Vehicle crowd sensing task allocation load balancing method based on differential privacy

Abstract

The invention discloses a vehicle crowd sensing task allocation load balancing method based on differential privacy, belongs to the field of vehicle crowd sensing, and particularly relates to the vehicle crowd sensing task allocation load balancing method based on the differential privacy. The purpose of the invention is to solve the problem of user sensitive data leakage caused by the fact that most privacy protection mechanisms in the existing task allocation process emphasize the protection of user position information and neglect the protection of user perception duration in the existing vehicle crowd sensing. The method comprises the following steps: enabling U = {u0, u1,..., ui,..., un-1} to represent a user set; s = {t0, t1,..., tm-1} represents a set of all tasks; tm-1 represents the (m-1) th task in the set S of all tasks; the tasks are sorted according to the progressive increase sequence of the completion time, and a task sorting list is R = {t0 ', t1',..., t'm-1}; t'm-1 represents the (m-1) th task sequenced according to the progressive increase sequence of the completion time; and distributing each task in the task sorting list to the user in a circulating manner.

Description

Technical Field

[0001] The present invention belongs to the field of vehicle crowd intelligence perception, and specifically relates to a vehicle crowd intelligence perception task allocation load balancing method based on differential privacy. Background Art

[0002] Task allocation is an important research direction in vehicle crowd sensing, which involves the server reasonably allocating perceived tasks to vehicle users while considering system optimization goals and practical constraints. With the rapid growth of the number of vehicle users and the number of perception tasks in the crowd sensing system, an efficient task allocation mechanism is essential to improve user participation and achieve actual benefits.

[0003] Vehicle crowd sensing faces potential security threats due to the large amount of data streams and the lack of strong security mechanisms. The interaction between vehicle users, platforms, and third parties creates a scenario where privacy leakage becomes a serious issue. In vehicle crowd sensing, users need to upload perceived data to the platform, which may contain sensitive information such as user identity, location, and other private attributes. If the platform suffers from security vulnerabilities or loses credibility, the user's privacy will be exposed to unauthorized entities. When the available time of each vehicle user is limited, it is particularly important to supervise the time required to perform each task.

[0004] The goal of the present invention is to achieve task load balancing in the process of vehicle crowd-sensing task allocation, and to ensure that the user's perception time privacy needs are met while considering the travel time between tasks. Summary of the invention

[0005] The purpose of this invention is to solve the problem that most privacy protection mechanisms in the existing task allocation process in existing vehicle crowd perception emphasize the protection of user location information, but ignore the protection of user perception time, resulting in the leakage of user sensitive data, and propose a vehicle crowd perception task allocation load balancing method based on differential privacy.

[0006] A vehicle crowd sensing task allocation load balancing method based on differential privacy. The specific process is as follows:

[0007] Step 1: Let U = {u 0 ,u 1 ,…,u i ,…,u n-1} represents a user collection;

[0008] Step 2: Let S = {t 0 ,t 1 ,…,t m-1} represents the set of all tasks; S includes the tasks assigned to user u i The task set {t 0 ,t1 ,…,t j ,…,t k-1}, t k-1 ≤t m-1 ;

[0009] Among them, t 0 represents the first task in the set S of all tasks, t 1 represents the second task in the set S of all tasks, t j represents the jth task in the set S of all tasks, t k-1 represents the k-1th task in the set S of all tasks, t m-1 represents the m-1th task in the set S of all tasks;

[0010] Sort the tasks in ascending order of completion time. The task sorting list is R = {t′ 0 ,t′ 1 ,…,t′ m-1};

[0011] Among them, t′ 0 represents the first task sorted in ascending order of completion time, t′ 1 represents the second task sorted in ascending order of completion time, t′ m-1 It represents the m-1th task sorted in ascending order of completion time;

[0012] Step 3: sort the task list R = {t′ 0 ,t′ 1 ,…,t′ m-1 Each task in} is assigned to the user U = {u 0 ,u 1 ,…,u i ,…,u n-1}.

