A data privacy protection method based on data segmentation under mobile crowd sensing

Abstract

The application discloses a kind of mobile crowd sensing based on data segmentation data privacy protection method, it is related to data privacy protection field.The present application is to solve the problem of low security of existing data segmentation based privacy protection method.The present application is specifically as follows: the sensing user who wants to participate in task applies for registration to sensing platform, sensing platform receives sensing user registration request and allocates two pseudonyms for each sensing user, sensing user applies for task to sensing platform using pseudonym, sensing platform verifies the legitimacy of sensing user applying for task, and allocates task and collaboration user trust list for legitimate sensing user;Legitimate sensing user selects collaboration user for sensing data transmission according to collaboration user trust list;Sensing user applying for task and collaboration user selected in step three interact to complete sensing data upload.The present application is used for private data transmission.

Description

Technical Field

[0001] This invention relates to the field of data privacy protection, and in particular to a data privacy protection method based on data segmentation under mobile crowd sensing. Background Technology

[0002] In the field of data privacy protection, data segmentation technology protects personal data privacy by dividing datasets into multiple parts and processing and analyzing them at different nodes. With the increasing number of mobile crowdsourcing applications, users' demand for privacy protection is also growing rapidly, especially regarding the privacy and security of users during data interaction.

[0003] In mobile crowdsensing, classic privacy protection methods based on data segmentation protect privacy by isolating data through fragmentation, effectively ensuring the integrity of the original data. However, traditional data segmentation-based privacy protection methods suffer from low data transmission path reliability, making them vulnerable to background knowledge attacks and Sybil attacks. Attackers can leverage the correlation between the perceived user's identity information and the perceived data to further infer user privacy, thus threatening the security of the perceived user's private data. Therefore, existing data segmentation-based privacy protection methods still suffer from low security issues. Summary of the Invention

[0004] The purpose of this invention is to address the low security issues of existing data segmentation-based privacy protection methods, and to propose a data privacy protection method based on data segmentation under mobile crowd sensing.

[0005] A data privacy protection method based on data segmentation under mobile crowd sensing, specifically:

[0006] Step 1: Perception users who want to participate in the task apply for registration on the perception platform. Upon receiving the registration request, the perception platform assigns two pseudonyms to each user. The user then uses the pseudonyms to apply for the task. The perception platform verifies the legitimacy of the user applying for the task and assigns the task and a trust list of collaborating users to the legitimate user.

[0007] Step 2: Legitimate sensing users select collaborating users for sensing data transmission based on the trusted list of collaborating users;

[0008] Step 3: The user applying for the task interacts with the collaborating user selected in Step 2 to complete the upload of the perception data.

[0009] Furthermore, in step one, the sensing platform receives a user registration request and assigns two pseudonyms to each user, specifically:

[0010] U i1 =list[n1 mod len(list)]

[0011] U i2 =list[n2 mod len(list)]

[0012] Among them, U i1 U i2 User U i The two aliases are n1 and n2, which are random numbers. list is a list of aliases, i∈[1,n], n is the number of users who are aware of the aliases, len(list) is the length of list, and mod is the modulo operation.

[0013] Furthermore, in step one, the sensing user applies for a task from the sensing platform using a pseudonym. The sensing platform verifies the legitimacy of the sensing user applying for the task, specifically as follows:

[0014] First, the sensing platform obtains the tasks submitted by the task publishers and forms a task set Task = {task1, task2, ..., task...} m’};

[0015] Where m′ is the number of tasks, and Task is the set of tasks submitted by all requesters;

[0016] Then, the perception platform broadcasts the task set Task = {task1, task2, ..., task...} to all perception users. m’ Users can apply to the perception platform to participate in tasks using any pseudonym.

[0017] Finally, the perception platform determines whether each perception user applying to participate in the task uses two aliases to apply for the same task. If the current perception user applies for the task using only one alias, then the current perception user is a legitimate perception user and a task is assigned to the legitimate perception user. If the current perception user applies for the same task using two aliases, then the current perception user is an illegitimate perception user and a task is not assigned to the illegitimate perception user.

