Vehicle cooperative task unloading method based on incentive mechanism

Abstract

The invention belongs to the technical field of vehicle collaborative task unloading, and particularly relates to a vehicle collaborative task unloading method based on an incentive mechanism, which comprises the following steps of: 1, establishing a network model, a computable vehicle utility model and an unloaded vehicle utility model; step 2, proposing a vehicle private information estimation method based on an exploration and development algorithm, predicting private information of vehicles, performing type division on the vehicles with idle computing resources, and designing an optimal contract item; 3, satisfying the constraints of individual rationality, excitation compatibility and monotonicity, and establishing an optimization problem of maximizing the expected utility of the unloaded vehicle; according to the mobile vehicle private information estimation method, under the condition that the IR and IC constraint conditions are met, the expected utility of unloading the vehicle is maximized, and in order to solve the problem that the calculable vehicle private information is unknown, the mobile vehicle private information estimation algorithm is provided, so that the cost can be effectively reduced, and the benefits of unloading the vehicle can be effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle collaborative task unloading, and in particular relates to a vehicle collaborative task unloading method based on an incentive mechanism. Background Art

[0002] In the scenario of information asymmetry, mobile vehicles usually do not proactively provide their private information to vehicles that need resource requests, such as preferences for resource sharing and the total amount of available resources. Moreover, since computing tasks consume resources and energy, selfish vehicles may not proactively share their resources without any incentives. Contract theory is considered to be a powerful tool for microeconomics to solve the problem of information asymmetry. Therefore, in recent years, some researchers have conducted in-depth research on the incentive mechanism of mobile vehicles in vehicular edge computing (VEC) networks.

[0003] By adopting centralized infrastructure in VEC networks, contracts are designed for mobile vehicles to incentivize computable vehicles to participate in resource sharing. In this regard, some research results are as follows: (1) Computational resource allocation and task allocation optimization in vehicle fog computing. In order to incentivize vehicles to participate in resource sharing, an effective incentive mechanism based on contract theory is designed. The contract mechanism stipulates the relationship between the amount of computing resources shared by vehicles and rewards. The contract items are designed by the base station and the expected utility of the BS is maximized. Considering the particularity of the high-speed mobile environment and the possibility that vehicles may deceive the server. (2) Task division and offloading scheme in vehicle-mounted edge computing networks. By designing a channel model to capture the dynamic characteristics of the channel, and using contract theory to solve the problem of vehicle deception, the contract includes the division point of the task and the price of the roadside unit renting the vehicle. (3) Vehicle threat sharing based on federated learning. According to the multi-dimensional private characteristics of vehicles, an incentive mechanism based on contract theory is designed. Considering the situation where selfish vehicles use cost-related private information to obtain private benefits, vehicles are incentivized to provide resources by utilizing self-disclosure attributes.

[0004] Due to the mobility of vehicles, the network topology between vehicles may change in different time slots, and the private information of vehicles may also change. Therefore, in the scenario of dynamic changes in mobile vehicles, designing a reasonable and effective incentive mechanism is still a problem that needs to be solved. Existing work designs contracts based on the characteristics of each vehicle type, such as dividing vehicles into types based on private characteristics such as the maximum remaining computing resources, stay time, acceleration or cost of the vehicle, and solving the problem of information asymmetry by designing a series of contract items that can be selected by computable vehicles.

[0005] However, due to the uneven deployment of base stations, the communication range cannot cover a large geographical area, which makes it easy for mobile vehicles in non-communication coverage areas to be unwilling to share idle resources, and unloaded vehicles cannot perform task calculations in time, leading to road hazards. In addition, due to the uneven distribution of task traffic, it is easy to cause infrastructure network congestion during peak hours, and it is impossible to design an effective incentive mechanism to encourage mobile computable vehicles to participate in resource sharing. Most of the existing literature uses centralized infrastructure to design incentives, and rarely considers the incentive design between vehicles in distributed VEC networks. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a vehicle collaborative task unloading method based on an incentive mechanism, comprising the following steps:

[0007] Step 1: Establish a network model, a computable vehicle utility model and an unloaded vehicle utility model;

[0008] Step 2: A vehicle private information estimation method based on exploration and exploitation algorithm is proposed to predict the private information of vehicles, classify the types of vehicles with idle computing resources, and design the optimal contract items;

[0009] Step 3: Satisfy the constraints of individual rationality, incentive compatibility and monotonicity, and establish an optimization problem to maximize the expected utility of unloading vehicles;

[0010] Step 4: For the optimization problem, reduce the number of IR and IC constraints by exploring the relationship between adjacent vehicle types;

[0011] Step 5: Use the Lagrange multiplier method in the convex optimization algorithm to solve the optimization problem.

