networking service provider S2 q Is a shared secret of (a);
a2a, first pair<u p > 1 、<u q > 1 、<u p > 2 And<u q > 2 giving an initial value;
a3a, on the Internet of vehicles service provider S1, calculate<u p > 1 ·<u q > 1 Is calculated at the internet of vehicles service provider S2<u p > 2 ·<u q > 2 Is a value of (2);
a4a, adopting an additive homomorphic encryption method to calculate<u p > 1 ·<u q > 2 And<u p > 2 ·<u q > 1 is a value of (2);
a5a according to<u p > 1 ·<u q > 2 、<u p > 2 ·<u q > 1 、<u p > 1 ·<u q > 1 And<u p > 2 ·<u q > 2 the value is obtained, and the sharing value v of the vehicle user p and the vehicle user q is obtained pq 。
The beneficial effects of the above-mentioned further scheme are: the scheme for exchanging information can reduce a large amount of encryption and decryption calculation, reduce the load of communication and calculation on equipment, and effectively reduce the risk of privacy disclosure.
Further, the step A4a includes the following sub-steps:
a4a1, setting a shared public key pk of the internet of vehicles service provider S1 and the internet of vehicles service provider S2,
a4a2 shared secret on the Internet of vehicles service provider S1 using the shared public key pk<u p > 1 Encrypting to obtain first encrypted content Enc pk (<u p > 1 )；
A4a3, first encryptingCapacity Enc pk (<u p > 1 ) Transmitting to another internet of vehicles service provider S2;
a4a5, first encrypted content Enc received from internet of vehicles service provider S2 pk (<u p > 1 ) And a shared secret on the internet of vehicles service provider S2<u p > 2 Calculate the first content to be decrypted
A4a6, according to the formula
Obtaining first temporary decrypted content Enc pk (<u p > 1 ·<u q > 2 )；
A4a7, corresponding decryption method using symmetric encryption to decrypt the first temporary decryption content Enc pk (<u p > 1 ·<u q > 2 ) Decrypting again to obtain<u p > 1 ·<u q > 2 ；
A4a8 shared secret on the internet of vehicles service provider S2 using the shared public key pk<u p > 2 Encrypting to obtain second encrypted content Enc pk (<u p > 2 )；
A4a9, encrypting the second encrypted content Enc pk (<u p > 2 ) Transmitting to another internet of vehicles service provider S1;
a4a10, second encrypted content Enc received from the internet of vehicles service provider S1 pk (<u p > 2 ) And a shared secret on the internet of vehicles service provider S1<u q > 1 Calculating a second content to be decrypted
A4a11 according to
Obtaining the second temporary decrypted content Enc pk (<u p > 2 ·<u q > 1 )；
A4a12, corresponding decryption method of symmetric encryption is adopted to decrypt the second temporary decryption content Enc pk (<u p > 2 ·<u q > 1 ) Decrypting again to obtain<u p > 2 ·<u q > 1 。
The beneficial effects of the above-mentioned further scheme are: the internet of vehicles service provider S1 and the internet of vehicles service provider S2 encrypt each other when transmitting data, so that the data will not leak.
Further, the step S54a obtains a shared value<v pq > 1 And shared value<v pq > 2 The method of (1) comprises the following sub-steps:
a1b shared local model parameter secrets for vehicle user p on the Internet of vehicles service provider S1<u p > 1 Shared local model parameter secret with vehicle user q<u q > 1 Encrypting to obtain a first encrypted value<u p > 1* And a second encryption value<u q > 1* ；
A2b shared local model parameter secrets to vehicle user p at the Internet of vehicles service provider S2<u p > 2 Shared local model parameter secret with vehicle user q<u q > 2 Encrypting to obtain a third secret value<u p > 2* And a fourth encryption value<u q > 2* ；
A3b, generating an encryption vector R on the Internet of vehicles service provider S2, sending the encryption vector R to the Internet of vehicles service provider S1, and calculating a shared value of the Internet of vehicles service provider S1 on the Internet of vehicles service provider S1<v pq > 1 ；
<v pq > 1 ＝<u p > 1* ·<u q > 1* -R
A4b, first encrypting value<u p > 1* And a second encryption value<u q > 1* Sending to the internet of vehicles service provider S2;
a5b, according to the first encryption value<u p > 1* Second encryption value<u q > 1* Third encryption value<u p > 2* And a fourth encryption value<u q > 2* Calculating the shared value of the Internet of vehicles service provider S2 on the Internet of vehicles service provider S2<v pq > 2 ；
<v pq > 2 ＝<u p > 1* ·<u q > 1* +<u p > 1* ·<u q > 2* +<u p > 2* ·<u q > 1* +R。
The beneficial effects of the above-mentioned further scheme are: by means of the encryption vector R, the problem of privacy leakage during batch sharing of multiplication triples is solved, and the degree of privacy protection is improved.