[0013] The beneficial effects of the present invention are:

[0014] In vehicle crowd intelligence perception, most privacy protection mechanisms in the existing task allocation process emphasize the protection of user location information, but ignore the protection of user perception time, and traditional task allocation technology focuses on maximizing user profits. The present invention provides a task allocation load balancing method based on privacy protection, which solves the problem of user sensitive data leakage caused by ignoring the privacy protection of user perception time, and generates optimal load balancing allocation under certain assumptions to allocate as many tasks as possible.

[0015] The present invention designs tasks to be allocated in a round-robin manner to ensure that no single vehicle user monopolizes all tasks in the system, thereby providing optimal load balancing and allocating as many tasks as possible.

[0016] Compared with the existing methods, the present invention not only provides optimal load balancing in task allocation, but also protects the privacy of user perceived time while considering the travel time between tasks, thereby avoiding leakage of user sensitive data. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0018] Specific implementation method 1: This implementation method is a vehicle crowd intelligence perception task allocation load balancing method based on differential privacy. The specific process is as follows:

[0019] A privacy-preserving vehicle crowd-sensing task allocation load balancing scheme is proposed. It allocates tasks to multiple resources (users) and protects the privacy of user perception time while considering the travel time between tasks. In this method, tasks are designed to be allocated in a round-robin manner to ensure that no single vehicle user monopolizes all tasks in the system, thereby providing fairness in task load allocation. Differential privacy technology is applied to the perception duration of vehicle users, and Laplace noise is added to the user's real stay time to confuse the data, thereby meeting the privacy protection requirements.

[0020] Step 1: Let U = {u 0 ,u 1 ,…,u i ,…,u n-1} represents a user collection;

[0021] Step 2: Let S = {t 0 ,t 1 ,…,t m-1} represents the set of all tasks; S includes the tasks assigned to user u i The task set {t 0 ,t 1 ,…,t j ,…,t k-1}, t k-1 ≤t m-1 ;

[0022] Among them, t 0 represents the first task in the set S of all tasks, t 1 represents the second task in the set S of all tasks, t j represents the jth task in the set S of all tasks, t k-1represents the k-1th task in the set S of all tasks, t m-1 represents the m-1th task in the set S of all tasks;

[0023] Sort the tasks in ascending order of completion time. The task sorting list is R = {t′ 0 ,t′ 1 ,…,t′ m-1};

[0024] Among them, t′ 0 represents the first task sorted in ascending order of completion time, t′ 1 represents the second task sorted in ascending order of completion time, t′ m-1 It represents the m-1th task sorted in ascending order of completion time;

[0025] Step 3: sort the task list R = {t′ 0 ,t′ 1 ,…,t′ m-1 Each task in} is assigned to the user U = {u 0 ,u 1 ,…,u i ,…,u n-1}.

[0026] Specific implementation method 2: This implementation method is different from the specific implementation method 1 in that in step 3, the task sorting list R={t′ 0 ,t′ 1 ,…,t′ m-1 Each task in} is assigned to the user U = {u 0 ,u 1 ,…,u i ,…,u n-1}; The specific process is:

[0027] Step 31:

[0028] Set the judgment condition Assign[j]. When Assign[j] is true, it means that task t j Assign to D users, 1≤D≤n-1; when Assign[j] is false, it means that task t j Not assigned to any one user;

[0029] Set the number of allocations No-times-tried j , No-times-tried j Represents task t j The number of times assigned to a user; No-times-tried jThe initial value of is 0;

[0030] Step 32: For task t j , judgment task t j Is it assigned to vehicle user u? i ;

[0031] If so, task t j Assigned to vehicle user u i , execute step 33;

[0032] If not, task t j Not assigned to vehicle user u i , execute steps three and four;

[0033] Step 3. Use task t j Update the value of last[i];

[0034] Let vehicle user u i =u i+1 Repeat steps 32 to 33 until the number of times is No-times-tried j =n;

[0035] Among them, last[i] represents the last one assigned to vehicle user u i mission;

[0036] Step 34: Set the allocation frequency to No-times-tried j Add 1, and vehicle user u i =u i+1 Repeat steps 32 to 33 until the number of times is No-times-tried j =n.

