Complex road condition non-uniform vehicle task unloading method based on VNF instance sharing

Abstract

The invention discloses a VNF instance sharing-based non-uniform vehicle task offloading method for complex road conditions, and belongs to the technical field of communication. The invention provides a non-uniform vehicle task unloading method based on VNF instance sharing, and aims to solve the problem of task unloading failure caused by instability or frequent interruption of inter-vehicle links due to high time-varying characteristics of vehicle speed and direction in scenes such as large transportation hubs and complex interchange interchange. According to the method, candidate unloading vehicles are selected and a corresponding unloading path set is constructed according to unloading demands of sub-tasks in a vehicle task request and a vehicle driving track change trend, and according to available computing resources and VNF instance distribution of the candidate unloading vehicles, a resource utilization rate is optimized and an optimal unloading decision is constructed. Therefore, the task completion rate of the system is effectively improved, and the unit task unloading delay is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of communication networks, and in particular relates to a method for unloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing. Background Art

[0002] With the development of wireless communication technology, vehicles are no longer simple means of transportation, but intelligent connected cars with powerful communication and computing capabilities. Vehicles are equipped with advanced sensing, wireless communication, and on-board computing modules when they leave the factory, and realize vehicle-to-vehicle (V2V) communication through technologies such as dedicated short-range communication based on the IEEE 802.11p protocol. Network function virtualization can realize a variety of virtual network functions (VNF) on the physical network by softwareizing and modularizing network functions.

[0003] Existing research on vehicle network task offloading is mainly based on scenarios where the vehicle direction and speed are fixed, which is difficult to meet the needs of actual application scenarios. For example, in scenarios such as large transportation hubs and complex interchanges, the direction and speed of the vehicle will be adjusted in real time as the road conditions change, resulting in rapid and disorderly changes in the network connection status. This situation will cause the interruption of the communication link between vehicles and the failure of task offloading, thereby reducing the utilization of network resources and worsening the user's service experience.

[0004] In view of the above problems, the present invention proposes a method for unloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing. According to the maximum unloading tolerance delay constraint of the task in the vehicle unloading request, the VNF type requirement, and the trend of the vehicle's driving trajectory, a candidate unloading vehicle is selected for each subtask and a corresponding unloading path is constructed. According to the remaining available computing resources of the candidate unloading vehicle and the distribution of VNF instances, the particle swarm algorithm is used to optimize the task unloading decision, and the optimal unloading vehicle and unloading path are found for each subtask, thereby maximizing the system task completion rate and minimizing the system unit task unloading delay. Summary of the invention

[0005] The present invention aims to solve the above problems of the prior art. A method for unloading tasks of vehicles with non-uniform speed in complex road conditions based on VNF instance sharing is proposed. The technical solution of the present invention is as follows:

[0006] A method for offloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing comprises the following steps:

[0007] 101. Use particle swarm algorithm to establish vehicle network task offloading optimization model. Suppose task vehicle i *The unloading request J = {j}, where j is a parallel subtask, the particle swarm is a set N = {n}, and the unloading decision vector, migration vector and optimal fitness of each particle n in N are ψ n =(ψ n,j | j∈J ),s n =(s n,j | j∈J )and Among them, ψ n,j is the offloading path of subtask j in particle n, s n,j is n,j The transformation step size is η, and the maximum number of iterations of the initialized particle swarm is η max , iteration count variable η=0, temporary set J′=J;

[0008] 102. If Take the first element j from the set J′, and according to the amount of data a of subtask j j , the required virtual network function VNF type v j , computational complexity c j , establish a candidate offloading path set P for subtask j j ={p j,k}, where p j,k If it is a candidate offloading path coded as k for subtask j, jump to step 102; otherwise, jump to step 103;

[0009] 103. For each particle n in N, from each set P j | j∈J Choose any path p j,k Initialize the offload decision vector From the value interval [0,|P j |) randomly selects an integer to initialize the migration vector s n Each component s in n,j | j∈J , calculate the particle fitness t n , initialize the optimal fitness of particle n Optimal unloading decision for particle n

[0010] 104. Let the global optimal unloading decision of particle swarm N be Iteration count variable η=η+1, if η≤η max , let the temporary set N′=N, jump to step 105, otherwise, jump to step 107;

[0011] 105. If Take the first element n from the set N′ and update the unloading decision vector ψn and the migration vector s n , calculate the particle fitness t n , jump to step 106, otherwise, jump to step 104;

[0012] 106. If Let the optimal fitness of particle n be Optimal unloading decision for particle n Jump to step 105, otherwise, jump to step 105;

[0013] 107. Output the global optimal unloading decision ψ best , the step ends.