[0018] Furthermore, the trusted list of collaborating users in step one is obtained in the following way:

[0019] First, obtain the trust value for each perceived user:

[0020]

[0021] Among them, T i It is a user-sensing U i Trust value, m i It is a user-sensing U i Number of tasks completed, f i It is a user-sensing U i Returns the number of correctly calculated results, si It is a user-sensing U i Returns the number of incorrect calculation results;

[0022] Then, users with trust values ​​greater than a preset trust threshold are identified, and these users are sorted from highest to lowest trust value and stored in the collaborative user trust list.

[0023] Furthermore, in step two, the legitimate sensing user selects the collaborating user for sensing data transmission based on the collaborating user trust list, specifically as follows:

[0024] Step 21: Designate two users with message association as an explicit user pair, and two users without message association as an implicit user pair.

[0025] Step 22: Obtain the association weights between perceived users who have an explicit relationship:

[0026]

[0027] Among them, J(U i′ U j′ ) is a user U that is associated with messages. i′ and U j′ The correlation weight between them It is a user-sensing U i′ Total number of interactions during the task. It is a user-sensing U j′ Total number of interactions during the task. User U with message association i′ and U j′ The number of interactions between them, i′, j′∈[1, n′], where n′ is the total number of users who are associated with messages;

[0028] Steps 2 and 3: Obtain the perceived non-interaction weights between users who do not have message associations;

[0029] Step 24: Select collaborative users for sensing data transmission using the association weights between sensing users with message associations, the non-interaction weights between sensing users without message associations, and the collaborative user trust list.

[0030] Furthermore, the non-interaction weights between perceived users who do not have message associations in steps two and three are obtained in the following way:

[0031] First, obtain the latitude and longitude value sets of the sensing users, and normalize the latitude and longitude values ​​of each sensing user:

[0032]

[0033] Among them, lng(n a )=(a1,a2,…,a n ) is the set of longitude values ​​perceived by the user, lat(n) b )=(b1,b2,…,b n ) is the set of dimension values ​​for perceiving the user, a′ i It is the normalized perceived user U i The longitude value, b′ i It is the normalized perceived user U i The latitude value, a i It is a hidden user U i The longitude value, b i It is a hidden user U i The latitude value, min() is the minimum value in the set, max() is the maximum value in the set, a n It is a user-sensing U n The longitude value, b n It is a user-sensing U n The latitude value;

[0034] Then, obtain the probability density function of the latitude and longitude values ​​of the latent user relative to the perceived user:

[0035]

[0036] Where, k′ i* It is to take a′ i* or b′ i* μ is the mean, σ is the standard deviation, and pdf(a′) i* ) is the user perception U i* Longitude probability density function, pdf(b′) i* ) is the user perception U i* Dimensional probability density function, i*∈[1,n*], where n* is the total number of perceptual users without message association, U i* These are users who are not associated with any messages;

[0037] Finally, the non-interaction weights between perceived users who do not have message associations are obtained by utilizing the distance between them:

[0038]

[0039] p i* =pdf(a′ i* )

[0040] q i* =pdf(b′) i* )

[0041] Among them, JS(U i*U j* ) is a user U who does not have message association. i* U j* No interaction weights between them It is a user-sensing U i* U j* The association weights between them, i*∈[1,n*], where n* is the total number of users who perceive the message without association, and p i* It is a user-sensing U i* The longitude probability density value, q i* It is a user-sensing U i* The latitudinal probability density values, P={p1,p2,…,p n*}, where P is the longitude probability density matrix, and Q = {q1, q2, ..., q n*}, Q is the latitudinal probability density matrix, M is the average matrix of matrices P and Q, D JS (P||Q) is the similarity value between P and Q, D KL (P||M) is the difference between P and M, D KL (Q||M) is the difference between Q and M.