Further, the step S5 includes the following sub-steps:
s51b, constructing the collected shared local model parameter secrets of all vehicle users on the same Internet of vehicles service provider Si as an N1 secret matrix, wherein N is the number of rows of the secret matrix, namely N shared local model parameter secrets;
s52b, on the same Internet of vehicles service provider Si, calculating the inner product of the N1 secret matrix and the 1*N transposed secret matrix thereof to obtain the N secret matrix;
s53b, superposing each row of data in the N-N secret matrix to obtain shared content of N vehicle users;
s54b, selecting the shared content belonging to the vehicle user p from the shared content of N vehicle users on the same Internet of vehicles service provider Si to obtain a third shared content of the vehicle user p on the Internet of vehicles service provider S1<ρ p ′> 1 Fourth shared content on a car networking service provider S2<ρ p ′> 2 ；
S55b, thirdSharing content<ρ p ′> 1 And fourth shared content<ρ p ′> 2 Reconstructing to obtain a similarity value rho of the vehicle user p p 。
Further, the reconstruction method in step S53a, S56a or S55b includes the following sub-steps:
b1a, on the same Internet of vehicles service provider Si, calculating the shared characterization parameters<z> i ；
<z> i ＝<r> 1 +<r> 2 mod2 l
Wherein, the liquid crystal display device comprises a liquid crystal display device,<z> 1 for the shared characterization parameters of the internet of vehicles service provider S1,<z> 2 for the shared characterization parameters of the internet of vehicles service provider S2,<r> 1 for characterizing step S53a<e> 1 At the same time<r> 2 Characterization in step S53a<e> 2 Or (b)<r> 1 For characterizing step S53a<f> 1 At the same time<r> 2 Characterization in step S53a<f> 2 Or (b)<r> 1 For characterizing step S56a<ρ p > 1 At the same time<r> 2 Characterization step S56a<ρ p > 2 Or (b)<r> 1 For characterizing in step S55b<ρ p ′> 1 At the same time<r> 2 Characterization in step S55b<ρ p ′> 2 L is the number of bits of the data of the local model trained each time;
b2a, to share characterization parameters on the Internet of vehicles service provider Si<z> i Transmitting to another internet of vehicles service provider S (2-i);
b3a, sharing characterization parameters sent by the internet of vehicles service provider Si on another internet of vehicles service provider S (2-i)<z> i Shared characterization parameters computed with the internet of vehicles service provider S (2-i) itself<z> i Adding to obtain a reconstruction value z;
wherein, in<r> 1 For characterizing step S53a<e> 1 At the same time<r> 2 Characterization in step S53a<e> 2 When the reconstruction value z is the first reconstruction similarity parameter e; at the position of<r> 1 For characterizing step S53a<f> 1 At the same time<r> 2 Characterization in step S53a<f> 2 When the reconstruction value z is the second reconstruction similarity parameter f; at the position of<r> 1 For characterizing step S56a<ρ p > 1 At the same time<r> 2 Characterization step S56a<ρ p > 2 The reconstruction value z is the similarity value ρ of the vehicle user p in step S56a p The method comprises the steps of carrying out a first treatment on the surface of the At the position of<r> 1 For characterizing in step S55b<ρ p ′> 1 At the same time<r> 2 Characterization in step S55b<ρ p ′> 2 The reconstructed value z is the similarity value ρ of the vehicle user p in step S55b p 。
Further, the reconstruction method in step S53a, S56a or S55b includes the following sub-steps:
B1B, on the same Internet of vehicles service provider Si, take two random numbers within 2l<a> i And<b> i ；
B2B based on random numbers<a> i And<b> i setting sharing parameters<c> i ；
<c> i ＝<a> i ·<b> i mod2 l
Wherein, l is the bit number of the data of the local model trained each time;
B3B based on random numbers<a> i Random number<b> i Sharing parameters<c> i Calculating a first reconstruction factor<α> i And a second reconstruction factor<β> i ；
<α> i ＝<u>-<a> i
<β> i ＝<v>-<b> i
Wherein, the liquid crystal display device comprises a liquid crystal display device,<u>for characterizing step S53a<e> 1 At the same time<v>Characterization in step S53a<e> 2 Or (b)<u>For characterizing step S53a<f> 1 At the same time<v>Characterization stepIn S53a<f> 2 Or (b)<u>For characterizing step S56a<ρ p > 1 At the same time<v>Characterization step S56a<ρ p > 2 Or (b)<u>For characterizing in step S55b<ρ p ′> 1 At the same time<v>Characterization in step S55b<ρ p ′> 2 ；
B4B, respectively to<α> i And<β> i reconstructing to obtain a first reconstruction coefficient alpha and a second reconstruction coefficient beta;
B5B, calculating the shared characterization parameters on the same Internet of vehicles service provider Si<z′> i ；
<z′> i ＝γ·α·β+β·<a> i +α·<β> i +<c> i
When i=1, the parameter γ=0, and when i=2, the parameter γ=1;
B6B will share characterization parameters on the internet of vehicles service provider Si<z> i Transmitting to another internet of vehicles service provider S (2-i);
B7B, sharing characterization parameters sent by the internet of vehicles service provider Si on another internet of vehicles service provider S (2-i)<z′> i Shared characterization parameters computed with the internet of vehicles service provider S (2-i) itself<z′> i And adding to obtain a reconstruction value z.
Further, the reconstruction method in step S53a, S56a or S55b includes the following sub-steps:
b1c, on the Internet of vehicles service provider Si, parameters are set<r> 1 Converting to a binary format;
b2c, carrying out multi-round circulation on another Internet of vehicles service provider S (2-i), carrying out multi-round careless transmission in the circulation process, and outputting (-a) when the t-th round careless transmission is carried out t，0 ，a t，1 ) To a car networking service provider Si, where a t，0 Is a random number, a t，1 ＝2 t ·<r> 1 -a t，0 ，a t，1 Is an intermediate variable;
b3c, on-vehicle trainOn the network service provider Si, a plurality of rounds of looping are performed, and at the time of the t-th round of looping, a (-a) is received t，0 ，a t，1 )；
B4c (-a) according to the reception of the t-th round t，0 ，a t，1 ) Calculating shared characterization parameters on a car networking service provider Si<z″> i ；
Wherein T is the total round of circulation, r < T >, and]is a parameter<r> 2 Is a group consisting of a (c) and (d) in the (c) th bit,<r> 1 for characterizing step S53a<e> 1 At the same time<r> 2 Characterization in step S53a<e> 2 Or (b)<r> 1 For characterizing step S53a<f> 1 At the same time<r> 2 Characterization in step S53a<f> 2 Or (b)<r> 1 For characterizing step S56a<ρ p > 1 At the same time<r> 2 Characterization step S56a<ρ p > 2 Or (b)<r> 1 For characterizing in step S55b<ρ p ′> 1 At the same time<r> 2 Characterization in step S55b<ρ p ′> 2 ；
B5c, calculating shared characterization parameters on another Internet of vehicles service provider S (2-i)<z″> 2-i ；
B6c to share characterization parameters on the Internet of vehicles service provider Si<z″> i Transmitting to another internet of vehicles service provider S (2-i);
b7c sharing characterization parameters sent by the internet of vehicles service provider Si on another internet of vehicles service provider S (2-i)<z″> i Shared characterization parameters computed with the internet of vehicles service provider S (2-i) itself<z″> 2-i And adding to obtain a reconstruction value z.
The beneficial effects of the above-mentioned further scheme are: the two internet of vehicles service providers do not directly model parameters, so that the privacy of the vehicle user is prevented from being revealed.
Drawings
FIG. 1 is an application scenario diagram of the present invention;
fig. 2 is a flowchart of a method for secure sharing and privacy protection of internet of vehicles data based on secret sharing.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and all the inventions which make use of the inventive concept are protected by the spirit and scope of the present invention as defined and defined in the appended claims to those skilled in the art.
As shown in fig. 1, in the application scenario of the present invention, a malicious vehicle user may cause a local model parameter to change, so that a toxic local model parameter also participates in global model training, and the accuracy of the global model is reduced.
Vehicle users include honest and malicious. They all participate in the training of the local model, and the sharing of the update model is completed according to the same scheme. However, malicious users tamper with their own data, and their trained local models may compromise the convergence of the global model, or even cause divergence, so as to threaten the security of data sharing in the system.