[0012] Preferably, the establishing of a computable vehicle utility model comprises:

[0013] The computable vehicle utility model is expressed as a function:

[0014] O h =η h -E h

[0015] Among them, O h represents the function used to calculate the vehicle utility model, η h E represents the unit communication association time reward given by the unloading vehicle to the computable vehicle; h It means that the energy consumption price paid by the vehicle in the unit communication association time can be calculated;

[0016] E h express:

[0017]

[0018] Among them, ε h Represents the unit cost of energy consumption, s h Indicates the calculable vehicle capacitance coefficient, f h represents the computing resources that can be shared by computing vehicles, t h Indicates the communication association time between vehicles;

[0019] t h express:

[0020]

[0021] Where S represents the relative displacement between the calculable vehicle and the unloaded vehicle, v h represents the speed of the calculable vehicle, and v represents the speed of the unloaded vehicle;

[0022] Unloading vehicles distinguish computable vehicles with different association times, energy consumption costs, and capacitance coefficients by the type of computable vehicles, classify vehicles into multiple discrete types, and represent the vehicle type βl as:

[0023]

[0024] Among them, t h represents the communication association time between vehicles, ε h Represents the unit cost of energy consumption, s h represents the capacitance coefficient;

[0025] Therefore, the vehicle type is β l The computable vehicle utility function is updated as:

[0026]

[0027] Among them, O l represents the updated computable vehicle utility model function, η l Indicates that after the update, uninstalling the vehicle gives the vehicle type to β l The unit communication association time reward of the vehicle can be calculated, f l Indicates that the updated vehicle type is β l Computational resources that can be shared by computing vehicles, β l Indicates the vehicle type.

[0028] Preferably, establishing the unloading vehicle utility model comprises:

[0029] The benefit of unloading a vehicle is related to the computational latency of the task. The computational time required for the unloading vehicle to offload the task to the local execution is expressed as:

[0030]

[0031] Among them, f 0 represents the computing power of the unloaded vehicle, d m represents the size of the mth task, k m Indicates the number of CPU cycles required to process a unit task;

[0032] Given a task size, the computational delay of the task is The unloading vehicle unloads the task to the vehicle type l, and the function of calculating and establishing the unloading vehicle utility model is updated as follows:

[0033]

[0034] Among them, U l represents the updated unloading vehicle utility model function, η l Indicates that the vehicle type is β l The time price reward of the calculable vehicle, α represents the unit benefit factor of delay, and f 0 represents the computing power of the unloaded vehicle, f l Indicates that the vehicle type is β l Computable vehicles can share computing resources.

[0035] Preferably, the vehicle private information estimation method based on the exploration and exploitation algorithm comprises:

[0036] The private information of the vehicle represents the unit cost of energy consumption. At the initial stage, the maximum and minimum values ​​of the unit cost of energy consumption are known to the unloaded vehicle, and it is divided into L equal intervals, each of which corresponds to the initial energy consumption cost ε of a vehicle type l ;

[0037] Unloading the vehicle will contract items (f l , η l ) is broadcast to the computable vehicles. After receiving the feedback, the average unit cost of energy consumption of all computable vehicles in vehicle type l at time slot t is calculated. l,t , uncertainty measure A for vehicle type l，t , update the unit cost value of energy consumption ε l ,express:

[0038]

[0039] Among them, ψ l，t and A l，t Respectively represent:

[0040]

[0041]

[0042] Among them, πl Indicates that the vehicle type is β l The vehicle probability, b l Indicates vehicle type β l The balance factor, |H(t)| represents the total number of computable vehicle sets at time slot t, Q l，t Indicates acceptance of the contract at time slot t (f l , η l ), the vehicle type is β l The average energy consumption of the calculable vehicle set, ε l，h It indicates that the unit cost of energy consumption of vehicle h can be calculated.