[0042] Furthermore, in step two-four, the selection of collaborative users for sensing data transmission using the association weights between sensing users with message associations, the non-interaction weights between sensing users without message associations, and the collaborative user trust list specifically involves:

[0043] Sort explicit user pairs in descending order of association weight between perceived users with message association, and compare the association weight of explicit user pairs with a first threshold, saving explicit user pairs with association weights greater than the first threshold.

[0044] The hidden user pairs are sorted from largest to smallest according to the non-interaction weight between perceived users who have no message association, and the non-interaction weight of the hidden user pairs is compared with the second threshold. The hidden user pairs corresponding to the non-interaction weights that are greater than the second threshold are saved.

[0045] The stored explicit user pairs, implicit user pairs, and the perceived users in the collaborative user trust list are combined to form a collaborative user selection dataset. A preset number of perceived users are randomly selected from the collaborative user selection dataset as collaborative users for the perception data transmission.

[0046] Furthermore, in step three, the sensing user applying for the task interacts with the collaborating user selected in step two to complete the uploading of sensing data, specifically as follows:

[0047] Step 3: First, designate the users who applied for the task as distributors and the collaborating users obtained in Step 2 as verifiers.

[0048] Step 32: The distributor selects prime numbers p' and q', satisfying q'|p'-1, selects generator g, and uses t as the threshold value for secret reconstruction;

[0049] Where p' is a prime number with a length greater than or equal to 512 bits, q' is a prime number with a length greater than or equal to 160 bits, and p' > q'. It is a multiplicative group modulo p'. And the order of g is q', and q'|p'-1 is q' divides p'-1;

[0050] Step 3: The distributor randomly selects t-1 elements a”1, a”2, ..., a” t-1 And using random numbers and a secret S, a polynomial of degree t-1 is constructed: f(x) = a”0 + a”1x + a”2x 2 +…+a” t-1 x t-1 ;

[0051] in, a”0=S, f(x) is the secret polynomial, and x is the identity information variable of the verifier;

[0052] Steps 3 and 4: The distributor calculates the validator's collaboration weight value based on the validator's trust value.

[0053]

[0054] Where, β, It is a constant. Verifier Trust value, It is the number of validators. It is the validator U i The collaboration weight value;

[0055] Step 35: The distributor is for each validator. Assign an identity information Then calculate the sub-secret based on the verifier's identity. The distributor distributes the sub-secret values ​​in descending order of the validator's collaboration weight value. Distribute the information to each corresponding verifier; then the distributor calculates the verification information. And verification information The broadcast is sent to each validator;

[0056] in, It is a random number;

[0057] Step 36: After receiving the sub-secret and verification information, the verifier uses the verification information to verify whether the sub-secret is correct. If the sub-secret is correct, it is sent to the perception platform; otherwise, it is not sent.

[0058] Step 37: After the sensing platform receives the sub-secret, it calculates the sum of the collaboration weights of the verifiers who sent the sub-secret. If the sum of the collaboration weights is greater than or equal to t, the secret polynomial is reconstructed to obtain the reconstructed secret; otherwise, secret reconstruction is not performed.

[0059] Furthermore, the step three-six, which involves verifying the correctness of the sub-secret using verification information, specifically involves:

[0060] If the following equation holds true, then the subsecret is correct; otherwise, the subsecret is incorrect:

[0061]

[0062] in, It represents the total number of validators.

[0063] Furthermore, the reconstruction secret polynomial in step three-seven obtains the reconstruction secret, specifically as follows:

[0064] First, reconstruct the secret polynomial:

[0065]

[0066] Where l′ and l* are the sub-secret labels received by the sensing platform, and r is the total number of sub-secrets received by the sensing platform. It is a sub-secret received by the sensing platform. x l* It is the verifier's identity information. It is a parameter variable;

[0067] Then, let Obtain the reconstructed secret S = f(0).