The following technical scheme is adopted in the embodiment to improve the accuracy of the global model and avoid privacy disclosure of a vehicle user.
As shown in fig. 2, a method for secure sharing and privacy protection of internet of vehicles data based on secret sharing includes the following steps:
s1, acquiring local data of a vehicle user, and constructing a local training data set;
s2, carrying out privacy training on the local model of the vehicle user by adopting a local training data set and global model parameters to obtain a local model for each training;
s3, performing fuzzy processing on the local model trained each time, and dividing the local model into two parts to obtain a shared local model parameter secret of the two parts;
the step S3 comprises the following sub-steps:
s31, taking random value in 2l range and assigning value to the first part shared local model parameter secret u 1 Where l is the number of bits of the data of the local model per training;
s32, sharing the local model parameter secret u according to the first portion 1 Through u 2 ＝u-u 1 mod2 l Where u is the data of the local model for each training, mod is the modulo operation, u 2 The local model parameter secret is shared for the second part.
S4, sending a part of the shared local model parameter secret to the Internet of vehicles service provider S1, and sending another part of the shared local model parameter secret to the Internet of vehicles service provider S2;
in the present embodiment, for the vehicle user p, if the first part shares the local model parameter secret u 1 Sent to the internet of vehicle service provider S1, the first part shares the local model parameter secret u 1 Shared local model parameter secrets for vehicle user p on a car networking service provider S1<u p > 1 The method comprises the steps of carrying out a first treatment on the surface of the If the second part shares the local model parameter secret u 2 Sent to the internet of vehicle service provider S2, the second part shares the local model parameter secret u 2 Shared local model parameter secrets for vehicle user p on a car networking service provider S2<u p > 2 。
S5, calculating the similarity according to the secret sharing of local model parameters of the Internet of vehicles service provider S1 and the Internet of vehicles service provider S2;
by using the similarity to calculate the distance between parameter updates, parameter updates that deviate too much from the vehicle user will be discriminated as malicious users. The shared local model parameter secret may be discarded, thereby guaranteeing the accuracy of the global model.
Step S5 includes two implementation methods, the first one:
the step S5 includes the following sub-steps:
s51a, constructing a shared value of a vehicle user p and a vehicle user q according to the shared local model parameter secret, v pq ＝u p ·u q Wherein u is p Shared local model parameter secret for vehicle user p, u q Shared local model parameter secret for vehicle user q, v pq Sharing a value for vehicle user p and vehicle user q;
the method for constructing the shared value of the vehicle user p and the vehicle user q in the step S51a comprises the following steps:
a1a, will v pq ＝u p ·u q The construction is as follows:
u p ·u q ＝(<u p > 1 +<u p > 2 )·(<u q > 1 +<u q > 2 )
＝<u p > 1 ·<u q > 1 +<u p > 1 ·<u q > 2 +<u p > 2 ·<u q > 1 +<u p > 2 ·<u q > 2
wherein, the liquid crystal display device comprises a liquid crystal display device,<u p > 1 to multiply u in triples on a car networking service provider S1 p Is used to determine the shared secret of (1),<u q > 1 to multiply u in triples on a car networking service provider S1 q Is used to determine the shared secret of (1),<u p > 2 to multiply u in triples on a car networking service provider S2 p Is used to determine the shared secret of (1),<u q > 2 to multiply u in triples on a car networking service provider S2 q Is a shared secret of (a);
a2a, first pair<u p > 1 、<u q > 1 、<u p > 2 And<u q > 2 giving an initial value;
a3a, on the Internet of vehicles service provider S1, calculate<u p > 1 ·<u q > 1 Is calculated at the internet of vehicles service provider S2<u p > 2 ·<u q > 2 Is a value of (2);
a4a, adopting an additive homomorphic encryption method to calculate<u p > 1 ·<u q > 2 And<u p > 2 ·<u q > 1 is a value of (2);
the step A4a comprises the following sub-steps:
a4a1, setting a shared public key pk of the internet of vehicles service provider S1 and the internet of vehicles service provider S2,
a4a2 shared secret on the Internet of vehicles service provider S1 using the shared public key pk<u p > 1 Encrypting to obtain first encrypted content Enc pk (<u p > 1 )；
A4a3, encrypting the first encrypted content Enc pk (<u p > 1 ) Transmitting to another internet of vehicles service provider S2;
a4a5, first encrypted content Enc received from internet of vehicles service provider S2 pk (<u p > 1 ) And a shared secret on the internet of vehicles service provider S2<u p > 2 Calculate the first content to be decrypted
A4a6, according to the formula
Obtaining first temporary decrypted content Enc pk (<u p > 1 ·<u q > 2 )；
A4a7, corresponding decryption method using symmetric encryption to decrypt the first temporary decryption content Enc pk (<u p > 1 ·<u q > 2 ) Decrypting again to obtain<u p > 1 ·<u q > 2 ；
A4a8、Shared secret on a car networking service provider S2 using a shared public key pk<u p > 2 Encrypting to obtain second encrypted content Enc pk (<u p > 2 )；
A4a9, encrypting the second encrypted content Enc pk (<u p > 2 ) Transmitting to another internet of vehicles service provider S1;
a4a10, second encrypted content Enc received from the internet of vehicles service provider S1 pk (<u p > 2 ) And a shared secret on the internet of vehicles service provider S1<u q > 1 Calculating a second content to be decrypted
A4a11 according to
Obtaining the second temporary decrypted content Enc pk (<u p > 2 ·<u q > 1 )；