[0037] The other steps and parameters are the same as those in the first embodiment.

[0038] Specific implementation method three: This implementation method is different from specific implementation methods one or two in that in step three two, for task t j , judgment task t j Is it assigned to vehicle user u? i ;

[0039] If so, task t j Assigned to vehicle user u i , execute step 33;

[0040] If not, task t j Not assigned to vehicle user u i , execute steps three and four;

[0041] The specific process is:

[0042] Step 321: Calculate vehicle user u i The probability P of making at least one call within the sensing period c (the time interval for performing the task) and the sensing area a (the area for performing the task) is c ' ,a (u i );

[0043] If P c ' ,a (u i )>0, execute step 322;

[0044] If P c ' ,a (u i )≤0, continue to determine the probability that the next user makes at least one call in the sensing period c and sensing area a;

[0045] Step 322: Real stay time l u,c,a Add Laplace noise η to obtain the chaotic residence time of each user in the sensing cycle c and sensing area a Differential privacy technology is used here;

[0046] Determine the chaotic residence time of each user in the perception cycle c and perception area a Is it equal to or greater than the perceptual duration δ required for the task? If so, then If it is 1, execute step 323; if not, then is 0;

[0047] Step 3: According to the duration of the chaos and privacy budget parameter ∈, calculate the vehicle user u i The probability P of not less than the task required perception duration δ in the perception cycle c and perception area a c ' , ' a (u i );

[0048] Step 324: Based on P c ' ,a (u i ) and P c ' , ' a (u i ) Calculate each vehicle user u i The probability P of completing the perception task in the perception cycle c and the perception area a c,a (u i ), execute step 325;

[0049] Step 325: Calculate travel time;

[0050] Step 326: Determine whether the travel time T or T′ is less than the task t j The start time of task t j Assigned to vehicle user u i , execute step 33; if not, perform task t j Not assigned to vehicle user u i , execute steps three and four.

[0051] The other steps and parameters are the same as those in the first or second embodiment.

[0052] Specific implementation method 4: This implementation method is different from specific implementation methods 1 to 3 in that the vehicle user u in step 321 i The probability P of making at least one call within the sensing period c (the time interval for performing the task) and the sensing area a (the area for performing the task) is c ' ,a (u i ) is:

[0053]

[0054] in,

[0055] λ u,c,a It represents the Poisson intensity, and the Poisson intensity estimate is the vehicle user u in the sensing period c. i The average number of calls within the sensing area a;

[0056] P c,a (u i ,n) represents the probability that the user makes at least n calls in the sensing period c and sensing area a;

[0057]

[0058] in,

[0059] represents the Poisson intensity raised to the power of n.

[0060] The other steps and parameters are the same as those in Specific Embodiments 1 to 3.

[0061] Specific implementation mode 5: This implementation mode is different from the specific implementation modes 1 to 4 in that the actual residence time l in step 322 is u,c,a Add Laplace noise η to obtain the chaotic residence time of each user in the sensing cycle c and sensing area a

[0062] Determine the chaotic residence time of each user in the perception cycle c and perception area a Is it equal to or greater than the perceptual duration δ required for the task? If so, then If it is 1, execute step 323; if not, then is 0;

[0063] The specific process is:

[0064]

[0065] Among them, l u,c,a represents the vehicle user u i The actual residence time in the perception cycle c and the perception area a;

[0066] represents the chaotic residence time generated by adding Laplace noise to the real residence time;

[0067] η represents Laplace noise, η is a variable that obeys Laplace distribution. The smaller the privacy budget parameter ∈, the greater the added noise and the higher the privacy protection level.