[0068] The beneficial effects of this invention are as follows:

[0069] This invention proposes a data segmentation-based method for protecting the privacy of sensory data. In a mobile crowdsourcing sensing environment, this invention uses data segmentation and user trust values ​​to assign weights to sensing users, protecting their privacy and making data transmission more secure. Furthermore, this invention proposes a method for selecting trusted collaborating users. It utilizes message relationships between sensing users and combines user trust values ​​obtained from evaluating data quality to construct a set of collaborating users. The collected privacy-preserving sensory data is uploaded using a secret sharing method to hide the original data, restricting low-weight collaborating users from reconstructing the sensing data and preventing attackers from colluding to obtain the sensing data. Simultaneously, sensing users use different identities to collect and upload sensing data, protecting the identity information of sensing users and the relationships between sensing data, thus enhancing the security of the privacy protection method. Attached Figure Description

[0070] Figure 1 This is a framework diagram of the present invention. Detailed Implementation

[0071] Specific implementation method one: as follows Figure 1 As shown, the specific process of a data privacy protection method based on data segmentation under mobile crowd sensing in this embodiment is as follows:

[0072] Step 1: Users who wish to participate in the task apply for registration on the Perception Platform. Upon receiving the registration request, the Perception Platform assigns two aliases to each user. Users then use these aliases to apply for tasks. The Perception Platform verifies the legitimacy of the users applying for tasks and assigns tasks and a trusted list of collaborating users to legitimate users.

[0073] Step 11: Users who wish to participate in the task apply for registration on the perception platform. Upon receiving the registration request, the perception platform assigns two pseudonyms to each user, specifically:

[0074] U i1 =list[n1 mod len(list)]

[0075] U i2 =list[n2 mod len(list)]

[0076] Among them, U i1 U i2 User U i The two aliases, n1 and n2 are random numbers, list is a list of aliases, i∈[1,n], n is the number of users who are aware of the aliases, len(list) is the length of the list, and mod is the modulo operation;

[0077] In this step, the Task Management Center (TC) of the perception platform generates a list of pseudonyms (list), and generates two random numbers (n1 and n2) using the random number generation function (srand(x)). Based on the existing list of pseudonyms, the random numbers are mapped to pseudonyms to obtain the user U. i The two pseudonyms; the perception platform includes: a task management center (TC) and a data management center (DC); the task management center is used to receive task requests, verify the legitimacy of perception users, complete task allocation, and generate a trust list of collaborative users; the data management center is used to obtain user trust values, receive privacy data uploaded by collaborative users, and reconstruct privacy data.

[0078] Steps 1 and 2: The user applies for a task using a pseudonym on the sensing platform, and the sensing platform verifies the legitimacy of the user applying for the task.

[0079] First, the perception platform obtains a task set consisting of all tasks submitted by task publishers: Task = {task1, task2, ..., task...} m’};

[0080] Where m′ is the number of tasks, and Task is the set of tasks submitted by all requesters;

[0081] Then, the perception platform broadcasts the task set Task = {task1, task2, ..., task...} to all perception users. m’ Users can apply to the perception platform to participate in tasks using any pseudonym.

[0082] Finally, the perception platform determines whether each perception user applying to participate in the task uses two pseudonyms to apply for the same task. If the current perception user only uses one pseudonym to apply for the task, then the current perception user is a legitimate perception user and a task is assigned to the legitimate perception user. If the current perception user uses two pseudonyms to apply for the same task, then the current perception user is an illegitimate perception user and a task is not assigned to the illegitimate perception user.

[0083] Step 13: The perception platform issues a list of trusted collaborating users to each perception user assigned a task.

[0084] The trusted list of collaborating users is obtained through the following methods:

[0085] First, obtain the trust value for each perceived user:

[0086]

[0087] Among them, T i It is a user-sensing U i Trust value, m i It is a user-sensing U i Number of tasks completed, fi It is a user-sensing U i Returns the number of correctly calculated results, s i It is a user-sensing U i Returns the number of incorrect calculation results;

[0088] The correct calculation result is the perceived user U. i It can return accurate results when participating in tasks, such as returning accurate temperature values ​​in temperature measurement tasks.