A4a12, corresponding decryption method of symmetric encryption is adopted to decrypt the second temporary decryption content Enc pk (<u p > 2 ·<u q > 1 ) Decrypting again to obtain<u p > 2 ·<u q > 1 。
A5a according to<u p > 1 ·<u q > 2 、<u p > 2 ·<u q > 1 、<u p > 1 ·<u q > 1 And<u p > 2 ·<u q > 2 the value is obtained, and the sharing value v of the vehicle user p and the vehicle user q is obtained pq 。
S52a, calculating a first similar parameter on the Internet of vehicles service provider Si<e> i And a second similar parameter<f> i ；
<e> i ＝<O p > i -<u p > i
<f> i ＝<O q > i -<u q > i
Wherein, the liquid crystal display device comprises a liquid crystal display device,<u p > i 、<u q > i is multiplication triplet v pq ＝u p ·u q Shared secret in internet of vehicles service provider Si, 1 or 2,u p ＝<u p > 1 +<u p > 2 ，u q ＝<u q > 1 +<u q > 2 ，<O p >i is a shared local model parameter vector for the vehicle user p of the set internet of vehicles service provider Si,<O q > i a shared local model parameter vector for a vehicle user q of a set internet of vehicles service provider Si;
in step S52a, a first similar parameter is calculated<e> i And a second similar parameter<f> i Two forms are included, the first:
<e> 1 ＝<O p > 1 -<u p > 1
<f> 1 ＝<O q > 1 -<u q > 1
in step S52a, a first similar parameter is calculated<e> i And a second similar parameter<f> i Two forms are included, the second:
<e> 2 ＝<O p > 2 -<u p > 2
<f> 2 ＝<O q > 2 -<u q > 2
s53a, for the first similar parameters<e> i And a second similar parameter<f> i Respectively carrying out reconstruction to obtain a first reconstruction similar parameter e and a second reconstruction similar parameter f;
s54a, calculating an inner product of the vehicle user p and the vehicle user q on the internet of vehicles service provider Si according to the first reconstruction similar parameter e and the second reconstruction similar parameter f:
<O p ·O q > 1 ＝f·<u p > 1 +e·<u q > 1 +<v pq > 1
<O p ·O q > 2 ＝e·f+f·<u p > 2 +e·<u q > 2 +<v pq > 2
wherein, the liquid crystal display device comprises a liquid crystal display device,<O p ·O q > 1 for the inner product of vehicle user p and vehicle user q calculated on the internet of vehicles service provider S1,<O p ·O q > 2 for the inner product of vehicle user p and vehicle user q calculated on the internet of vehicles service provider S2,<u p > 1 to multiply u in triples on a car networking service provider S1 p Is used to determine the shared secret of (1),<u q > 1 to multiply u in triples on a car networking service provider S1 q Is used to determine the shared secret of (1),<v pq > 1 to multiply v in triples on a car networking service provider S1 pq Is used to determine the shared secret of (1),<u p > 2 to multiply u in triples on a car networking service provider S2 p Is used to determine the shared secret of (1),<u q > 2 to multiply u in triples on a car networking service provider S2 q Is used to determine the shared secret of (1),<v pq > 2 to multiply v in triples on a car networking service provider S2 pq V pq ＝<v pq > 1 +<v pq > 2 ；
S55a, calculating inner products between the vehicle user p and all other vehicle users on the same Internet of vehicles service provider Si through the methods of the steps S51 to S53, and superposing all inner products to obtain first shared content on the Internet of vehicles service provider S1<ρ p > 1 Second shared content on a car networking service provider S2<ρ p > 2 ；
S56a, the first shared content<ρ p > 1 And a second shared content<ρ p > 2 Reconstructing to obtain similarity value of vehicle user pρ p 。
Step S5 includes two implementation methods, the second:
the step S5 includes the following sub-steps:
s51b, constructing the collected shared local model parameter secrets of all vehicle users on the same Internet of vehicles service provider Si as an N1 secret matrix, wherein N is the number of rows of the secret matrix, namely N shared local model parameter secrets;
s52b, on the same Internet of vehicles service provider Si, calculating the inner product of the N1 secret matrix and the 1*N transposed secret matrix thereof to obtain the N secret matrix;
s53b, superposing each row of data in the N-N secret matrix to obtain shared content of N vehicle users;
s54b, selecting the shared content belonging to the vehicle user p from the shared content of N vehicle users on the same Internet of vehicles service provider Si to obtain a third shared content of the vehicle user p on the Internet of vehicles service provider S1<ρ p ′> 1 Fourth shared content on a car networking service provider S2<ρ p ′> 2 ；
S55b, third shared content<ρ p ′> 1 And fourth shared content<ρ p ′> 2 Reconstructing to obtain a similarity value rho of the vehicle user p p 。
S6, judging whether the similarity is larger than a threshold value, if so, jumping to the step S7, otherwise, discarding the shared local model parameter secret, and jumping to the step S1;
s7, in the Internet of vehicles service provider S1 and the Internet of vehicles service provider S2, weighting and aggregating the shared local model parameter secrets to obtain a global model;
s8, judging whether the global model converges, if so, obtaining the trained global model, and if not, jumping to the step S2;
s9, sharing the trained global model among the vehicle users, and realizing safe sharing and privacy protection of the Internet of vehicles data.
The global model in the present invention is understood to be a data processor, which itself is derived by training the data. In the past, when the vehicle p wanted to obtain one piece of relatively complete information, it was necessary to obtain data of other vehicles and then analyze all the data itself to obtain useful information. The global model can only take own data as one input, and available information is output through the global model. In the process, direct sharing of data is avoided, and the problem of privacy leakage is avoided.
