A trajectory data privacy protection method and system based on venn diagram and differential privacy

Abstract

The application relates to a trajectory data privacy protection method and system based on a Venn diagram and differential privacy, and relates to the field of privacy protection.The application is used for solving the problem of low safety of an existing trajectory privacy data safety protection method.The application comprises the following steps: obtaining friend users of a to-be-protected trajectory user A, and discretizing the coordinates of the locations on each friend user trajectory into a geographical space after the geographical space is divided into grids;obtaining a location coordinate set contained in an i-th friend user trajectory of the user A;obtaining an intersection C of the location coordinate sets in all the friend user trajectories of the user A; and performing privacy protection on the location coordinates in the intersection C to obtain a trajectory of the user A after privacy protection.The application is used for performing privacy protection on trajectory data of a user.

Description

A method and system for protecting trajectory data privacy based on Venn graphs and differential privacy. Technical Field

[0001] This invention relates to the field of privacy protection, and in particular to a method and system for protecting trajectory data privacy based on Venn diagrams and differential privacy. Background Technology

[0002] With the widespread use of mobile devices and location services, users' trajectory data is being collected and used extensively. While this improves service quality and user experience, it also brings serious privacy risks. Unprotected trajectory data may expose users' personal privacy information and could even be used maliciously.

[0003] Currently, trajectory data privacy protection technologies mainly focus on two aspects: online application trajectory data privacy protection and offline application trajectory data privacy protection. Online application trajectory data privacy protection provides users with corresponding efficient services through the real-time location of moving objects, such as intelligent traffic management and mobile medical services. Offline application trajectory data privacy protection analyzes and mines data collected from individuals or specific institutions to collect sensitive information about other individuals, such as home address, workplace, lifestyle, and health status. However, adversaries with background knowledge can make malicious inferences about the privacy of individual trajectory data. Therefore, existing methods cannot effectively curb malicious attacks and privacy leaks based on background knowledge, resulting in low security of existing trajectory privacy data security protection methods. Summary of the Invention

[0004] The purpose of this invention is to address the low security of existing trajectory privacy data protection methods, and to propose a trajectory data privacy protection method and system based on Venn graphs and differential privacy.

[0005] A trajectory data privacy protection method based on Venn graphs and differential privacy includes the following steps:

[0006] Step 1: Obtain the friends of user A whose trajectory is to be protected, and discretize the coordinates of the locations on each friend's trajectory into a gridded geographic space;

[0007] Step 2: Obtain the set of location coordinates F contained in the trajectory of user A's i-th friend. i =(x i,1 ,x i,2 ,x i,n ,....,x i,N );

[0008] Among them, F i It is the set of location coordinates contained in the trajectory of the i-th friend user, x i,nThese are the coordinates of the nth location on the trajectory of the i-th friend user; N is the total number of locations on the trajectory of the i-th friend user.

[0009] Step 3: Obtain the intersection C of the location coordinates sets of all of user A's friends' user trajectories, and perform privacy protection on the location coordinates in the intersection C to obtain user A's trajectory after privacy protection.

[0010] Furthermore, in step three, the intersection C of the location coordinates sets of all of user A's friends' user trajectories is obtained, and the location coordinates in the intersection C are protected for privacy to obtain user A's trajectory after privacy protection. Specifically:

[0011] Step 3: 1. Obtain the intersection C of the location coordinates sets from the trajectories of all of User A's friends:

[0012]

[0013] Where s is the total number of user A's friends. These are the coordinates of the i*th location in the intersection C. It is the coordinate of the N*th location in the intersection C, where N* is the total number of locations in the intersection C;

[0014] Step 3.2: Use the intersection C to obtain the trajectory of user A after privacy protection.

[0015] Furthermore, the step 3.2, obtaining the privacy-protected trajectory of user A using the intersection C, specifically involves:

[0016] Step 321: Initialize user A's friend user IDs i = 1;

[0017] Step 322, Using r as the center, draw a circle with radius r to obtain a circular region. and initialize Location number j = 1 within;

[0018] Step 3: Obtaining Points utility value

[0019] in, yes The coordinates of the j-th location within;

[0020] Steps 3, 2, and 4: Point-based utility value Get The probability of the center of the circle

[0021] Step 325: Determine if j equals k. If j < k, then set j = j + 1 and return to step 323; if j = k, then... The location corresponding to the maximum value is used as The center of confusion;

[0022] Where k is The total number of locations within;

[0023] Step 326: Determine if i* is less than N*. If i* < N*, let i* = i* + 1 and return to step 322. If i* = N*, output all obfuscation centers, the privacy-protected trajectory of user A composed of all obfuscation centers, and publish the privacy-protected trajectory of user A.