[0089] Then, users with a trust value greater than the preset trust threshold of 0.5 are identified, and these users are sorted from largest to smallest trust value and stored in the collaborative user trust list.

[0090] Step 2: The authorized sensing user selects the collaborating user for sensing data transmission based on the task and the trusted list of collaborating users. Specifically:

[0091] Step 21: Designate two user pairs with message association as explicit user pairs, and two user pairs with no message association as implicit user pairs.

[0092] There is an explicit relationship between explicit users and perceived users, and an implicit relationship between implicit users and implicit users.

[0093] Step 22: Calculate the association weights between explicit user pairs:

[0094]

[0095] Among them, J(U i' U j' ) is the explicit user's view on U i' and U j' The correlation weight between them It is a user-sensing U i' Total number of interactions during the task. It is a user-sensing U j' Total number of interactions during the task. It is explicit user to U i' and U j' The number of interactions between them, i',j'∈[1,n'], where n' is the total number of users who perceive the relationship with an explicit one;

[0096] Steps 2 and 3: Obtain the non-interaction weights of implicit user pairs:

[0097] First, obtain the latitude and longitude value sets of the sensing users, and normalize the latitude and longitude values ​​of each sensing user:

[0098]

[0099] Among them, lng(n a )=(a1,a2,…,a n ) is the set of longitude values ​​perceived by the user, lat(n) b )=(b1,b2,…,b n ) is the set of dimension values ​​for perceiving the user, a′ i It is the normalized perceived user U i The longitude value, b′ i It is the normalized perceived user U i The latitude value, a i It is a hidden user U i The longitude value, b i It is a hidden user U i The latitude value, min() is the minimum value in the set, and max() is the maximum value in the set;

[0100] Then, obtain the latent user pairs, and obtain the probability density function of the latitude and longitude values ​​of the perceived users in the latent user pairs:

[0101]

[0102] Where, k′ i* It is to take a′ i* or b′ i* μ = 0.5 is the mean, σ = 0.1 is the standard deviation, and pdf(a′) i* ) is the user perception U i* Longitude probability density function, pdf(b′) i* ) is the user perception U i* The dimensional probability density function, i*∈[1,n*], where n* is the total number of users with implicit relationships;

[0103] The distance between the probability density functions of the latitude and longitude values ​​of latent users and perceived users is:

[0104]

[0105] Among them, D i*j* Is it a latent user versus a perceived user U? i* U j* The distance between the probability density functions of latitude and longitude values, pdf(a' j* ) is the user perception U j* Longitude probability density function, pdf(b') j* ) is the user perception U j* Latitude probability density function, D i*j* User U is a user with a latent relationship. i* and U j* The distance between them, i*, j*∈[1,n*];

[0106] Finally, the distance between users with implicit relationships is used to obtain the non-interaction weights between them:

[0107]

[0108] p i* =pdf(a′ i* )

[0109] q i* =pdf(b′) i* )

[0110] Among them, JS(U i* U j* () is the user U who perceives a implicit relationship. i* U j* No interaction weights between them It is a user-sensing U i* U j* The correlation weight between them, p i* It is a user-sensing U i* The probability density value of longitude, q i* It is a user-sensing U i* The probability density value of the latitude value, P={p1,p2,…,p n*}, where P is the longitude probability density matrix, and Q = {q1, q2, ..., q n*}, Q is the latitudinal probability density matrix, M is the average matrix of matrices P and Q, D JS (P||Q) is the similarity value between P and Q, D KL (P||M) is the difference between P and M, D KL (Q||M) is the difference between Q and M;

[0111] The range of JS divergence is [0,1], with 0 for the same value and 1 for different values.