[0014] Furthermore, in step 102, a candidate offloading path set P is established for subtask j. j ={p j,k}, specifically including the following steps:

[0015] 1) Initialize the set of all vehicles I = {i}, where i is any vehicle, and the set of candidate unloading vehicles The set of candidate offloading paths for subtask j The maximum tolerable delay of task offloading t max ;

[0016] 2) If Take the first element i from the set I, according to the amount of data a of subtask j j , Required virtual network function VNF type v j , computational complexity c j , calculate the minimum computing resources that vehicle i needs to allocate for subtask j Jump to step 3), otherwise, jump to step 4);

[0017] 3) If the remaining available computing resources f of vehicle i i r Greater than or equal to Add vehicle i to the candidate unloading vehicle set I′ and jump to step 2); otherwise, jump to step 2);

[0018] 4) If Take the first element i from the set I′ and use the depth-first search algorithm to construct the task vehicle i * The set P′ of all feasible paths to candidate unloading vehicle i j ={p j,k}, where p j,k is a feasible offloading path coded as k for subtask j, jump to step 5), otherwise, jump to step 9);

[0019] 5) If According to the data volume a of subtask j j , Required virtual network function VNF type v j , computational complexity c j and the remaining available computing resources f of candidate unloading vehicle i i r , calculate the minimum computational delay of candidate unloading vehicle i to process subtask j Jump to step 6), otherwise, jump to step 4);

[0020] 6) For P′ j Each path p in j,k , let subtask j be along p j,k The upload delay to candidate unloading vehicle i is Candidate unloading vehicle i along p j,k The estimated return delay of the original return result is Among them, r l is the transmission rate of physical link l, σ j is the ratio of the uploaded and returned data volumes of subtask j, r min is the minimum transmission rate between vehicles, and the path p is calculated j,k Duration Delay the completion time of offloading subtask j Greater than All p j,k From the set P′ j Delete from;

[0021] 7) If P′ j Each path p in j,k , calculated in The minimum computing resources required for candidate unloading vehicle i to allocate for subtask j within time Calculate the computing resources f that candidate unloading vehicle i needs to allocate for subtask j under fair allocation of on-board resources Fr , jump to step 8), otherwise, jump to step 4);

[0022] 8) From P′ j Choose to meet A path p j,k Add candidate offloading path set P j , jump to step 4);

[0023] 9) Output candidate offloading path set P j , the step ends.

[0024] Furthermore, in step 2), the minimum computing resources that vehicle i needs to allocate for subtask j is The calculation method is shown in formula (1):

[0025]

[0026] In formula (1), a j represents the amount of data for subtask j, c j represents the computational complexity of subtask j, t max represents the maximum tolerable delay of task offloading, f0 represents the basic computing resources required to instantiate VNF, and λ i,j is a binary variable, indicating whether vehicle i has a sharable VNF instance. If not, the corresponding VNF needs to be instantiated for subtask j. Let λ i,j =1, otherwise, let λ i,j =0, the calculation method is shown in formula (2):

[0027]

[0028] In formula (2), v j represents the VNF type required for subtask j, V i Represents the set of VNF types contained in vehicle i.

[0029] Furthermore, the minimum computational delay of the candidate unloading vehicle i in step 5) for processing subtask j is The calculation method is shown in formula (3):

[0030]

[0031] In formula (3), f i r represents the remaining available computing resources of candidate unloading vehicle i.