[0024] Furthermore, the acquisition point in step three-two-three utility value Specifically:

[0025]

[0026] in, It is a point Time The Euclidean distance, k is The total number of locations within the area.

[0027] Furthermore, the point-based steps in steps three, two, and four... utility value Get The probability of the center of the circle Specifically:

[0028]

[0029] Where ΔU is the sensitivity and ε is the privacy budget parameter. yes As The probability of the center of the circle.

[0030] A trajectory data privacy protection system based on Venn diagrams and differential privacy includes: a module for obtaining the location coordinates of friends' user trajectories, a module for obtaining the intersection of friends' user trajectories, a module for obtaining the obfuscated center location, and a module for generating user trajectories after privacy protection;

[0031] The friend user trajectory location coordinate acquisition module is used to acquire the location coordinates in the friend user trajectory of user A whose trajectory is to be protected, and discretize the location coordinates into a gridded geographic space to obtain the set F of location coordinates contained in the i-th friend user trajectory of user A. i =(x i,1 ,xi,2 ,x i,n ,....,x i,N );

[0032] Among them, F i It is the set of location coordinates contained in the trajectory of the i-th friend user, x i,n These are the coordinates of the nth location on the trajectory of the i-th friend user; N is the total number of locations on the trajectory of the i-th friend user.

[0033] The friend user trajectory intersection acquisition module obtains the intersection C of the location coordinate sets in the trajectories of all friend users of user A;

[0034] The module for obtaining the confusion center location is used to draw circles with each location in the intersection C as the center, and obtain the confusion center location of each circle;

[0035] The privacy-protected user trajectory generation module uses each obfuscated center location to form the privacy-protected trajectory of user A and publishes it.

[0036] Furthermore, the friend user trajectory intersection acquisition module obtains the intersection C of the set of location coordinates in the trajectories of all of user A's friends, specifically as follows:

[0037]

[0038] Where s is the total number of user A's friends. These are the coordinates of the i*th location in the intersection C. It is the coordinate of the N*th location in the intersection C, where N* is the total number of locations in the intersection C.

[0039] Furthermore, the confusion center location acquisition module is used to draw circles with each location in the intersection C as the center, and obtain the confusion center location of each circle, specifically:

[0040] S1. Initialize the user IDs of user A's friends, i = 1;

[0041] S2, will Using r as the center, draw a circle with radius r to obtain a circular region. and initialize Location number j = 1 within;

[0042] S3, Acquisition Point utility value

[0043] in, yes The coordinates of the j-th location within;

[0044] S4, Point-based utility value Get The probability of the center of the circle

[0045] S5. Determine if j equals k. If j < k, set j = j + 1 and return to S3; if j = k, then... The location corresponding to the maximum value is used as The center of confusion;

[0046] Where k is The total number of locations within;

[0047] S6. Determine if i* is less than N*. If i* < N*, let i* = i* + 1 and return to step S2. If i* = N*, output all confusion centers.

[0048] Furthermore, the acquisition point in S3 utility value Specifically:

[0049]

[0050] in, It is a point Time The Euclidean distance, k is The total number of locations within the area.

[0051] Furthermore, the point-based method in S4 utility value Get The probability of the center of the circle Specifically:

[0052]

[0053] Where ΔU is the sensitivity and ε is the privacy budget parameter. yes As The probability of the center of the circle.