The reconstruction method in step S53a, S56a or S55b includes three implementation methods, the first one:
b1a, on the same Internet of vehicles service provider Si, calculating the shared characterization parameters<z> i ；
<z> i ＝<r> 1 +<r> 2 mod2 l
Wherein, the liquid crystal display device comprises a liquid crystal display device,<z> 1 for the shared characterization parameters of the internet of vehicles service provider S1,<z> 2 for the shared characterization parameters of the internet of vehicles service provider S2,<r> 1 for characterizing step S53a<e> 1 At the same time<r> 2 Characterization in step S53a<e> 2 Or (b)<r> 1 For characterizing step S53a<f> 1 At the same time<r> 2 Characterization in step S53a<f> 2 Or (b)<r> 1 For characterizing step S56a<ρ p > 1 At the same time<r> 2 Characterization step S56a<ρ p > 2 Or (b)<r> 1 For characterizing in step S55b<ρ p ′> 1 At the same time<r> 2 Characterization in step S55b<ρ p ′> 2 L is the number of bits of the data of the local model trained each time;
b2a, to share characterization parameters on the Internet of vehicles service provider Si<z> i Transmitting to another internet of vehicles service provider S (2-i);
b3a, sharing characterization parameters sent by the internet of vehicles service provider Si on another internet of vehicles service provider S (2-i)<z> i Shared characterization parameters computed with the internet of vehicles service provider S (2-i) itself<z> i Adding to obtain a reconstruction value z;
wherein, in<r> 1 For characterizing step S53a<e> 1 At the same time<r> 2 Characterization in step S53a<e> 2 When the reconstruction value z is the first reconstruction similarity parameter e; at the position of<r> 1 For characterizing step S53a<f> 1 At the same time<r> 2 Characterization in step S53a<f> 2 When the reconstruction value z is the second reconstruction similarity parameter f; at the position of<r> 1 For characterizing step S56a<ρ p > 1 At the same time<r> 2 Characterization step S56a<ρ p > 2 The reconstruction value z is the similarity value ρ of the vehicle user p in step S56a p The method comprises the steps of carrying out a first treatment on the surface of the At the position of<r> 1 For characterizing in step S55b<ρ p ′> 1 At the same time<r> 2 Characterization in step S55b<ρ p ′> 2 The reconstructed value z is the similarity value ρ of the vehicle user p in step S55b p 。
The reconstruction method in the step S53a, S56a or S55b includes three implementation methods, the second one:
B1B, on the same Internet of vehicles service provider Si, at 2 l Taking two random numbers in range<a> i And<b> i ；
B2B based on random numbers<a> i And<b> i setting sharing parameters<c> i ；
<c> i ＝<a> i ·<b> i mod2 l
Wherein, l is the bit number of the data of the local model trained each time;
B3B based on random numbers<a> i Random number<b> i Sharing parameters<c> i Calculating a first reconstruction factor<α> i And a second reconstruction factor<β> i ；
<α> i ＝<u>-<a> i
<β> i ＝<v>-<b> i
Wherein, the liquid crystal display device comprises a liquid crystal display device,<u>for characterizing step S53a<e> 1 At the same time<v>Characterization in step S53a<e> 2 Or (b)<u>For characterizing step S53a<f> 1 At the same time<v>Characterization in step S53a<f> 2 Or (b)<u>For characterizing step S56a<ρ p > 1 At the same time<v>Characterization step S56a<ρ p > 2 Or (b)<u>For characterizing in step S55b<ρ p ′> 1 At the same time<v>Characterization in step S55b<ρ p ′> 2 ；
B4B, respectively to<α> i And<β> i reconstructing to obtain a first reconstruction coefficient alpha and a second reconstruction coefficient beta;
B5B, calculating the shared characterization parameters on the same Internet of vehicles service provider Si<z′> i ；
<z′> i ＝γ·α·β+β·<a> i +α·<β> i +<c> i
When i=1, the parameter γ=0, and when i=2, the parameter γ=1;
B6B will share characterization parameters on the internet of vehicles service provider Si<z> i Transmitting to another internet of vehicles service provider S (2-i);
B7B, sharing characterization parameters sent by the internet of vehicles service provider Si on another internet of vehicles service provider S (2-i)<z′> i Shared characterization parameters computed with the internet of vehicles service provider S (2-i) itself<z′> i And adding to obtain a reconstruction value z.
The reconstruction method in the step S53a, S56a or S55b includes three implementation methods, the third one:
b1c, on the Internet of vehicles service provider Si, parameters are set<r> 1 Converting to a binary format;
b2c, carrying out multi-round circulation on another Internet of vehicles service provider S (2-i), wherein the circulation process is carried out for multiple roundsIntentionally transmitted, and output (-a) when the t-th round is unintentionally transmitted t，0 ，a t，1 ) To a car networking service provider Si, where a t，0 Is a random number, a t，1 ＝2 t ·<r> 1 -a t，0 ，a t，1 Is an intermediate variable;
B3 c on the Internet of vehicles service provider Si, a plurality of rounds of circulation are carried out, and at the time of the t-th round of circulation, (-a) is received t，0 ，a t，1 )；
B4c (-a) according to the reception of the t-th round t，0 ，a t，1 ) Calculating shared characterization parameters on a car networking service provider Si<z″> i ；
Wherein T is the total round of circulation, r < T >, and]is a parameter<r> 2 Is a group consisting of a (c) and (d) in the (c) th bit,<r> 1 for characterizing step S53a<e> 1 At the same time<r> 2 Characterization in step S53a<e> 2 Or (b)<r> 1 For characterizing step S53a<f> 1 At the same time<r> 2 Characterization in step S53a<f> 2 Or (b)<r> 1 For characterizing step S56a<ρ p > 1 At the same time<r> 2 Characterization step S56a<ρ p > 2 Or (b)<r> 1 For characterizing in step S55b<ρ p ′> 1 At the same time<r> 2 Characterization in step S55b<ρ p ′> 2 ；
B5c, calculating shared characterization parameters on another Internet of vehicles service provider S (2-i)<z″> 2-i ；
B6c to share characterization parameters on the Internet of vehicles service provider Si<z″> i Send to anotherA vehicle networking service provider S (2-i);
b7c sharing characterization parameters sent by the internet of vehicles service provider Si on another internet of vehicles service provider S (2-i)<z″> i Shared characterization parameters computed with the internet of vehicles service provider S (2-i) itself<z″> 2-i And adding to obtain a reconstruction value z.
Claims (6)
1. The internet of vehicles data safety sharing and privacy protecting method based on secret sharing is characterized by comprising the following steps:
s1, acquiring local data of a vehicle user, and constructing a local training data set;
s2, carrying out privacy training on the local model of the vehicle user by adopting a local training data set and global model parameters to obtain a local model for each training;