[0032] Furthermore, in step 6), the passage p j,k Duration The calculation method is shown in formula (4):

[0033]

[0034] In formula (4), τ l represents the link duration of the physical link l between vehicle i′ and vehicle i″, where vehicle i′ and vehicle i″ are two vehicles that can communicate directly, τ l The value of is calculated by formula (5):

[0035]

[0036] In formula (5), D represents the effective communication range of the vehicle, (x i′ ,y i′ )、(xi″ ,y i″ ) represent the current position coordinates of vehicle i′ and vehicle i″, μ max represents the maximum speed of the vehicle, γ i′ , γ i″ They respectively represent the angles between the driving directions of vehicle i′ and vehicle i″ and the positive direction of the x-axis after the direction change, and are calculated as shown in formula (6):

[0037]

[0038] In formula (6), α i′ , α i″ They represent the angles between the current driving directions of vehicle i′ and vehicle i″ and the positive direction of the x-axis, and β represents the angle change when vehicle i′ and vehicle i″ change from the current driving directions to the opposite directions. The calculation method is shown in formula (7):

[0039]

[0040] In formula (7), θ represents the instantaneous maximum steering angle of the vehicle.

[0041] Further, in the step 7) The minimum computing resources required for candidate unloading vehicle i to allocate for subtask j within time The computing resources f that candidate unloading vehicle i needs to allocate for subtask j under fair allocation of on-board resources Fr The calculation methods are shown in formula (8) and formula (9) respectively:

[0042]

[0043] In formula (9), f i represents the total computing resources of vehicle i, |I * | represents the total number of mission vehicles, |J| max represents the maximum number of subtasks contained in each unloading request, and |I| represents the total number of vehicles.

[0044] Furthermore, the fitness t of particle n in step 103 is n The calculation method of is shown in formula (10):

[0045]

[0046] In formula (10), represents subtask j along path p j,k The unloading completion delay to vehicle i is calculated as shown in formula (11): is a binary variable, indicating whether subtask j follows path p for particle n.j,k Uninstall to p j,k The last node vehicle i, if yes, let Otherwise, let The calculation method is shown in formula (12):

[0047]

[0048] Furthermore, in step 105, the unloading decision vector ψ of each particle n is updated. n , that is, update ψ n Each component ψ n,j , the calculation method is shown in formula (13):

[0049]

[0050] Update the migration vector s of each particle n n The calculation method of is shown in formula (14):

[0051]

[0052] In formula (14), round() means rounding to the nearest integer, ω means the inertia factor, ε1 and ε2 mean the learning factors, and rand() means a random number in the numerical interval (0,1).

[0053] The advantages and beneficial effects of the present invention are as follows:

[0054] The present invention discloses a method for unloading tasks of non-uniform-speed vehicles in complex road conditions based on VNF instance sharing. Existing research on the problem of vehicle network task unloading usually only considers uniformly moving vehicles on single / bidirectional lanes, and does not consider the real-time changing characteristics of vehicle speed and direction in large transportation hubs or complex interchange environments. In response to the above problems, the present invention selects unloading vehicles for each subtask and constructs an unloading path based on the maximum tolerable delay constraint of the task in the vehicle unloading request, the VNF type requirement, the shareability of the VNF instance, and the changing trend of the vehicle's driving trajectory to ensure the availability of computing resources and the reliability of the unloading path. The unloading decision vector is optimized through the particle swarm algorithm to maximize resource utilization, thereby improving the system task completion rate and reducing the system unit task unloading delay. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 The present invention provides a preferred embodiment of a flowchart of a method for unloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing. DETAILED DESCRIPTION

[0056] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0057] The technical solution of the present invention to solve the above technical problems is:

[0058] The concepts and models involved in the present invention are as follows:

[0059] The present invention assumes that in a vehicle network with complex road conditions, the driving direction and speed of each vehicle change dynamically. Based on the effective communication distance constraint between vehicles, V2V direct link or V2V multi-hop link forwarding task can be adopted. Each vehicle is equipped with a server with certain computing power and can run multiple VNF instances at the same time. Vehicle tasks are composed of several different types of subtasks, which can be unloaded in parallel to local or other idle vehicles. Among them, different subtasks with the same required VNF type can share the VNF instance on the same vehicle node, and the same task can only be unloaded to one vehicle at most.