[0068] ~ means to obey;

[0069] Laplace(0,1 / ∈) represents the Laplace distribution with parameters μ=0 and b=1 / ∈, where ∈ represents the privacy budget parameter;

[0070]

[0071] in The cloud platform can be calculated, so the above formula is regarded as a probability problem of continuous random variables; then:

[0072]

[0073] Where f(η) represents the probability density function of the variable η:

[0074]

[0075] The other steps and parameters are the same as those in Specific Embodiments 1 to 4.

[0076] Specific implementation method 6: This implementation method is different from the specific implementation methods 1 to 5 in that the steps 3, 2 and 3 are based on the chaotic residence time. and privacy budget parameter ∈, calculate the vehicle user u i The probability P of not less than the task required perception duration δ in the perception cycle c and perception area a c ' , ' a (u i ); the specific process is:

[0077]

[0078] The other steps and parameters are the same as those in Specific Implementations 1 to 5.

[0079] Specific implementation method 7: This implementation method is different from the specific implementation methods 1 to 6 in that the steps 3, 2, and 4 are based on P c ' ,a (u i ) and P c ' , ' a (u i ) Calculate each vehicle user u i The probability P of completing the perception task in the perception cycle c and the perception area a c,a (u i ), execute step 325; the specific process is:

[0080] Each vehicle user u i The probability P of completing the perception task in the perception cycle c and the perception area a c,a (u i ) is:

[0081] P c,a (u i )=P c ' ,a (u i ) c ' , ' a (u i ).

[0082] The other steps and parameters are the same as those in Specific Embodiments 1 to 6.

[0083] Specific implementation eight: This implementation differs from specific implementations one to seven in that the travel time is calculated in steps three to two and five; the specific process is as follows:

[0084] Step 3251: Assume k tasks are assigned to user u i , assigned to user u i The task set is {t 0 ,t 1 ,…,t j ,…,t k-1};

[0085] Among them, t 0 Assigned to user u i The first task in the task set, t 1 Assigned to user u i The second task in the task set, t j Assigned to user u iThe jth task in the task set, t k-1 Assigned to user u i The kth task in the task set;

[0086] Step 3252: Calculate user u i From the initial position to task t j The travel time T of the location; the specific process is:

[0087] User i First move to task t 0 Execute the first task at the location of task t, and then 0 Move to task t 1 , and so on, until user u i From the task k-2 Move to task t k-1 Execute the k-1th task at the position of user u i From the initial position to task t j The travel time to the location is:

[0088]

[0089] Among them, Euclidean (u i ,t j ) is user u i From the initial position to task t j The Euclidean distance of the position, t 0 ≤t j ≤t k-1 ; s is the speed of the vehicle user;

[0090] Step 3253: Calculate user u i From the task j The position to task t j+1 travel time T′ of the location;

[0091] The expression is:

[0092]

[0093] Among them, Euclidean (t j ,t j+1 ) is the task t j The position to task t j+1 Euclidean distance of locations.

[0094] The other steps and parameters are the same as those in Specific Embodiments 1 to 7.

[0095] Specific implementation method 9: This implementation method is different from specific implementation methods 1 to 8 in that the task t in step 326 j Assigned to vehicle user u i The specific process is:

[0096] Consider a set of n vehicle users and m tasks, where both the vehicle users and the tasks are associated with an initial location. Assume that the tasks remain fixed in their locations, but the vehicle users are mobile. Each task t j are associated with a start time and an end time, which represent the time when the task must start and end, respectively. Therefore, to successfully execute the task t j , it must be started at its start time, even though task t j The execution of a task t may end before its end time. j Can be assigned to at most one user.