[0112] Step 24: Select collaborative users for data transmission awareness using the association weights between explicit user pairs, the non-interaction weights of implicit user pairs, and the trust list of collaborative users. Specifically:

[0113] The association weights between explicit user pairs are sorted from largest to smallest, and explicit user pairs with association weights greater than the first threshold are saved.

[0114] The non-interaction weights of latent user pairs are sorted from largest to smallest, and latent user pairs with non-interaction weights greater than the second threshold are saved.

[0115] The saved explicit user pairs, implicit user pairs, and users in the collaborative user trust list are combined to form a collaborative user selection dataset. A preset number of users are randomly selected from the collaborative user selection dataset as collaborative users for perceptual data transmission.

[0116] Step 3: The user requesting the task interacts with the collaborating user selected in Step 2 to complete the upload of the sensing data. Specifically:

[0117] Step 3: First, designate the users who applied for the task as distributors and the collaborating users obtained in Step 2 as verifiers.

[0118] Step 3.2: The distributor selects prime numbers p' and q', satisfying q'|p'-1, and selects generators. And the order of g is q', and the threshold value t for secret reconstruction is set;

[0119] Where p' > q', p' is a prime number with a length greater than or equal to 512 bits, and q' is a prime number with a length greater than or equal to 160 bits. It is the multiplication group modulo p', which contains all integers coprime to p'. These integers form a group in the multiplication operation modulo p', and q'|p'-1 means that q' divides p'-1.

[0120] Step 3: The distributor secretly and randomly selects t-1 elements. Together with the secret S, construct a polynomial of degree t-1: f(x) = a”0 + a”1x + a”2x 2 +…+a” t-1 x t-1 ;

[0121] Where a”0=S, f(x) is the secret polynomial, and x is the identity information variable of the verifier;

[0122] Steps 3 and 4: The distributor calculates the validator's collaboration weight value based on the validator's trust value.

[0123]

[0124] in, It is a constant. It is to perceive the user Trust value, It is the number of validators. Verifier The collaboration weight value;

[0125] Step 35: The distributor is for each validator. Assign an identity information This identity information is typically a unique identifier used to distinguish different verifiers during the secret sharing process. The distributor then calculates the sub-secret value based on the verifier's identity information. The sub-secret value will be distributed according to the verifier's collaboration weight value in descending order. Distribute to each corresponding verifier; the distributor calculates the verification information. in, And broadcast verification information;

[0126] Step 36: After receiving the sub-secret from the distributor, the verifier uses the verification information to verify whether the sub-secret is correct. If the sub-secret is correct, it sends the sub-secret to the DC of the perception platform; otherwise, it does not send the sub-secret.

[0127]

[0128] in, This is the total number of validators; (This step first obtains the number of validators) Get again ).

[0129] If true, it proves that the members received the correct share and the distributor was honest; if false, it proves that the distributor was dishonest and there was fraud during the secret distribution phase.

[0130] Step 37: After the DC of the sensing platform receives the sub-secret, it calculates the sum of the collaboration weights of the verifiers who sent the sub-secret. If the sum of the collaboration weights reaches the weight threshold, the secret polynomial is reconstructed using the Lagrange interpolation formula. Obtain the reconstructed secret S = f(0); if the secret share is incorrect, stop the secret reconstruction:

[0131] The sum of the sub-secret weights received by the DC of the perception platform reaches the weight threshold, as shown in the following formula:

[0132]

[0133] Where t is the weight threshold and r is the total number of sub-secrets received by the perception platform;

[0134] The Secret Polynomial of Reconstruction As shown in the following formula:

[0135] in, l* is the received sub-secret identifier. It is a sub-secret received by the sensing platform. x l* It is the verifier's identity information. It is a parameter variable;

[0136] make Obtain the reconstructed secret S = f(0).