s3, performing fuzzy processing on the local model trained each time, and dividing the local model into two parts to obtain a shared local model parameter secret of the two parts;
s4, sending a part of the shared local model parameter secret to the Internet of vehicles service provider S1, and sending another part of the shared local model parameter secret to the Internet of vehicles service provider S2;
s5, calculating the similarity according to the secret sharing of local model parameters of the Internet of vehicles service provider S1 and the Internet of vehicles service provider S2;
s6, judging whether the similarity is larger than a threshold value, if so, jumping to the step S7, otherwise, discarding the shared local model parameter secret, and jumping to the step S1;
s7, in the Internet of vehicles service provider S1 and the Internet of vehicles service provider S2, weighting and aggregating the shared local model parameter secrets to obtain a global model;
s8, judging whether the global model converges, if so, obtaining the trained global model, and if not, jumping to the step S2;
s9, sharing the trained global model among the vehicle users to realize safe sharing and privacy protection of the vehicle networking data;
the step S3 comprises the following sub-steps:
s31, at the collection
Is a random value within the range of (1) to form a first partially shared local model parameter secret u 1 Where l is the number of bits of the data of the local model per training;
s32, sharing the local model parameter secret u according to the first portion 1 Through u 2 ＝u-u 1 mod2 l Where u is the data of the local model for each training, mod is the modulo operation, u 2 Sharing the local model parameter secret for the second portion;
the step S5 includes the following sub-steps:
s51a, constructing a shared value of a vehicle user p and a vehicle user q according to the shared local model parameter secret, v pq ＝u p ·u q Wherein u is p Shared local model parameter secret for vehicle user p, u q Shared local model parameter secret for vehicle user q, v pq Sharing a value for vehicle user p and vehicle user q;
s52a, calculating a first similar parameter < e > on the Internet of vehicles service provider Si i And a second similarity parameter < f > i ；
＜e＞ i ＝＜O p ＞ i -＜u p ＞ i
＜f＞ i ＝＜O q ＞ i -＜u q ＞ i
Wherein < u p ＞ i 、＜u q ＞ i Is multiplication triplet v pq ＝u p ·u q Shared secret in internet of vehicles service provider Si, 1 or 2,u p ＝＜u p ＞ 1 +＜u p ＞ 2 ，u q ＝＜u q ＞ 1 +＜u q ＞ 2 ，＜O p ＞ i Shared local model parameter vector for vehicle user p of set-up internet of vehicles service provider Si, < O q ＞ i A shared local model parameter vector for a vehicle user q of a set internet of vehicles service provider Si;
s53a, for the first similarity parameter < e > i And a second similarity parameter < f > i Respectively carrying out reconstruction to obtain a first reconstruction similar parameter e and a second reconstruction similar parameter f;
s54a, calculating an inner product of the vehicle user p and the vehicle user q on the internet of vehicles service provider Si according to the first reconstruction similar parameter e and the second reconstruction similar parameter f:
＜O p ·O q ＞ 1 ＝f·＜u p ＞ 1 +e·＜u q ＞ 1 +＜v pq ＞ 1
＜O p ·O q ＞ 2 ＝e·f+f·＜u p ＞ 2 +e·＜u q ＞ 2 +＜v pq ＞ 2
wherein < O p ·O q ＞ 1 For the inner product of vehicle user p and vehicle user q calculated on the internet of vehicles service provider S1, < O p ·O q ＞ 2 For the inner product of vehicle user p and vehicle user q calculated on the internet of vehicles service provider S2, < u p ＞ 1 To multiply u in triples on a car networking service provider S1 p Is less than u q ＞ 1 To multiply u in triples on a car networking service provider S1 q Is less than v pq ＞ 1 To multiply v in triples on a car networking service provider S1 pq Is less than u p ＞ 2 To multiply u in triples on a car networking service provider S2 p Is less than u q ＞ 2 To multiply u in triples on a car networking service provider S2 q Is less than v pq ＞ 2 To multiply v in triples on a car networking service provider S2 pq V pq ＝＜v pq ＞ 1 +＜v pq ＞ 2 ；
S55a, on the same Internet of vehicles service provider Si, passing through the parties of steps S51 to S53Calculating the inner products between the vehicle user p and all other vehicle users by a method, and superposing all the inner products to obtain the first shared content < rho on the Internet of vehicles service provider S1 p ＞ 1 Second shared content < ρ on internet of vehicles service provider S2 p ＞ 2 ；
S56a, the first shared content is less than ρ p ＞ 1 And second shared content < ρ p ＞ 2 Reconstructing to obtain a similarity value rho of the vehicle user p p The method comprises the steps of carrying out a first treatment on the surface of the Or (b)
The step S5 includes the following sub-steps:
s51b, constructing the collected shared local model parameter secrets of all vehicle users on the same Internet of vehicles service provider Si as an N1 secret matrix, wherein N is the number of rows of the secret matrix, namely N shared local model parameter secrets;
s52b, on the same Internet of vehicles service provider Si, calculating the inner product of the N1 secret matrix and the 1*N transposed secret matrix thereof to obtain the N secret matrix;
s53b, superposing each row of data in the N-N secret matrix to obtain shared content of N vehicle users;
s54b, selecting the shared content belonging to the vehicle user p from the shared content of N vehicle users on the same Internet of vehicles service provider Si to obtain a third shared content < ρ of the vehicle user p on the Internet of vehicles service provider S1 p ′＞ 1 Fourth shared content < ρ on internet of vehicles service provider S2 p ′＞ 2 ；
S55b, the third shared content is less than ρ p ′＞ 1 And fourth shared content < ρ p ′＞ 2 Reconstructing to obtain a similarity value rho of the vehicle user p p 。
2. The method for securely sharing and protecting the privacy of internet of vehicles data based on secret sharing according to claim 1, wherein the method for constructing the sharing value between the vehicle user p and the vehicle user q in the step S51a comprises the following sub-steps:
a1a, will v pq ＝u p ·u q The construction is as follows:
u p ·u q ＝(＜u p ＞ 1 +＜u p ＞ 2 )·(＜u q ＞ 1 +＜u q ＞ 2 )
＝＜u p ＞ 1 ·＜u q ＞ 1 +＜u p ＞ 1 ·＜u q ＞ 2 +＜u p ＞ 2 ·＜u q ＞ 1 +＜u p ＞ 2 ·＜u q ＞ 2
wherein < u p ＞ 1 To multiply u in triples on a car networking service provider S1 p Is less than u q ＞ 1 To multiply u in triples on a car networking service provider S1 q Is less than u p ＞ 2 To multiply u in triples on a car networking service provider S2 p Is less than u q ＞ 2 To multiply u in triples on a car networking service provider S2 q Is a shared secret of (a);
a2a, first pair < u p ＞ 1 、＜u q ＞ 1 、＜u p ＞ 2 Sum < u q ＞ 2 Giving an initial value;
a3a, on the Internet of vehicles service provider S1, calculate < u p ＞ 1 ·＜u q ＞ 1 Is calculated to < u on the internet of vehicles service provider S2 p ＞ 2 ·＜u q ＞ 2 Is a value of (2);