[0097] Step 3261

[0098] Assume A g Assigned to user u i The task set:

[0099]

[0100] x ij ={0,1}

[0101] In the formula, |A g |≤m-1,A g ∈S;

[0102] x ij is a decision variable. If task t j Assigned to user u i , then x ij =1, otherwise x ij =0;

[0103] |A g | indicates A g The cardinality of

[0104] Let st(t j ) and ft(t j ) represent tasks t j The start and end time of ft(t d )+traveltime((t d ,t l ))>st(t l ), then task t d and t l overlapping;

[0105] t 0 ≤t d ≤t m-1 ;t 0 ≤t l ≤t m-1 ; d≠l≠j;

[0106] Among them, traveltime((t d ,t l )) indicates that from task t d The position to task t l The time spent in the location

[0107] ft(t d ) represents task t d completion time;

[0108] ft(t l ) represents task t l completion time;

[0109] st(t l ) represents task t l The start time of

[0110] Overlap refers to the overlap of the time windows of two tasks; if task t d The end time plus the time from t d To the next task l The travel time of task t is greater than l The start time of the two tasks will overlap.

[0111] Step 3262: Based on the user set, the set of all tasks, assign to user u i The task set A g , decision variable x ij , build a task load balancing distribution model.

[0112] The other steps and parameters are the same as those in Specific Embodiments 1 to 8.

[0113] Specific implementation method 10: This implementation method is different from the specific implementation methods 1 to 9 in that the steps 3262 are based on the user set, the set of all tasks, and the number of tasks assigned to user u. i The task set A g , decision variable x ij , build a task load balancing distribution model, the task load balancing distribution model expression is:

[0114] Maximizef 1 (),f 2 ()st

[0115]

[0116] x ij ={0,1}, 0≤i≤n-1, 0≤j≤m-1

[0117] t l ∈A g , d≠l, 0≤i≤n-1, t d and t l No Overlap

[0118] In the formula, f 1 () is the objective function, i.e., maximizing the number of assigned tasks;

[0119] f 2 () is the objective function, i.e., optimal load balancing among users;

[0120] A b Assigned to user u i A set of tasks; |A b | indicates A b The cardinality of the set A b The number of elements in );

[0121] 0≤j≤m-1 represents a task t j Can be assigned to at most one user;

[0122] x ij ={0,1}, 0≤i≤n-1, 0≤j≤m-1 are decision variables;

[0123] t l ∈A a , d≠l, 0≤i≤n-1, t d and t l Non-overlapping,corresponds to the overlapping condition, i.e., the two tasks assigned to a user should not overlap.

[0124] The other steps and parameters are the same as those in Specific Embodiments 1 to 9.

[0125] The present invention may also have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

Claims

1. A load balancing method for vehicle crowd sensing task allocation based on differential privacy, characterized by: The specific process of the method is: Step 1: Let U = {u0,u1,…,u i ,…,u n-1 } represents a user collection; Step 2: Let S = {t0, t1, …, t m-1 } represents the set of all tasks; S includes the tasks assigned to user u i The task set {t0,t1,…,t j ,…,t k-1 }, t k-1 ≤t m-1 ; Among them, t0 represents the first task in the set S of all tasks, t1 represents the second task in the set S of all tasks, and t j represents the jth task in the set S of all tasks, t k-1 represents the k-1th task in the set S of all tasks, t m-1 represents the m-1th task in the set S of all tasks; Sort the tasks in ascending order of completion time. The task sorting list is R = {t′0, t′1, …, t′ m-1 }; Among them, t′0 represents the first task after sorting in ascending order of completion time, t′1 represents the second task after sorting in ascending order of completion time, and t′ m-1 It represents the m-1th task sorted in ascending order of completion time; Step 3: sort the task list R = {t′0, t′1, …, t′ m-1 Each task in} is assigned to users U = {u0,u1,…,u i ,…,u n-1 }.