[0043] Preferably, the constraints of individual rationality, incentive compatibility and monotonicity are satisfied, and the optimization problem of maximizing the expected utility of unloading vehicles is established as:

[0044]

[0045]

[0046] C3: β 1 <...<β l <...<β L , 1≤l≤L

[0047] Among them, C1 represents the IR constraint, C2 represents the IC constraint, C3 represents the constraint that vehicle types are sorted in ascending order, L represents the size of the vehicle type set, l and k represent the index of the vehicle type set, and β l , β k Indicates the vehicle type, f 0 represents the computing power of the unloaded vehicle, f l 、f k Indicates that the vehicle type is β l , β k The computing resources that can be shared by the computing vehicles, α represents the unit benefit factor of delay, η l , η k Indicates that the unloading vehicle is given a vehicle type of β l , β k The unit communication association time reward of the vehicle can be calculated

[0048] Preferably, the number of constraints of IR and IC is reduced by exploring the relationship between adjacent vehicle types, and the optimization problem is updated as follows:

[0049]

[0050]

[0051] C3: β 1 <...<βl <...<β L , 1≤l≤L

[0052] Among them, C1 represents the IR constraint, C2 represents the IC constraint, C3 represents the vehicle type sorting constraint in ascending order, and β l Indicates the vehicle type, f 0 represents the computing power of the unloaded vehicle, f l Indicates that the vehicle type is β l The computing resources that can be shared by the computing vehicles, α represents the unit benefit factor of delay, η l Indicates that the unloading vehicle is given a vehicle type of β l The unit communication association time reward of the vehicle can be calculated, and L represents the size of the vehicle type set.

[0053] Beneficial effects of the present invention:

[0054] The present invention provides a vehicle collaborative task unloading method based on an incentive mechanism, and studies how to design a reasonable and effective incentive mechanism to encourage computable vehicles to participate and share resources. First, the utility model of the unloading vehicle and the utility model of the computable vehicle are modeled. By comprehensively considering the impact of vehicle mobility and private information on incentives, a contract-based incentive mechanism scheme is designed to maximize the expected utility of the unloading vehicle under the conditions of satisfying IR and IC constraints. Secondly, in order to solve the problem of unknown private information of computable vehicles, a mobile vehicle private information estimation algorithm is proposed. Finally, the feasibility of the proposed scheme is verified by simulation. Compared with other incentive schemes, the scheme proposed by the present invention can effectively reduce costs and increase the benefits of unloading vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a principle block diagram of the present invention;

[0056] Figure 2 This is a network model diagram in the present invention;

[0057] Figure 3 A comparison chart of unit rewards obtained by different calculable vehicle types in the present invention;

[0058] Figure 4 A comparison diagram of vehicle computing resources obtained for different computable vehicle types in the present invention;

[0059] Figure 5 This is a comparison chart of the benefits of the vehicles that can be calculated in the present invention.

[0060] Figure 6 The cost comparison diagram of the vehicle that can be calculated in the present invention is

[0061] Figure 7 This is a comparison chart of the benefits of unloading vehicles in the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] Various embodiments of the present invention may be presented in the form of a range. It should be understood that the description in the form of a range is only for convenience and simplicity and should not be understood as a rigid limitation on the scope of the present invention; therefore, it should be considered that the range description has specifically disclosed all possible sub-ranges and single values ​​within the range. For example, it should be considered that the range description from 1 to 6 has specifically disclosed sub-ranges, such as from 1 to 3, from 1 to 4, from 1 to 5, from 2 to 4, from 2 to 6, from 3 to 6, etc., as well as single numbers within the range, such as 1, 2, 3, 4, 5 and 6, regardless of the range. In addition, whenever a numerical range is indicated in the present invention, it is intended to include any quoted numbers (fractions or integers) within the indicated range. Unless otherwise specified, the various raw materials, reagents, instruments and equipment used in the present invention can be purchased from the market or can be prepared by existing equipment.

[0064] In the present invention, in the absence of any contrary description, the directional words used, such as "upper" and "lower", are specifically the directions of the drawings in the accompanying drawings. In addition, in the present invention, the terms "include", "comprise", etc. refer to "including but not limited to". In the present invention, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. In the present invention, "and / or" describes the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. Wherein A and B can be singular or plural. In the present invention, "at least one" refers to one or more, and "plural" refers to two or more. "At least one", "at least one of the following" or similar expressions refer to any combination of these items, including any combination of singular items or plural items. For example, "at least one of a, b, or c" or "at least one of a, b and c" can both mean: a, b, c, ab, i.e. a and b, ac, bc or abc, where a, b, c can be single or plural, respectively.