a4a, adopting additive homomorphic encryption method to calculate < u p ＞ 1 ·＜u q ＞ 2 Sum < u p ＞ 2 ·＜u q ＞ 1 Is a value of (2);
a5a is according to < u p ＞ 1 ·＜u q ＞ 2 、＜u p ＞ 2 ·＜u q ＞ 1 、＜u p ＞ 1 ·＜u q ＞ 1 Sum < u p ＞ 2 ·＜u q ＞ 2 The value is obtained, and the sharing value v of the vehicle user p and the vehicle user q is obtained pq 。
3. The method for secure sharing and privacy protection of internet of vehicles data based on secret sharing according to claim 2, wherein the step A4a comprises the following sub-steps:
a4a1, setting a shared public key pk of the internet of vehicles service provider S1 and the internet of vehicles service provider S2,
a4a2 shared secret < u on the Internet of vehicles service provider S1 using the shared public key pk p ＞ 1 Encrypting to obtain first encrypted content Enc pk (＜u p ＞ 1 )；
A4a3, encrypting the first encrypted content Enc pk (＜u p ＞ 1 ) Transmitting to another internet of vehicles service provider S2;
a4a5, first encrypted content Enc received from internet of vehicles service provider S2 pk (＜u p ＞ 1 ) And a shared secret < u on the internet of vehicles service provider S2 p ＞ 2 Calculate the first content to be decrypted
A4a6, according to the formula
Obtaining first temporary decrypted content Enc pk (＜u p ＞ 1 ·＜u q ＞ 2 )；
A4a7, corresponding decryption method using symmetric encryption to decrypt the first temporary decryption content Enc pk (＜u p ＞ 1 ·＜u q ＞ 2 ) Decrypting again to get < u p ＞ 1 ·＜u q ＞ 2 ；
A4a8 shared secret < u on the Internet of vehicles service provider S2 using the shared public key pk p ＞ 2 Encrypting to obtain second encrypted content Enc pk (＜u p ＞ 2 )；
A4a9, encrypting the second encrypted content Enc pk (＜u p ＞ 2 ) Transmitting to another internet of vehicles service provider S1;
a4a10, second encrypted content Enc received from the internet of vehicles service provider S1 pk (＜u p ＞ 2 ) And a shared secret < u on the internet of vehicles service provider S1 q ＞ 1 Calculating a second content to be decrypted
A4a11 according to
Obtaining the second temporary decrypted content Enc pk (＜u p ＞ 2 ·＜u q ＞ 1 )；
A4a12, corresponding decryption method of symmetric encryption is adopted to decrypt the second temporary decryption content Enc pk (＜u p ＞ 2 ·＜u q ＞ 1 ) Decrypting again to obtain < u p ＞ 2 ·＜u q ＞ 1 。
4. The method for secure sharing and privacy protection of internet of vehicles data based on secret sharing according to claim 1, wherein the reconstruction method in step S53a, S56a or S55b comprises the following sub-steps:
b1a, on the same Internet of vehicles service provider Si, calculating a shared characterization parameter < z > i ；
＜z＞ i ＝＜r＞ 1 +＜r＞ 2 mod2 l
Wherein < z > 1 For the shared characterization parameters of the Internet of vehicles service provider S1, < z > 2 For the shared characterization parameters of the Internet of vehicles service provider S2, < r > 1 For characterizing < e > in step S53a 1 At the same time < r > 2 Characterization step S53a < e > 2 Or < r >, of 1 For characterizing step S53a＜f＞ 1 At the same time < r > 2 Characterization step S53a < f > 2 Or < r >, of 1 For characterizing < ρ in step S56a p ＞ 1 At the same time < r > 2 Characterization step S56a < ρ p ＞ 2 Or < r >, of 1 For characterizing < ρ in step S55b p ′＞ 1 At the same time < r > 2 Characterization step S55b < ρ p ′＞ 2 L is the number of bits of the data of the local model trained each time;
b2a sharing characterization parameters < z > on the Internet of vehicles service provider Si i Transmitting to another internet of vehicles service provider S (2-i);
b3a, transmitting the shared characterization parameter < z > from the internet of vehicles service provider Si on another internet of vehicles service provider S (2-i) i Shared characterization parameters < z > calculated with the internet of vehicles service provider S (2-i) itself i Adding to obtain a reconstruction value z;
wherein, at < r > 1 For characterizing < e > in step S53a 1 At the same time < r > 2 Characterization step S53a < e > 2 When the reconstruction value z is the first reconstruction similarity parameter e; at < r > 1 For characterizing < f > in step S53a 1 At the same time < r > 2 Characterization step S53a < f > 2 When the reconstruction value z is the second reconstruction similarity parameter f; at < r > 1 For characterizing < ρ in step S56a p ＞ 1 At the same time < r > 2 Characterization step S56a < ρ p ＞ 2 The reconstruction value z is the similarity value ρ of the vehicle user p in step S56a p The method comprises the steps of carrying out a first treatment on the surface of the At < r > 1 For characterizing < ρ in step S55b p ′＞ 1 At the same time < r > 2 Characterization step S55b < ρ p ′＞ 2 The reconstructed value z is the similarity value ρ of the vehicle user p in step S55b p 。
5. The method for secure sharing and privacy protection of internet of vehicles data based on secret sharing according to claim 1, wherein the reconstruction method in step S53a, S56a or S55b comprises the following sub-steps:
B1B, on the same Internet of vehicles service provider Si, at 2 l Taking two random numbers < a > -in the range i Sum < b > i ；
B2b, according to random number < a > i Sum < b > i Setting sharing parameter < c >, and i ；
＜c＞ i ＝＜a＞ i ·＜b＞ i mod2 l
wherein, l is the bit number of the data of the local model trained each time;
B3B is based on the random number < a > i Random number < b > i And sharing parameter < c > i Calculate the first reconstruction factor < alpha > i And a second reconstruction factor < beta > i ；
＜α＞ i ＝＜u＞-＜a＞ i
＜β＞ i ＝＜v＞-＜b＞ i
Wherein < u > is used to characterize < e > in step S53a 1 At the same time < v > characterizes < e > in step S53a 2 Or < u > is used to characterize < f > in step S53a 1 At the same time < v > characterizes < f > in step S53a 2 Or < u > is used to characterize < ρ in step S56a p ＞ 1 At the same time < v > characterizes < ρ in step S56a p ＞ 2 Or < u > is used to characterize < ρ in step S55b p ′＞ 1 At the same time < v > characterizes < ρ in step S55b p ′＞ 2 ；
B4B for < alpha >, respectively i Sum < beta >, and i reconstructing to obtain a first reconstruction coefficient alpha and a second reconstruction coefficient beta;
B5B, calculating a sharing characterization parameter < z' >, on the same Internet of vehicles service provider Si i ；
＜z′＞ i ＝γ·α·β+β·＜a＞ i +α·＜β＞ i +＜c＞ i
When i=1, the parameter γ=0, and when i=2, the parameter γ=1;
B6B, in-car networking service providerThe characterization parameters < z > -will be shared on Si i Transmitting to another internet of vehicles service provider S (2-i);
B7B transmitting the shared characterizing parameters < z' >, transmitted by the internet of vehicles service provider Si, on another internet of vehicles service provider S (2-i) i Shared characterization parameters < z' >, calculated by the internet of vehicles service provider S (2-i) itself i And adding to obtain a reconstruction value z.