2. According to a method for load balancing of vehicle crowd sensing task allocation based on differential privacy in claim 1, it is characterized by: In step 3, the task sorting list R = {t′0, t′1, …, t′ m-1 Each task in} is assigned to users U = {u0,u1,…,u i ,…,u n-1 }; The specific process is: Step 31: Set the judgment condition Assign[j]. When Assign[j] is true, it means that task t j Assign to D users, 1≤D≤n-1; when Assign[j] is false, it means that task t j Not assigned to any one user; Set the number of allocations No-times-tried j , No-times-tried j Represents task t j The number of times assigned to a user; No-times-tried j The initial value of is 0; Step 32: For task t j , judgment task t j Is it assigned to vehicle user u? i ; If so, task t j Assigned to vehicle user u i , execute step 33; If not, task t j Not assigned to vehicle user u i , execute steps three and four; Step 3. Use task t j Update the value of last[i]; Let vehicle user u i =u i+1 Repeat steps 32 to 33 until the number of times is No-times-tried j =n; Among them, last[i] represents the last one assigned to vehicle user u i mission; Step 34: Set the allocation frequency to No-times-tried j Add 1, and vehicle user u i =u i+1 Repeat steps 32 to 33 until the number of times is No-times-tried j =n.

3. According to a method for load balancing of vehicle crowd sensing task allocation based on differential privacy in claim 2, it is characterized by: In step 32, for task t j , judgment task t j Is it assigned to vehicle user u? i ; If so, task t j Assigned to vehicle user u i , execute step 33; If not, task t j Not assigned to vehicle user u i , execute steps three and four; The specific process is: Step 321: Calculate vehicle user u i The probability P of making at least one call in the sensing period c and sensing area a c ' ,a (u i ); If P c ' ,a (u i )>0, execute step 322; If P c ' ,a (u i )≤0, continue to determine the probability that the next user makes at least one call in the sensing period c and sensing area a; Step 322: Real stay time l u,c,a Add Laplace noise η to obtain the chaotic residence time of each user in the sensing cycle c and sensing area a Determine the chaotic residence time of each user in the perception cycle c and perception area a Is it equal to or greater than the perceptual duration δ required for the task? If so, then If it is 1, execute step 323; if not, then is 0; Step 3: According to the duration of the chaos and privacy budget parameter ∈, calculate the vehicle user u i The probability P of not less than the task required perception duration δ in the perception cycle c and perception area a c ' , ' a (u i ); Step 324: Based on P c ' ,a (u i ) and P c ' , ' a (u i ) Calculate each vehicle user u i The probability P of completing the perception task in the perception cycle c and the perception area a c,a (u i ), execute step 325; Step 325: Calculate travel time; Step 326: Determine whether the travel time T or T′ is less than the task t j The start time of task t j Assigned to vehicle user u i , execute step 33; if not, task t j Not assigned to vehicle user u i , execute steps three and four.

4. According to a method for load balancing of vehicle crowd sensing task allocation based on differential privacy in claim 3, it is characterized by: In step 321, the vehicle user u i The probability P of making at least one call within the sensing period c (the time interval for performing the task) and the sensing area a (the area for performing the task) is c ' ,a (u i ) is: in, λ u,c,a represents Poisson intensity; P c,a (u i ,n) represents the probability that the user makes at least n calls in the sensing period c and sensing area a; in, represents the Poisson intensity raised to the power of n.

5. According to a method for load balancing of vehicle crowd sensing task allocation based on differential privacy according to claim 4, it is characterized by: In step 322, the actual residence time l u,c,a Add Laplace noise η to obtain the chaotic residence time of each user in the sensing cycle c and sensing area a Determine the chaotic residence time of each user in the perception cycle c and perception area a Is it equal to or greater than the perceptual duration δ required for the task? If so, then If it is 1, execute step 323; if not, then is 0; The specific process is: Among them, l u,c,a represents the vehicle user u i The actual residence time in the perception cycle c and the perception area a; represents the chaotic residence time generated by adding Laplace noise to the real residence time; η represents Laplace noise; ~ means to obey; Laplace(0,1 / ∈) represents the Laplace distribution, ∈ represents the privacy budget parameter; in Where f(η) represents the probability density function of the variable η:

6. The method for load balancing of vehicle crowd sensing task allocation based on differential privacy according to claim 5 is characterized by: In the step 323, according to the chaotic residence time and privacy budget parameter ∈, calculate the vehicle user u i The probability P″ of not less than the task-required perception duration δ in the perception cycle c and perception area a c,a (u i ); the specific process is:

7. The method for load balancing of vehicle crowd sensing task allocation based on differential privacy according to claim 6 is characterized by: In step 324, based on P′ c,a (u i ) and P″ c,a (u i ) Calculate each vehicle user u i The probability P of completing the perception task in the perception cycle c and the perception area a c,a (u i ), execute step 325; The specific process is: Each vehicle user u i The probability P of completing the perception task in the perception cycle c and the perception area a c,a (u i ) is: P c,a (in i )=P′ c,a (in i )P″ c,a (in i )。 8. The method for load balancing of vehicle crowd sensing task allocation based on differential privacy according to claim 7 is characterized by: In said step 325, the travel time is calculated; The specific process is: Step 3251: Assume k tasks are assigned to user u i , assigned to user u i The task set is {t0,t1,…,t j ,…,t k-1 }; Among them, t0 is the value assigned to user u. i The first task in the task set, t1 is assigned to user u i The second task in the task set, t j Assigned to user u i The jth task in the task set, t k-1 Assigned to user u i The kth task in the task set; Step 3252: Calculate user u i From the initial position to task t j The travel time T of the location; the specific process is: User i First, it moves to the position of task t0 to perform the first task, then moves from the position of task t0 to the position of task t1 to perform the second task, and so on, until user u i From the task k-2 Move to task t k-1 Execute the k-1th task at the position of user u i From the initial position to task t j The travel time to the location is: Among them, Euclidean (u i ,t j ) is user u i From the initial position to task t j The Euclidean distance of the position, t0≤t j ≤t k-1 ; s is the speed of the vehicle user; Step 3253: Calculate user u i From the task j The position to task t j+1 travel time T′ of the location; The expression is: Among them, Euclidean (t j ,t j+1 ) is the task t j The position to task t j+1 Euclidean distance of locations.

9. The method for load balancing of vehicle crowd sensing task allocation based on differential privacy according to claim 8 is characterized in that: Task t in step 326 j Assigned to vehicle user u i The specific process is: Step 3261. Set A g Assigned to user u i The task set: x ij ＝{0,1} In the formula, |A g |≤m-1,A g ∈S; x ij is a decision variable. If task t j Assigned to user u i , then x ij =1, otherwise x ij =0; |A g | indicates A g The cardinality of Let st(t j ) and ft(t j ) represent tasks t j The start and end time of ft(t d )+traveltime((t d ,t l ))>st(t l ), then task t d and t l overlapping; t0≤t d ≤t m-1 ;t0≤t l ≤t m-1 ;d≠l≠j; Among them, traveltime((t d ,t l )) indicates that from task t d The position to task t l The time spent in the location ft(t d ) represents task t d completion time; ft(t l ) represents task t l completion time; st(t l ) represents task t l The start time of Step 3262: Based on the user set, the set of all tasks, assign to user u i The task set A g , decision variable x ij , build a task load balancing distribution model.

10. The method for load balancing of vehicle crowd sensing task allocation based on differential privacy according to claim 9 is characterized in that: In the step 3262, based on the user set, the set of all tasks, and the task set assigned to user u i The task set A g , decision variable x ij , build a task load balancing distribution model, the task load balancing distribution model expression is: Maximizef1(),f2()st x ij ={0,1},0≤i≤n-1,0≤j≤m-1 t l ∈A g , d≠l, 0≤i≤n-1, t d and t l No Overlap Where f1() is the objective function, which is to maximize the number of assigned tasks; f2() is the objective function, i.e., the optimal load balancing among users; A b Assigned to user u i A set of tasks; |A b | indicates A b The cardinality of .

Images