[0065] like Figure 1-Figure 7 As shown, an embodiment of the present invention provides a vehicle collaborative task unloading method based on an incentive mechanism, comprising the following steps:

[0066] Step 1: Establish a network model, a computable vehicle utility model and an unloaded vehicle utility model;

[0067] For scenarios with asymmetric information, the mobility and private information of vehicles are considered to establish a network model, a computable vehicle utility model, and an unloaded vehicle utility model.

[0068] Network models such as Figure 2 As shown in Figure 1, a distributed VEC network consisting of an unloading vehicle and multiple computable vehicles is considered. The unloading vehicle can offload multiple subtasks to different computable vehicles. Computable vehicles with idle resources will not unconditionally process the tasks sent by the unloading vehicle and will not actively provide their private information. Therefore, a contract-based incentive mechanism scheme is adopted to incentivize selfish computable vehicles to perform task calculations and provide their private information.

[0069] When unloading a vehicle at time slot T, multiple subtasks may be generated for task unloading calculation. The unloading task is represented as: Subtask d m Expressed as a binary: d m =(d m ,k m ), where d m represents the subtask size, k m It represents the CPU cycles required to process a unit task. When the unloading vehicle is in time slot t, there may be multiple computable vehicles moving in the same direction within its communication range, which can be represented by a set: H(t) = {H h},h∈H,H={1,...,h,...,H}.

[0070] Step 2: A vehicle private information estimation method based on exploration and exploitation algorithm is proposed to predict the private information of vehicles, classify the types of vehicles with idle computing resources, and design the optimal contract items;

[0071] Considering that mobile vehicles will not provide their private information, a vehicle private information estimation method based on exploration and exploitation algorithm is proposed to predict the private information of vehicles, so that vehicles with idle computing resources can be classified into types to design the optimal contract items.

[0072] Step 3: Satisfy the constraints of individual rationality, incentive compatibility and monotonicity, and establish an optimization problem to maximize the expected utility of unloading vehicles;

[0073] In order to motivate computable vehicles to participate and share resources and perform task computation, an optimization problem of maximizing the expected utility of unloading vehicles is established under the constraints of individual rationality, incentive compatibility and monotonicity;

[0074] Step 4: For the optimization problem, reduce the number of IR and IC constraints by exploring the relationship between adjacent vehicle types;

[0075] Step 5: Use the Lagrange multiplier method in the convex optimization algorithm to solve the optimization problem.

[0076] The establishing of a computable vehicle utility model comprises:

[0077] Computable vehicles consume their own energy when performing task calculations. Therefore, the utility function of a computable vehicle is defined as the reward obtained per unit communication association time minus the energy consumption price paid when performing task calculations. The computable vehicle utility model is represented by the function:

[0078] O h =η h -E h

[0079] Among them, O h represents the function used to calculate the vehicle utility model, η h E represents the unit communication association time reward given by the unloading vehicle to the computable vehicle; h It means that the energy consumption price paid by the vehicle in the unit communication association time can be calculated;

[0080] E h express:

[0081]

[0082] Among them, ε h Indicates the unit cost of energy consumption, s h Indicates the calculable vehicle capacitance coefficient, f h represents the computing resources that can be shared by computing vehicles, t h Indicates the communication association time between vehicles;

[0083] When the relative speed between the calculable vehicle and the unloading vehicle is positive, the relative displacement is: Otherwise, the relative displacement is: Among them, (x, y) is the position coordinate of the main vehicle, (x h ,y h ) is the position coordinate of the hth computable vehicle, t h express:

[0084]

[0085] Where S represents the relative displacement between the calculable vehicle and the unloaded vehicle, v h represents the speed of the calculable vehicle, and v represents the speed of the unloaded vehicle;

[0086] The longer the communication association time t between vehicles is, the more stable the communication link between V2V will be, the greater the willingness of the unloading vehicle to unload the computable vehicle will be, and it will also be willing to pay a higher reward.