[0054] The beneficial effects of this invention are as follows:

[0055] This invention utilizes a grid to divide real-world geographic space, transforming user trajectory datasets into grid location coordinates. To prevent attackers from inferring sensitive individual information using background knowledge, this invention employs a method of relationships between Venn diagram sets. It takes as input a set of location coordinates of the user's friends, outputs a set of duplicate location coordinates of the friends, and uses each location coordinate in the duplicate set as a center point. A differential privacy exponential mechanism is then used to select a location coordinate within the circular area to replace the center point, thus protecting the duplicate location coordinates and ultimately protecting the privacy of individual trajectory data. This invention effectively resists malicious inferences about individual trajectory data privacy by adversaries with background knowledge. While effectively protecting trajectory data privacy, it also ensures the usability of trajectory data, improving the security of trajectory privacy data protection methods. Attached Figure Description

[0056] Figure 1 shows the results of geographic grid spatial division;

[0057] Figure 2 shows the set relations in the Venn diagram;

[0058] Figure 3 is a schematic diagram of the privacy protection scheme. Detailed Implementation

[0059] Specific Implementation Method 1: The specific process of this implementation method for protecting trajectory data privacy based on Venn diagrams and differential privacy is as follows:

[0060] Step 1: Obtain the friends of user A whose trajectory is to be protected, and discretize the coordinates of the locations on each friend's trajectory into a gridded geographic space;

[0061] In this context, the location coordinates on each trajectory represent a point on a geospatial grid.

[0062] Step 2: Obtain the set F of location coordinates contained in the trajectory of the i-th friend user. i =(x i,1 ,x i,2 ,x i,n ,....,x i,N );

[0063] Among them, F i It is the set of location coordinates contained in the trajectory of the i-th friend user, x i,n These are the coordinates of the nth location on the trajectory of the i-th friend user; N is the total number of locations on the trajectory of the i-th friend user.

[0064] Step 3: Obtain sets F1, ..., F i ..., F sThe intersection C of the coordinates of the points in the intersection C is obtained by protecting the privacy of the location coordinates in the intersection C, and the trajectory of user A after privacy protection is obtained as follows:

[0065] Step 31: Obtain sets F1, ..., F i ..., F s The intersection C:

[0066]

[0067] Among them, F s Let be the s-th friend of user A, where s is the total number of friends of A. These are the coordinates of the i*th location in the intersection C. It is the coordinate of the N*th location in the intersection C, where N* is the total number of locations in the intersection C;

[0068] Step 3.2: Using the coordinates of each location in the intersection C as the center, draw a circle with radius r to obtain a circular region. Obtain the utility value of each location within each circular region, and based on the utility value of each location within the circular region, obtain the location coordinates of each obfuscated center position. Then, publish the trajectory of user A after privacy protection.

[0069] Step 321: Initialize i = 1;

[0070] Step 322, Using r as the center, draw a circle with radius r to obtain a circular region, and initialize it. Location number j = 1 within;

[0071] Step 3, obtain Locations within a circular area centered at the center utility value

[0072]

[0073] in, Therefore Let be the coordinates of the j-th location within the circular region centered at the center. It is a point Time The Euclidean distance, k is... This represents the total number of locations within the circular area centered on the circle.

[0074] Steps 3, 2, and 4: Based on the results obtained in Step 3, 2, and 3. Get The probability of the center of the circle

[0075] As The probability of the center of the circle is as follows:

[0076]

[0077] Where ΔU is the sensitivity setting of 1, and ε is the privacy budget parameter. yes As The probability of the center of the circle;

[0078] Step 325: Determine if j equals k. If j < k, then set j = j + 1 and return to step 323; if j = k, then... The location corresponding to the maximum value is used as The center of confusion;

[0079] Where k is The total number of locations within;

[0080] Step 326: Determine if i* is less than N*. If i < N*, let i* = i* + 1 and return to step 322. If i* = N*, output the privacy-protected trajectory of user A composed of all the obfuscated centers and publish it.

[0081] Specific Implementation Method Two: A trajectory data privacy protection system based on Venn diagram and differential privacy, including: a module for obtaining the location coordinates of friends' user trajectories, a module for obtaining the intersection of friends' user trajectories, a module for obtaining the obfuscated center location, and a module for generating user trajectories after privacy protection;

[0082] The friend user trajectory location coordinate acquisition module is used to acquire the location coordinates in the friend user trajectory of user A whose trajectory is to be protected, and discretize the location coordinates into a gridded geographic space to obtain the set F of location coordinates contained in the i-th friend user trajectory of user A. i =(x i,1 ,x i,2 ,x i,n ,....,x i,N );

[0083] Among them, F i It is the set of location coordinates contained in the trajectory of the i-th friend user, x i,n These are the coordinates of the nth location on the trajectory of the i-th friend user; N is the total number of locations on the trajectory of the i-th friend user.