Claims

1. A data privacy protection method based on data segmentation under mobile crowd sensing, characterized in that... The specific process of the method is as follows: Step 1: Perception users who want to participate in the task apply for registration on the perception platform. Upon receiving the registration request, the perception platform assigns two pseudonyms to each user. The user then uses the pseudonyms to apply for the task. The perception platform verifies the legitimacy of the user applying for the task and assigns the task and a trust list of collaborating users to the legitimate user. The trusted list of collaborating users is obtained through the following method: First, obtain the trust value for each perceived user: in, It is to perceive the user Trust value, It is to perceive the user Number of tasks completed It is to perceive the user Returns the number of correctly calculated results. It is to perceive the user Returns the number of incorrect calculation results; Then, users with trust values ​​greater than a preset trust threshold are identified, and these users are sorted from highest to lowest trust value and stored in the collaborative user trust list. Step 2: The legitimate sensing user selects the sensing data transmission collaborating user from the collaborating user trust list, specifically as follows: Step 21: Treat two users with message association as explicit user pairs and two users without message association as implicit user pairs. Step 22: Obtain the association weights between users who have message connections: in, Users with message association are aware of the user. and The correlation weight between them It is to perceive the user Total number of interactions during the task. It is to perceive the user Total number of interactions during the task. Users with message association are aware of the user. and Number of interactions between them , It is the total number of users who are associated with messages; Steps 2 and 3: Obtain the perceived non-interaction weights between users who have no message association; Step 24: Select collaborative users for sensing data transmission using the association weights between sensing users with message associations, the non-interaction weights between sensing users without message associations, and the collaborative user trust list. Step 3: The user applying for the task interacts with the collaborating user selected in Step 2 to complete the upload of the perception data.

2. The data privacy protection method based on data segmentation under mobile crowd sensing as described in claim 1, characterized in that: In step one, the sensing platform receives a user registration request and assigns two pseudonyms to each user, specifically: in, User Two kana characters, It is a random number. It is a list of kana characters. , where n is the number of perceived users. yes The length of the value, mod is the modulo operation.

3. The data privacy protection method based on data segmentation under mobile crowd sensing according to claim 2, characterized in that: In step one, the sensing user applies for a task from the sensing platform using a pseudonym. The sensing platform verifies the legitimacy of the sensing user applying for the task, specifically as follows: First, the sensing platform obtains the tasks submitted by task publishers and assembles them into a task set. ; Where m' is the number of tasks. It is the collection of tasks submitted by all requesters; Then, the perception platform broadcasts the task set to all perception users. Users can apply to the perception platform to participate in tasks using any pseudonym. Finally, the perception platform determines whether each perception user applying to participate in the task uses two aliases to apply for the same task. If the current perception user applies for the task using only one alias, then the current perception user is a legitimate perception user and a task is assigned to the legitimate perception user. If the current perception user applies for the same task using two aliases, then the current perception user is an illegitimate perception user and a task is not assigned to the illegitimate perception user.

4. A data privacy protection method based on data segmentation under mobile crowd sensing as described in claim 3, characterized in that: The non-interaction weights between perceived users who have no message association, obtained in steps two and three, are obtained as follows: First, obtain the latitude and longitude value sets of the sensing users, and normalize the latitude and longitude values ​​of each sensing user: in, It is a set of longitude values ​​for sensing users. It is a set of dimensions for perceiving users. It is the normalized perception of the user The longitude value, It is the normalized perception of the user latitude value, Hidden users The longitude value, Hidden users latitude value, It is the minimum value in the set. It is the maximum value in the set. It is to perceive the user The longitude value, It is to perceive the user The latitude value; Then, obtain the probability density function of the latitude and longitude values ​​of the latent user relative to the perceived user: in, Is to take or , It is the mean. It is the standard deviation. It is to perceive the user Longitude probability density function It is to perceive the user Latitude probability density function , This is the total number of users who are aware of the message association. These are users who are not associated with any messages; Finally, the non-interaction weights between perceived users who have no message association are obtained by utilizing the distance between them: in, Users who are not associated with messages , No interaction weights between them It is to perceive the user , The correlation weight between them , This is the total number of users who are aware of the message association. It is to perceive the user The longitude probability density value, It is to perceive the user The latitudinal probability density value, , It is the longitude probability density matrix. , It is the latitudinal probability density matrix. It is a matrix sum matrix The average matrix, yes and Similarity value between them yes and The difference between them yes and The difference between them.