[0087] Since the communication range of the unloading vehicle is limited, the number of computable vehicles within the communication range is usually limited, so the type set of computable vehicles belongs to a discrete finite space. In the scenario of information asymmetry in the VEC network, all computable vehicles are divided into L types according to the communication association time, capacitance coefficient and unit cost of energy consumption of the computable vehicles. The unit cost of energy consumption is regarded as the private information of the computable vehicle and is obtained through the private information estimation algorithm.

[0088] Unloading vehicles use the type of computable vehicles to distinguish the computable vehicle set with different communication association time, capacitance coefficient and energy consumption unit cost, divide the vehicles into multiple discrete types, and divide the vehicle type β l express:

[0089]

[0090] Among them, t h represents the communication association time between vehicles, ε h Indicates the unit cost of energy consumption, s h represents the capacitance coefficient;

[0091] Among them, t h represents the communication association time between vehicles, ε h Indicates the unit cost of energy consumption, s h represents the capacitance coefficient;

[0092] The vehicle type set represents: All computable vehicle types are sorted in ascending order, i.e., β 1 <...<β l <...<β L , 1≤l≤L.

[0093] Therefore, the vehicle type is β l The computable vehicle utility function is updated as:

[0094]

[0095] Among them: l represents the updated computable vehicle utility model function, η lIndicates that after the update, uninstalling the vehicle gives the vehicle type to β l The unit communication association time reward of the vehicle can be calculated, f l Indicates that the updated vehicle type is β l Computational resources that can be shared by computing vehicles, β l Indicates the vehicle type.

[0096] The establishing of the unloading vehicle utility model comprises:

[0097] The benefit of unloading a vehicle is related to the computational latency of the task. The computational time required for the unloading vehicle to offload the task to the local execution is expressed as:

[0098]

[0099] Among them, f 0 represents the computing power of the unloaded vehicle, d m represents the size of the mth task, k m Indicates the number of CPU cycles required to process a unit task;

[0100] Given a task size, the computational delay of the task is In a linear relationship, the unloading vehicle unloads the task to the vehicle type βl, and the function of calculating the utility model of the unloading vehicle is updated as follows:

[0101]

[0102] Among them, U l represents the updated unloading vehicle utility model function, η l Indicates that the vehicle type is β l The time price reward of the calculable vehicle, α represents the unit benefit factor of delay, and f 0 represents the computing power of the unloaded vehicle, f l Indicates that the vehicle type is β l Computable vehicles can share computing resources.

[0103] The algorithm for predicting the private information of a vehicle includes:

[0104] The private information of the vehicle represents the unit cost of energy consumption. At the initial stage, the maximum and minimum values ​​of the unit cost of energy consumption are known to the unloaded vehicle, and it is divided into L equal intervals, each of which corresponds to the initial energy consumption cost ε of a vehicle type l ;

[0105] Unloading the vehicle will contract items (f l , η l ) is broadcast to the computable vehicles. After receiving the feedback, the average unit cost of energy consumption of all computable vehicles in vehicle type l at time slot t is calculated.l，t , uncertainty measure A for vehicle type l，t , update the energy consumption unit cost value ε l ,express:

[0106]

[0107] Among them, ψ l，t and A l，t Respectively represent:

[0108]

[0109]

[0110] Among them, π l Indicates that the vehicle type is β l The vehicle probability, b l Indicates vehicle type β l The balance factor, |H(t)| represents the total number of computable vehicle sets at time slot t, Q l , t means accepting the contract at time slot t (f l , η l ), the vehicle type is β l The average energy consumption of the calculable vehicle set, ε l，h It indicates that the unit cost of energy consumption of vehicle h can be calculated.

[0111] In the case of information asymmetry, unloading vehicles will maximize their own benefits. At the same time, computable vehicles may deliberately provide false private information in order to obtain the highest reward. To solve this problem, contract theory is used to design a suitable incentive mechanism to encourage computable vehicles to participate in sharing resources and provide correct private information.

[0112] The unloading vehicle designs contract items of different vehicle types in order to recruit computable vehicles willing to help perform task calculations. The unloading vehicle designs contracts by maximizing its benefits, and each contract item represents a different type of computable vehicle.