[0060] The technical solution of the present invention is described as follows:

[0061] 1. The minimum computing resources that vehicle i needs to allocate for subtask j

[0062] The calculation method is shown in formula (1):

[0063]

[0064] In formula (1), a j represents the amount of data for subtask j, c j represents the computational complexity of subtask j, t max represents the maximum tolerable delay of task offloading, f0 represents the basic computing resources required to instantiate VNF, and λ i,j is a binary variable, indicating whether vehicle i has a sharable VNF instance. If not, the corresponding VNF needs to be instantiated for subtask j. Let λ i,j =1, otherwise, let λ i,j =0, the calculation method is shown in formula (2):

[0065]

[0066] In formula (2), v j represents the VNF type required for subtask j, V i Represents the set of VNF types contained in vehicle i;

[0067] 2. Minimum computational delay of candidate unloading vehicle i processing subtask j

[0068] The calculation method is shown in formula (3):

[0069]

[0070] In formula (3), f i r represents the remaining available computing resources of candidate unloading vehicle i;

[0071] 3. Pathway p j,k Duration

[0072] The calculation method is shown in formula (4):

[0073]

[0074] In formula (4), τ l represents the link duration of the physical link l between vehicle i′ and vehicle i″, where vehicle i′ and vehicle i″ are two vehicles that can communicate directly, τ l The value of is calculated by formula (5):

[0075]

[0076] In formula (5), D represents the effective communication range of the vehicle, (x i′ ,y i′ )、(x i″ ,y i″ ) represent the current position coordinates of vehicle i′ and vehicle i″, μ max represents the maximum speed of the vehicle, γ i′ , γ i″ They respectively represent the angles between the driving directions of vehicle i′ and vehicle i″ and the positive direction of the x-axis after the direction change, and are calculated as shown in formula (6):

[0077]

[0078] In formula (6), α i′ , α i″ They represent the angles between the current driving directions of vehicle i′ and vehicle i″ and the positive direction of the x-axis, and β represents the angle change when vehicle i′ and vehicle i″ change from the current driving directions to the opposite directions. The calculation method is shown in formula (7):

[0079]

[0080] In formula (7), θ represents the instantaneous maximum steering angle of the vehicle;

[0081] 4. The minimum computing resources required for candidate unloading vehicle i to allocate for subtask j within time The computing resources f that candidate unloading vehicle i needs to allocate for subtask j under fair allocation of on-board resources Fr

[0082] The calculation method is shown in formula (8) and formula (9):

[0083]

[0084] In formula (8), Indicates that subtask j is along p j,k The upload delay to candidate unloading vehicle i, represents candidate unloading vehicle i along p j,k The return delay of the original return result, in formula (9), f i represents the total computing resources of vehicle i, |I * | represents the total number of mission vehicles, |J| max represents the maximum number of subtasks contained in each unloading request, and |I| represents the total number of vehicles;

[0085] 5. Fitness t of particle n n

[0086] The calculation method is shown in formula (10):

[0087]

[0088] In formula (10), represents subtask j along path p j,k The unloading completion delay to vehicle i is calculated as shown in formula (11): is a binary variable, indicating whether subtask j follows path p for particle n. j,k Uninstall to p j,k The last node vehicle i, if yes, let Otherwise, let The calculation method is shown in formula (12):

[0089]

[0090] In formula (12), ψ n,j represents the offloading path of subtask j in particle n;

[0091] 6. Unloading decision vector ψ for each particle n n The component ψ n,j and the migration vector s of each particle n n

[0092] The calculation method is shown in formula (13) and formula (14):

[0093]

[0094] In formula (13), s n,j Represents ψ n,j The transformation step size, |P j | represents the total number of candidate offloading paths for subtask j. In formula (14), round() represents rounding, ω represents the inertia factor, ε1 and ε2 represent the learning factors, and rand() represents a random number in the numerical interval (0,1). represents the optimal unloading decision of particle n, ψ best represents the global optimal unloading decision of particle swarm N;

[0095] 7. Sub-algorithm 1: Establish a set of candidate offloading paths P for subtask j j ={p j,k}