[0084] The friend user trajectory intersection acquisition module obtains the intersection C of the set of location coordinates in the trajectories of all of user A's friends, specifically:

[0085]

[0086] Where s is the total number of user A's friends. These are the coordinates of the i*th location in the intersection C. It is the coordinate of the N*th location in the intersection C, where N* is the total number of locations in the intersection C;

[0087] The module for obtaining the confusion center location is used to draw circles with each location in the intersection C as the center, and obtain the confusion center location of each circle, specifically:

[0088] S1. Initialize the user IDs of user A's friends, i = 1;

[0089] S2, will Using r as the center, draw a circle with radius r to obtain a circular region. and initialize Location number j = 1 within;

[0090] S3, Acquisition Point utility value Specifically:

[0091]

[0092] in, It is a point Time The Euclidean distance, k is The total number of locations within;

[0093] in, yes The coordinates of the j-th location within;

[0094] S4, Point-based utility value Get The probability of the center of the circle Specifically:

[0095]

[0096] Where ΔU is the sensitivity and ε is the privacy budget parameter. yes As The probability of the center of the circle;

[0097] S5. Determine if j equals k. If j < k, set j = j + 1 and return to S3; if j = k, then... The location corresponding to the maximum value is used as The center of confusion;

[0098] Where k is The total number of locations within;

[0099] S6. Determine if i* is less than N*. If i* < N*, let i* = i* + 1 and return to step S2. If i* = N*, output all confusion centers.

[0100] The privacy-protected user trajectory generation module uses each obfuscated center location to form the privacy-protected trajectory of user A and publishes it.

[0101] Example: To verify the beneficial effects of the present invention, the following experiments were conducted in this example:

[0102] (1) The attack model set in this experiment is the background knowledge attack model: It is assumed that the attacker knows that Alice, John and Eric are good friends through background knowledge, and knows the trajectory data of Alice and John. By calculating the same trajectory points of Alice and John, Eric's trajectory data is inferred. The trajectory dataset of Alice, John and Eric is shown in Table 1:

[0103] Table 1 Trajectory Dataset

[0104]

[0105] (2) Divide the geospace into grids to ensure that each location point in the trajectory dataset is in the grid location points: Discretize Alice's trajectory data GPS points to the location point coordinates in the gridded geospace; Discretize John's trajectory data GPS points to the location point coordinates in the gridded geospace. The result of the geospace grid division is shown in Figure 1. The attack model uses the relationship between Venn graph sets to calculate the duplicate location point set of Alice and John, and clearly infers Eric's trajectory data, leading to the leakage of sensitive information, as shown in Figure 2.

[0106] The attack model uses the relationships between Venn graph sets to calculate the set of repeated locations for Alice and John:

[0107] C = A ∩ B = (School, Shopping Mall, Company)

[0108] A = (school, restaurant, shopping mall, company, home, playground)

[0109] B = (Company, School, Stadium, Shopping Mall, Hospital, Pet Shop)

[0110] (3) In this embodiment, the method described in the specific implementation method utilizes the differential privacy exponential mechanism to perform privacy protection processing on the set of repeated location coordinates, specifically as follows:

[0111] Within a circular region centered at a repeating location coordinate point and with radius r, calculate the utility value function for each location coordinate point in the set of repeating location coordinate points.

[0112] The utility value function is the coordinates of each location within the circular region. coordinates of the center point The Euclidean distance between them, thus obtaining the Euclidean distance between them. Locations within a circular area centered at the center utility value

[0113]

[0114] Confused about the coordinates of the center point The probability of:

[0115]

[0116] Where ε is the privacy budget parameter;

[0117] The location coordinates of each duplicate location point are obtained by obfuscating them in the above manner, and then the trajectory data that has been processed for privacy protection is published, as shown in Figure 3.

[0118] The above experiments demonstrate that the present invention can protect the privacy of users' trajectories, thereby preventing attack models from using background knowledge to infer sensitive individual information.