[0113] When a computable vehicle chooses to accept a contract, it is rational and will choose a contract that can guarantee its own interests, and will not choose a contract that has negative effects on it. Therefore, the design of the contract for the computable vehicle needs to meet the IR constraint, that is, In the case of information asymmetry, the computable vehicle will not actively share its computing resources and private information with the unloading vehicle, but the computable vehicle will provide its private information in order to select the best contract item for itself. Based on this, the design of the contract also needs to meet the IC constraint, so that the computable vehicle can select the most suitable contract item to obtain the maximum benefit, that is,

[0114] The above mentioned constraints of individual rationality, incentive compatibility and monotonicity are satisfied to establish the optimization problem of maximizing the expected utility of unloading vehicles:

[0115]

[0116]

[0117] C3: β 1 <...<β l <...<β L , 1≤l≤L

[0118] Among them, C1 represents the IR constraint, C2 represents the IC constraint, C3 represents the constraint that vehicle types are sorted in ascending order, L represents the size of the vehicle type set, l and k represent the index of the vehicle type set, and β l , β k Indicates the vehicle type, f 0 represents the computing power of the unloaded vehicle, f l 、f k Indicates that the vehicle type is β l , β k The computing resources that can be shared by the computing vehicles, α represents the unit benefit factor of delay, η l , η k Indicates that the unloading vehicle is given a vehicle type of β l , β k The unit communication association time reward of the vehicle can be calculated.

[0119] By exploring the relationship between adjacent vehicle types, the number of IR and IC constraints is reduced, and the optimization problem is updated as follows:

[0120]

[0121]

[0122] C3: β 1 <...<β l <...<β L , 1≤l≤L

[0123] Among them, C1 represents the IR constraint, C2 represents the IC constraint, C3 represents the vehicle type sorting constraint in ascending order, and β l Indicates the vehicle type, f 0 represents the computing power of the unloaded vehicle, f l represents the computing resources that can be shared by the computable vehicles of vehicle type βl, α represents the unit benefit factor of delay, η l Indicates that the unloading vehicle is given a vehicle type of β lThe unit communication association time reward of the vehicle can be calculated, and L represents the size of the vehicle type set.

[0124] Specifically: Since the constraints of the above optimization problem are relatively complex and difficult to solve directly, it is necessary to simplify the IR and IC constraints. Therefore, the following lemma is used to simplify the IR and IC constraints before solving the optimization problem.

[0125] Lemma 1: For any contract that satisfies the IC constraint, have

[0126] Proof: According to the IC constraint, we have: and Two inequalities, by adding the inequalities, we get (η l -η k )(β l -β t )≥0, so when β l ≥β k When η l ≥η k .

[0127] Lemma 2: For any contract (f l , η l ), if and only if have when have

[0128] prove:

[0129] Sufficiency: According to the IC constraints, if there is:

[0130]

[0131] That is, we get:

[0132]

[0133] Therefore, when η l >η k When f l >f k When η l =η k When f l =f k .

[0134] Necessity: According to IC constraints, if there are:

[0135]

[0136] That is, we get:

[0137]

[0138] Therefore, when f l >f k When η l >η k When f l =f k When η l =η k .

[0139] Lemma 3: For any IR-constrained contract, if vehicles of type 1 satisfy the IR constraint, then vehicles of all types satisfy the IR constraint, that is:

[0140] Proof: According to IC constraint and β monotonically increasing, we can get

[0141] Lemma 4: The IC constraint is monotonic and can be divided into local downward incentive compatible constraints:

[0142]

[0143] Similarly, it can be simplified to a local upward incentive compatible constraint:

[0144]

[0145] Prove the local downward incentive compatibility constraint: According to the IC constraint, we can get:

[0146]

[0147]

[0148] Since IC is monotonic, according to Lemma 1 and Lemma 2, we can get:

[0149]

[0150] Right now:

[0151]

[0152]

[0153] Therefore, it can be deduced that:

[0154]

[0155] The local downward incentive compatibility constraint has been proved. Similarly, the local upward incentive compatibility constraint can be proved.

[0156] By simplifying the IR and IC constraints mentioned above, the optimization problem can be updated as follows:

[0157]

[0158]

[0159] C3: β 1 <...<β l <...<β L , 1≤l≤L

[0160] Among them, C1 represents the IR constraint, C2 represents the IC constraint, there are L IR constraints and L(L-1) IC constraints in total, and C3 represents that the vehicle types are sorted in ascending order;

[0161] L represents the size of the vehicle type set, β l Indicates the vehicle type, f 0 Indicates the computing power of the unloaded vehicle, indicating that the vehicle type is β l The computing resources that can be shared by the computing vehicles, α represents the unit benefit factor of delay, η l Indicates that the unloading vehicle is given a vehicle type of β l The unit communication association time reward of the vehicle can be calculated.