[0096] Step 1: Initialize the set of all vehicles I = {i}, where i is any vehicle, and the set of candidate unloading vehicles The set of candidate offloading paths for subtask j The maximum tolerable delay of task offloading t max ;

[0097] Step 2: If Take the first element i from the set I, according to the amount of data a of subtask j j , Required virtual network function VNF type v j , computational complexity c j , calculate the minimum computing resources that vehicle i needs to allocate for subtask j Jump to step 3, otherwise, jump to step 4;

[0098] Step 3: If the remaining available computing resources f of vehicle i i r Greater than or equal to Add vehicle i to the candidate unloading vehicle set I′ and jump to step 2; otherwise, jump to step 2;

[0099] Step 4: If Take the first element i from the set I′ and use the depth-first search algorithm to construct the task vehicle i * The set P′ of all feasible paths to candidate unloading vehicle i j ={p j,k}, where p j,k If there is a feasible offloading path coded as k for subtask j, jump to step 5; otherwise, jump to step 9;

[0100] Step 5: If According to the amount of data a of subtask jj , Required virtual network function VNF type v j , computational complexity c j and the remaining available computing resources f of candidate unloading vehicle i i r , calculate the minimum computational delay of candidate unloading vehicle i to process subtask j Jump to step 6, otherwise, jump to step 4;

[0101] Step 6: P′ j Each path p in j,k , let subtask j be along p j,k The upload delay to candidate unloading vehicle i is Candidate unloading vehicle i along p j,k The estimated return delay of the original return result is Among them, r l is the transmission rate of physical link l, σ j is the ratio of the uploaded and returned data volumes of subtask j, r min is the minimum transmission rate between vehicles, and the path p is calculated j,k Duration Delay the completion time of offloading subtask j Greater than All p j,k From the set P′ j Delete from;

[0102] Step 7: If P′ j Each path p in j,k , calculated in The minimum computing resources required for candidate unloading vehicle i to allocate for subtask j within time Calculate the computing resources f that candidate unloading vehicle i needs to allocate for subtask j under fair allocation of on-board resources Fr , jump to step 8, otherwise, jump to step 4;

[0103] Step 8: From P′ j Choose to meet A path p j,k Add candidate offloading path set P j , jump to step 4;

[0104] Step 9: Output candidate offloading path set P j , the step ends.

[0105] A method for offloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing specifically comprises the following steps:

[0106] Step 1: Use particle swarm algorithm to establish vehicle network task offloading optimization model, assuming that task vehicle i * The unloading request J = {j}, where j is a parallel subtask, the particle swarm is a set N = {n}, and the unloading decision vector, migration vector and optimal fitness of each particle n in N are ψ n =(ψ n,j | j∈J ),s n =(s n,j | j∈J )and Among them, ψ n,j is the offloading path of subtask j in particle n, s n,j is n,j The transformation step size is η, and the maximum number of iterations of the initialized particle swarm is η max , iteration count variable η=0, temporary set J′=J;

[0107] Step 2: If Take the first element j from the set J′, and according to the amount of data a of subtask j j , the required virtual network function VNF type v j , computational complexity c j , call sub-algorithm 1 to establish a candidate offloading path set P for subtask j j ={p j,k}, where p j,k is a candidate offloading path coded as k for subtask j, jump to step 2, otherwise, jump to step 3;

[0108] Step 3: For each particle n in N, from each set P j | j∈J Choose any path p j,k Initialize the offload decision vector From the value interval [0,|P j |) randomly selects an integer to initialize the migration vector s n Each component s in n,j | j∈J , calculate the particle fitness t n , initialize the optimal fitness of particle n Optimal unloading decision for particle n

[0109] Step 4: Let the global optimal unloading decision of particle swarm N be Iteration count variable η=η+1, if η≤η max , let the temporary set N′=N, jump to step 5, otherwise, jump to step 7;

[0110] Step 5: If Take the first element n from the set N′ and update the unloading decision vector ψ n and the migration vector s n , calculate the particle fitness t n , jump to step 6, otherwise, jump to step 4;

[0111] Step 6: If Let the optimal fitness of particle n be Optimal unloading decision for particle n Jump to step 5, otherwise, jump to step 5;

[0112] Step 7: Output the global optimal offloading decision ψ best , the step ends.