6. The method for secure sharing and privacy protection of internet of vehicles data based on secret sharing according to claim 1, wherein the reconstruction method in step S53a, S56a or S55b comprises the following sub-steps:
b1c, on the Internet of vehicles service provider Si, parameter < r > 1 Converting to a binary format;
b2c, carrying out multi-round circulation on another Internet of vehicles service provider S (2-i), carrying out multi-round careless transmission in the circulation process, and outputting (-a) when the t-th round careless transmission is carried out t，0 ，a t，1 ) To a car networking service provider Si, where a t，0 Is a random number, a t，1 ＝2 t ·＜r＞ 1 -a t，0 ，a t，1 Is an intermediate variable;
b3c, carrying out multi-round circulation on the Internet of vehicles service provider Si, and receiving (-a) when the t th round circulation is carried out t,0 ，a t,1 )；
B4c (-a) according to the reception of the t-th round t，0 ，a t，1 ) Calculating a shared characterization parameter < z > on a vehicle networking service provider Si i ；
Wherein T is the total round of circulation, r < T >, and]for parameter < r > 2 T < r > 1 For characterizing < e > in step S53a 1 At the same time < r > 2 Characterization step S53a < e > 2 Or < r >, of 1 For characterizing step S53aMiddle < f >, a combination of 1 At the same time < r > 2 Characterization step S53a < f > 2 Or < r >, of 1 For characterizing < ρ in step S56a p ＞ 1 At the same time < r > 2 Characterization step S56a < ρ p ＞ 2 Or < r >, of 1 For characterizing < ρ in step S55b p ′＞ 1 At the same time < r > 2 Characterization step S55b < ρ p ′＞ 2 ；
B5c calculating the shared characterizing parameter < z' >, on another Internet of vehicles service provider S (2-i) 2-i ；
B6c sharing characterization parameters < z' >, on the Internet of vehicles service provider Si i Transmitting to another internet of vehicles service provider S (2-i);
b7c transmitting the shared characterization parameter < z' >, which is transmitted by the internet of vehicles service provider Si, on another internet of vehicles service provider S (2-i) i Shared characterization parameters < z ">, calculated by the internet of vehicles service provider S (2-i) itself 2-i And adding to obtain a reconstruction value z.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210616532.2A CN114944914B (en) | 2022-06-01 | 2022-06-01 | Internet of vehicles data security sharing and privacy protection method based on secret sharing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210616532.2A CN114944914B (en) | 2022-06-01 | 2022-06-01 | Internet of vehicles data security sharing and privacy protection method based on secret sharing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114944914A CN114944914A (en) | 2022-08-26 |
| CN114944914B true CN114944914B (en) | 2023-06-02 |
Family
ID=82909084
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210616532.2A Active CN114944914B (en) | 2022-06-01 | 2022-06-01 | Internet of vehicles data security sharing and privacy protection method based on secret sharing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114944914B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112770291A (en) * | 2021-01-14 | 2021-05-07 | 华东师范大学 | Distributed intrusion detection method and system based on federal learning and trust evaluation |
| CN113987551A (en) * | 2021-11-22 | 2022-01-28 | 湖南科技学院 | Self-adaptive federal learning method for Internet of vehicles |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11323249B2 (en) * | 2017-12-20 | 2022-05-03 | Lg Electronics, Inc. | Cryptographic methods and systems for authentication in connected vehicle systems and for other uses |
| CN112367305B (en) * | 2020-10-22 | 2022-05-20 | 香港中文大学（深圳） | Internet of vehicles protection method based on privacy block chain and mobile terminal |
| CN112583575B (en) * | 2020-12-04 | 2023-05-09 | 华侨大学 | Federal learning privacy protection method based on homomorphic encryption in Internet of vehicles |
| CN113612598B (en) * | 2021-08-02 | 2024-02-23 | 北京邮电大学 | Internet of vehicles data sharing system and method based on secret sharing and federal learning |
| CN114492833A (en) * | 2021-12-29 | 2022-05-13 | 上海智能网联汽车技术中心有限公司 | Internet of vehicles federal learning layered knowledge safe migration method based on gradient memory |
-
2022
- 2022-06-01 CN CN202210616532.2A patent/CN114944914B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112770291A (en) * | 2021-01-14 | 2021-05-07 | 华东师范大学 | Distributed intrusion detection method and system based on federal learning and trust evaluation |
| CN113987551A (en) * | 2021-11-22 | 2022-01-28 | 湖南科技学院 | Self-adaptive federal learning method for Internet of vehicles |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114944914A (en) | 2022-08-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107124268B (en) | Privacy set intersection calculation method capable of resisting malicious attacks | |
| CN112217626B (en) | Network threat cooperative defense system and method based on intelligence sharing | |
| Xue et al. | Security improvement on an anonymous key agreement protocol based on chaotic maps | |
| CN101977112B (en) | Public key cipher encrypting and decrypting method based on neural network chaotic attractor | |
| CN109474928A (en) | Realize that the true value of efficient secret protection finds method in mobile gunz sensory perceptual system | |
| Kumar et al. | Comparative study on AES and RSA algorithm for medical images | |
| CN105187425B (en) | Facing cloud calculus communication system safety without certificate thresholding decryption method | |
| DE102012111903A1 (en) | Method for establishing a secure connection between clients | |
| CA2639649A1 (en) | Cryptography method and system | |
| Diffie et al. | New Directions in cryptography (1976) | |
| Ahmad et al. | Multi-objective evolution of strong S-boxes using non-dominated sorting genetic algorithm-II and chaos for secure telemedicine | |
| CN116167088A (en) | Method, system and terminal for privacy protection in two-party federal learning | |
| Es-Sabry et al. | Securing images using high dimensional chaotic maps and DNA encoding techniques | |
| El-Zoghdy et al. | How good is the DES algorithm in image ciphering | |
| CN113132104A (en) | Active and safe ECDSA (electronic signature SA) digital signature two-party generation method | |
| CN114239862A (en) | anti-Byzantine attack federal learning method for protecting user data privacy | |
| CN114944914B (en) | Internet of vehicles data security sharing and privacy protection method based on secret sharing | |
| CN117134945A (en) | Data processing method, system, device, computer equipment and storage medium | |
| Sejad et al. | Proposal of a DH optimization model | |
| CN114900283A (en) | Deep learning user gradient aggregation method based on multi-party security calculation | |
| Fu et al. | A fast chaos-based colour image encryption algorithm using a hash function | |
| Gunasekaran et al. | Encrypting and decrypting image using computer visualization techniques | |
| Rabea et al. | Survey Image Cryptanalysis Using a Substitution Box Based Chaotic Map | |
| Li et al. | PSFL: Ensuring Data Privacy and Model Security for Federated Learning | |
| Ghosh et al. | An Analysis of Chaos-based Cryptographic Algorithms |
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 |