Claims

1. A method for protecting the privacy of trajectory data based on Venn graphs and differential privacy, characterized in that... The specific process of the method is as follows: Step 1: Obtain the friend users of user A whose trajectory is to be protected, and discretize the coordinates of the locations on the trajectory of each friend user into a gridded geographic space; Step 2: Obtain the set of location coordinates contained in the trajectory of the i-th friend user of user A. ;in, It is the set of location coordinates contained in the trajectory of the i-th friend user. These are the coordinates of the nth location on the trajectory of the i-th friend user; N is the total number of locations on the trajectory of the i-th friend user; Step 3: Obtain the intersection of the sets of location coordinates from the trajectories of all of user A's friends. and the intersection To protect the privacy of user A's location coordinates, we obtain the user A's trajectory after privacy protection. Specifically, Step 31: Obtain the intersection of the location coordinate sets from the trajectories of all of user A's friends. ： in, This is the total number of user A's friends. It is an intersection The first in Location coordinates, It is an intersection The first in Location coordinates, It is an intersection Total number of locations; Step 32: Utilize intersection After obtaining privacy protection, the trajectory of user A is as follows: Step 321, initialize user A's friend user ID i=1; Step 322, ... As the center, with Draw a circle with radius to obtain a circular region. and initialize Location labels within Step 3: Obtaining Points utility value Specifically: in, It is a point Time The Euclidean distance, k is The total number of locations within, yes The first Location coordinates; Steps 3-4: Utilizing points utility value Get The probability of the center of the circle Step 325: Judgment Is it equal to k? If Then let And return to step three; if Then The location corresponding to the maximum value is used as The center of confusion; among them, yes Total number of locations within; Step 326, Judgment Is it less than ,like Then let And return to step 322; if Then output all obfuscation centers, the privacy-protected trajectory of user A composed of all obfuscation centers, and publish the privacy-protected trajectory of user A.

2. The trajectory data privacy protection method based on Venn diagrams and differential privacy according to claim 1, characterized in that: The point-based steps in steps three, two, and four utility value Get The probability of the center of the circle Specifically: in, It's about sensitivity. It is a privacy budget parameter. yes As The probability of the center of the circle.

3. A trajectory data privacy protection system based on Venn graphs and differential privacy, characterized in that... The system includes: a friend user trajectory location coordinate acquisition module, a friend user trajectory intersection acquisition module, a confusion center location acquisition module, and a privacy-protected user trajectory generation module; the friend user trajectory location coordinate acquisition module is used to acquire the coordinates of locations contained in the friend user trajectories of user A whose trajectory is to be protected, and discretize the location coordinates into a gridded geographic space to obtain the set of location coordinates contained in the i-th friend user trajectory of user A. ;in, It is the set of location coordinates contained in the trajectory of the i-th friend user. These are the coordinates of the nth location on the trajectory of the i-th friend user; N is the total number of locations on the trajectory of the i-th friend user; the friend user trajectory intersection acquisition module obtains the intersection of the set of location coordinates in the trajectories of all friends of user A. Specifically: in, This is the total number of user A's friends. It is an intersection The first in Location coordinates, It is an intersection The first in Location coordinates, It is an intersection Total number of locations; the confusion center location acquisition module is used to find the intersection. Using each location as the center, draw circles to obtain the confusion center of each circle. Specifically: S1, initialize user A's friend user ID i=1; S2, ... As the center, with Draw a circle with radius to obtain a circular region. and initialize Location labels within S3, Acquisition Point utility value Specifically: in, It is a point Time The Euclidean distance, k is The total number of locations within, yes The first Location coordinates; S4, point-based utility value Get The probability of the center of the circle S5, Judgment Is it equal to k? If Then let And return to S3; if Then The location corresponding to the maximum value is used as The center of confusion; among them, yes Total number of locations within; S6, Judgment Is it less than ,like Then let And return to step S2; if Then all obfuscated center points are output; the privacy-protected user trajectory generation module uses the location corresponding to each obfuscated center point to form the privacy-protected trajectory of user A, and publishes it.

4. The trajectory data privacy protection system based on Venn diagrams and differential privacy according to claim 3, characterized in that: The point-based S4 utility value Get The probability of the center of the circle Specifically: in, It's about sensitivity. It is a privacy budget parameter. yes As The probability of the center of the circle.

Images