[0113] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions.

[0114] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0115] The above embodiments should be understood to be only used to illustrate the present invention and not to limit the protection scope of the present invention. After reading the contents of the present invention, technicians can make various changes or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims

1. A method for offloading tasks of vehicles with non-uniform speed in complex road conditions based on VNF instance sharing, characterized in that: The following steps are involved:

101. The particle swarm algorithm is used to establish a vehicle network task offloading optimization model. Assume that the offloading request of task vehicle i* is J={j}, where j is a parallel subtask, the particle swarm is a set N={n}, and the offloading decision vector, migration vector and optimal fitness of each particle n in N are ψ n =(ψ n，j | j∈J ),s n =(s n，j | j∈J )and Among them, ψ n，j is the offloading path of subtask j in particle n, s n，j is n，j The transformation step size is η, and the maximum number of iterations of the initialized particle swarm is η max , iteration count variable η=0, temporary set J′=J; 102. If Take the first element j from the set J′, according to the amount of data a of the subtask j , the required virtual network function VNF type v j , computational complexity c j , establish a candidate offloading path set P for subtask j j ={p j，k }, where p j，k If it is a candidate offloading path coded as k for subtask j, jump to step 102; otherwise, jump to step 103; 103. For each particle n in N, from each set P j | j∈J Choose any path p j，k Initialize the offload decision vector From the value interval [0, |P j |) randomly selects an integer to initialize the migration vector s n Each component s in n，j | j∈J , calculate the particle fitness t n , initialize the optimal fitness of particle n Optimal unloading decision for particle n 104. Let the global optimal unloading decision of particle swarm N be Iteration count variable η=η+1, if η≤η max , let the temporary set N′=N, jump to step 105, otherwise, jump to step 107; 105. If Take the first element n from the set N′ and update the unloading decision vector ψ n and the migration vector s n , calculate the particle fitness t n , jump to step 106, otherwise, jump to step 104; 106. If Let the optimal fitness of particle n be Optimal unloading decision for particle n Jump to step 105, otherwise, jump to step 105; 107. Output the global optimal unloading decision ψ best , the step ends.

2. According to claim 1, a method for offloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing is characterized in that: In step 102, a candidate offloading path set P is established for subtask j. j ={p j，k }, specifically including the following steps: 1) Initialize the set of all vehicles I = {i}, where i is any vehicle, and the set of candidate unloading vehicles The set of candidate offloading paths for subtask j The maximum tolerable delay of task offloading t max ; 2) If Take the first element i from the set I, according to the subtask, the amount of data a j , Required virtual network function VNF type v j , computational complexity c j , calculate the minimum computing resources that vehicle i needs to allocate for subtask j Jump to step 3), otherwise, jump to step 4); 3) If the remaining available computing resources f of vehicle i i r Greater than or equal to Add vehicle i to the candidate unloading vehicle set I′ and jump to step 2); otherwise, jump to step 2); 4) If Take the first element i from the set I′ and use the depth-first search algorithm to construct the task vehicle i * The set P′ of all feasible paths to candidate unloading vehicle i j ={p j，k }, where p j，k is a feasible offloading path coded as k for subtask j, jump to step 5), otherwise, jump to step 9); 5) If According to the data volume a of subtask j j , Required virtual network function VNF type v j , computational complexity c j and the remaining available computing resources f of candidate unloading vehicle i i r , calculate the minimum computational delay of candidate unloading vehicle i to process subtask j Jump to step 6), otherwise, jump to step 4); 6) For P′ j Each path p in j，k , let subtask j be along p j，k The upload delay to candidate unloading vehicle i is Candidate unloading vehicle i along p j，k The estimated return delay of the original return result is Among them, r l is the transmission rate of the physical link, σ j is the ratio of the uploaded and returned data volumes of subtask j, r min is the minimum transmission rate between vehicles, and the path p is calculated j，k Duration Delay the completion time of the uninstallation of the subtask Greater than All p j,k From the set P j ' is deleted; 7) If P j Each path p in ′ j,k , calculated in The minimum computing resources required for candidate unloading vehicle i to allocate for subtask j within time Calculate the computing resources f that candidate unloading vehicle i needs to allocate for subtask j under fair allocation of on-board resources Fr , jump to step 8), otherwise, jump to step 4); 8) From P j 'Select the one that satisfies A path p j,k Add candidate offloading path set P j , jump to step 4); 9) Output candidate offloading path set P j , the step ends.