5. A data privacy protection method based on data segmentation under mobile crowd sensing according to claim 4, characterized in that: The step two in step four, which utilizes the association weights between sensing users with message associations, the non-interaction weights between sensing users without message associations, and the trust list of collaborating users to select collaborating users for sensing data transmission, specifically involves: Sort explicit user pairs in descending order of association weight among users with message association, compare the association weight of explicit user pairs with a first threshold, and save explicit user pairs with association weights greater than the first threshold. The hidden user pairs are sorted from largest to smallest according to the non-interaction weight between perceived users who have no message association, and the non-interaction weight of the hidden user pairs is compared with the second threshold. The hidden user pairs corresponding to the non-interaction weights that are greater than the second threshold are saved. The stored explicit user pairs, implicit user pairs, and the perceived users in the collaborative user trust list are combined to form a collaborative user selection dataset. A preset number of perceived users are randomly selected from the collaborative user selection dataset as collaborative users for the perception data transmission.

6. A data privacy protection method based on data segmentation under mobile crowd sensing according to claim 5, characterized in that: In step three, the user applying for the task interacts with the collaborating user selected in step two to upload the sensing data. Specifically: Step 3: First, designate the users who applied for the task as distributors and the collaborating users obtained in Step 2 as verifiers. Step 3.2: The distributor selects a prime number. , And satisfy Select generator ,Will As a threshold value for secret reconstruction; in, A prime number with a length greater than or equal to 512 bits. It is a prime number with a length greater than or equal to 160 bits. , It is a model Multiplication group below and The order is Step, yes Divisible ; Step 33: Distributors are randomly selected. element And using random numbers and secrets Together we build Polynomial of degree: ; in, , , It is a secret polynomial. It is a variable representing the identity information of the verifier; Steps 3 and 4: The distributor calculates the validator's collaboration weight value based on the validator's trust value. in, It is a constant. Verifier Trust value, , It is the number of validators. Verifier The collaboration weight value; Step 35: The distributor is for each validator. Assign an identity information Then calculate the sub-secret based on the verifier's identity. The distributor distributes the sub-secret values ​​in descending order of the validator's collaboration weight value. Distribute the information to each corresponding verifier; then the distributor calculates the verification information. and verification information The broadcast is sent to each validator; in, , It is a random number; Step 36: After receiving the sub-secret and verification information, the verifier uses the verification information to verify whether the sub-secret is correct. If the sub-secret is correct, it is sent to the perception platform; otherwise, it is not sent. Step 37: After the sensing platform receives the sub-secret, it calculates the sum of the collaboration weights of the verifiers who sent the sub-secret. If the sum of the collaboration weights is greater than or equal to t, the secret polynomial is reconstructed to obtain the reconstructed secret; otherwise, secret reconstruction is not performed.

7. A data privacy protection method based on data segmentation under mobile crowd sensing as described in claim 6, characterized in that: Step 36, which involves verifying the correctness of the sub-secret using verification information, specifically involves: If the following equation holds true, then the subsecret is correct; otherwise, the subsecret is incorrect: in, It represents the total number of validators.

8. A data privacy protection method based on data segmentation under mobile crowd sensing as described in claim 7, characterized in that: The secret polynomial for reconstruction in step three-seven obtains the secret of reconstruction, specifically as follows: First, reconstruct the secret polynomial: in, , It is a sub-secret identifier received by the sensing platform. It is the total number of sub-secrets received by the sensing platform. It is a sub-secret received by the sensing platform. , It is the verifier's identity information. It is a parameter variable; Then, let To obtain the secret after reconstruction .

Images