[0162] Embodiment 1:

[0163] Problem Solving:

[0164] Incentive mechanism problem solving:

[0165] The Lagrange multiplier method in the convex optimization algorithm is used to solve the above optimization problem. First, we get f l , l∈L, and then bring it into Get η l , l∈L is the optimal solution, and its algorithm pseudo code is shown in Algorithm 1.

[0166]

[0167]

[0168] Private information estimation algorithm:

[0169] The private information of the vehicle represents the unit cost of energy consumption. At the initial stage t = 1, the maximum unit cost of energy consumption known by the unloaded vehicle is ε max and the minimum value ε min , and there are L types of vehicles. Unloading vehicles will close the interval [ε min , ε max] is divided into L equal intervals, and the minimum value of each interval is used as the initial unit cost value of energy consumption under different vehicle types, that is:

[0170]

[0171] The unloading vehicle designs the contract project and sends it to the computable vehicles within the communication range by broadcasting. When the unloading vehicle receives the feedback from the computable vehicle, the unloading vehicle can obtain the private information of the computable vehicles under different vehicle types. At the end of t = 1, the unit cost value of energy consumption is updated to:

[0172]

[0173] Among them, ψl, 1 Indicates that at time slot t = 1, the energy consumption unit cost ε of vehicle h can be calculated for vehicle type l l，h Find the average value, which means:

[0174]

[0175] Among them, Q l，1 Indicates that the contract is accepted at time slot t = 1 (f l , η l ), the vehicle type is β l The computable vehicle set, |Q l，1 | indicates vehicle type β l The number of l Indicates that the vehicle type is β l The probability of a vehicle |H(1)| represents the total number of computable vehicle sets at time slot t=1.

[0176] A l，1 is the uncertainty measurement factor for vehicle type l at time slot t = 1, expressed as:

[0177]

[0178] Among them, b l Represents the balance factor for vehicle type l.

[0179] When time slot t>1, the unloading vehicle updates the energy unit cost value at that moment according to the energy unit cost obtained in the previous time slot, that is:

[0180]

[0181] Among them, ψ l，t and A l，t Respectively represent:

[0182]

[0183]

[0184] Among them, π l Indicates that the vehicle type is β l The vehicle probability, b l Indicates vehicle type β l The balance factor, |H(t)| represents the total number of computable vehicle sets at time slot t, Q l , t means accepting the contract at time slot t (f l , η l ), the vehicle type is β l The computable vehicle set, ε l，h Indicates that the vehicle type is β l The unit cost of energy consumption of vehicle h can be calculated.

[0185] The private information ε of the vehicle is estimated by the private information estimation algorithm l To make an estimate, the vehicle types can be classified. The private information estimation algorithm is as shown in Algorithm 2:

[0186]

[0187]

[0188] Simulation results:

[0189] The optimal contract items (f l ,η l ), Figure 2 It shows that among different vehicle types, the unit reward η obtained by the vehicle can be calculated l The change of , where the unit reward η of the vehicle can be calculated l Increases with the type of vehicle. Figure 3 According to Lemma 1, under the IR constraint, since the vehicle type is monotonically increasing, the unit reward η of the vehicle can be calculated l It is also monotonically increasing.

[0190] Figure 4 The figure shows the changes in the computing resources shared by different vehicle types. As the number of vehicle types increases, the computing resources shared by the vehicles increases. l will also increase. Figure 4 And IC constraints, we can get the unit reward η of the calculable vehicle l When the number of vehicles increases, the computing resources shared by the vehicles can be calculated. l will also increase, which is consistent with Lemma 2.

[0191] In order to verify that the proposed solution meets the two constraints of IR and IC under information asymmetry, Figure 5 The benefits of computable vehicles when the vehicle types are 3, 5, 7, and 9 are shown, where the maximum benefit values ​​of different types are indicated by black arrows. The benefits of computable vehicles are the largest when they only choose the contract items corresponding to their types, and the benefits are all greater than zero. For example, for a computable vehicle belonging to vehicle type 5, the benefits of the computable vehicle are the highest only when the fifth contract item is selected. Choosing other contract items will reduce its benefits. For a computable vehicle belonging to vehicle type 3, when it does not choose the third contract item and chooses others, the computable vehicle may have a negative benefit.