3. According to claim 2, a method for offloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing is characterized in that: The minimum computing resources that vehicle i needs to allocate for subtask j in step 2) The calculation method is shown in formula (1): In formula (1), a j represents the amount of data for subtask j, c j represents the computational complexity of subtask j, t max represents the maximum tolerable delay of task offloading, f0 represents the basic computing resources required to instantiate VNF, and λ i,j is a binary variable, indicating whether vehicle i has a sharable VNF instance. If not, the corresponding VNF needs to be instantiated for subtask j. Let λ i,j =1, otherwise, let λ i,j =0, the calculation method is shown in formula (2): In formula (2), v j represents the VNF type required for subtask j, V i Represents the set of VNF types contained in vehicle i.

4. According to claim 3, a method for offloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing is characterized in that: The minimum computational delay of processing subtask j of candidate unloading vehicle i in step 5) The calculation method is shown in formula (3): In formula (3), f i r represents the remaining available computing resources of candidate unloading vehicle i.

5. According to claim 4, a method for offloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing is characterized in that: In step 6), path p j,k Duration The calculation method is shown in formula (4): In formula (4), τ l represents the link duration of the physical link l between vehicle i′ and vehicle i″, where vehicle i′ and vehicle i″ are two vehicles that can communicate directly, τ l The value of is calculated by formula (5): In formula (5), D represents the effective communication range of the vehicle, (x i′ ,y i′ )、(x i″ ,y i″ ) represent the current position coordinates of vehicle i′ and vehicle i″, μ max represents the maximum speed of the vehicle, γ i′ , γ i″ They respectively represent the angles between the driving directions of vehicle i′ and vehicle i″ and the positive direction of the x-axis after the direction change, and are calculated as shown in formula (6): In formula (6), α i′ , α i″ They represent the angles between the current driving directions of vehicle i′ and vehicle i″ and the positive direction of the x-axis, and β represents the angle change when vehicle i′ and vehicle i″ change from the current driving directions to the opposite directions. The calculation method is shown in formula (7): In formula (7), θ represents the instantaneous maximum steering angle of the vehicle.

6. According to claim 5, a method for offloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing is characterized in that: In the step 7) The minimum computing resources required for candidate unloading vehicle i to allocate for subtask j within time The computing resources f that candidate unloading vehicle i needs to allocate for subtask j under fair allocation of on-board resources Fr The calculation methods are shown in formula (8) and formula (9) respectively: In formula (9), f i represents the total computing resources of vehicle i, I * represents the total number of mission vehicles, J max It represents the maximum number of subtasks contained in each unloading request, and I represents the total number of vehicles.

7. According to claim 1, a method for offloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing is characterized in that: The fitness of particle n in step 103 is t n The calculation method of is shown in formula (10): In formula (10), represents subtask j along path p j,k The unloading completion delay to vehicle i is calculated as shown in formula (11): is a binary variable, indicating whether subtask j follows path p for particle n. j,k Uninstall to p j,k The last node vehicle i, if yes, let Otherwise, let The calculation method is shown in formula (12):

8. According to claim 1, a method for offloading tasks of vehicles with non-uniform speed under complex road conditions based on VNF instance sharing is characterized in that: In step 105, the unloading decision vector ψ of each particle n is updated n , that is, update ψ n Each component ψ n,j , the calculation method is shown in formula (13): Update the migration vector s of each particle n n The calculation method of is shown in formula (14): In formula (14), round() means rounding to the nearest integer, ω means the inertia factor, ε1 and ε2 mean the learning factors, and rand() means a random number in the numerical interval (0,1).

Images