[0192] Figure 6 It shows how the costs of the three schemes vary for different vehicle types under information asymmetry. Among them, the costs of the three schemes all increase with the increase in vehicle types. This is because with the increase in vehicle types, the computing resources shared by computable vehicles will also increase accordingly, resulting in an increase in the cost paid by the vehicles. Under information asymmetry, the cost of the scheme proposed in this chapter is the lowest, because the scheme proposed in this invention predicts the private information of computable vehicles through Algorithm 2, and designs different rewards by judging different types of vehicles, which can effectively reduce the cost of unloading vehicles. The Stackelberg-IA scheme cannot judge the type of vehicle, and needs to design rewards based on the amount of resources requested. It is impossible to design different reward strategies for vehicle types. The reward of the LC-IA scheme is linearly related to the shared computing resources. It does not consider the impact of the private information of computable vehicles on task offloading. When the shared computing resources increase, the cost paid will also increase accordingly. Therefore, by Figure 6 It can be seen that the costs of these two solutions are higher than the cost of the solution proposed in the present invention.

[0193] Figure 7 The figure shows the changes in the revenue of the unloader under different vehicle types under the three schemes. Figure 7 It can be seen that the revenue of the unloader in the three schemes increases with the increase of vehicle types. Among them, in the LC-LA scheme, when the impact of private information of the vehicle on the revenue is not considered, the unloader only performs linear pricing based on the shared computing resources to maximize its revenue. In the Stackelberg-IA scheme, when the vehicle type is unknown and the revenue is positive, the price strategy analysis is performed according to the required resource conditions. When the price of the calculable vehicle is increased, the revenue of the unloader will be reduced. Therefore, in the scenario of asymmetric information, the revenue of the unloader in the scheme proposed by the present invention is greater than that of the Stackelberg-IA scheme and the LC-LA scheme.

[0194] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined in the present invention can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown in the present invention, but should conform to the widest range consistent with the principles and novel features of the present invention.

Claims

1. A vehicle collaborative task unloading method based on an incentive mechanism, characterized in that: The steps include: Step 1: For the scenario of information asymmetry, considering the mobility and private information factors of vehicles, a network model, a computable vehicle utility model and an unloaded vehicle utility model are established; Step 2: Considering that mobile vehicles will not provide their private information, a vehicle private information estimation method based on exploration and exploitation algorithm is proposed to predict the private information of vehicles and classify the types of vehicles with idle computing resources to design the optimal contract items; Step 3: Satisfy the constraints of individual rationality, incentive compatibility and monotonicity, and establish an optimization problem of maximizing the expected utility of unloading vehicles to incentivize computable vehicles to participate and share resources and perform task calculations; Step 4: For the optimization problem, the optimization problem is updated by exploring the relationship between adjacent vehicle types to reduce the number of constraints on individual rationality and incentive compatibility; Step 5: Use the Lagrange multiplier method in the convex optimization algorithm to solve the updated optimization problem and obtain the optimal vehicle collaborative task unloading solution.

2. The vehicle collaborative task unloading method based on incentive mechanism according to claim 1 is characterized in that: The network model includes: a distributed VEC network of an unloading vehicle and multiple computable vehicles. The unloading vehicle can unload multiple subtasks to different computable vehicles. The computable vehicles with idle resources will not unconditionally process the tasks sent by the unloading vehicle and will not actively provide their private information. A contract-based incentive mechanism scheme is adopted to incentivize selfish computable vehicles to perform task calculations and provide their private information.

3. The vehicle collaborative task unloading method based on incentive mechanism according to claim 1 is characterized in that: The establishing of a computable vehicle utility model comprises: The computable vehicle utility model function is expressed as: THE h =η h -AND h Among them, O h represents the function used to calculate the vehicle utility model, η h E represents the unit communication association time reward given by the unloading vehicle to the computable vehicle; h It means that the energy consumption price paid by the vehicle in the unit communication association time can be calculated; E h express: Among them, ε h Indicates the unit cost of energy consumption, s h Indicates the calculable vehicle capacitance coefficient, f h represents the computing resources that can be shared by computing vehicles, t h Indicates the communication association time between vehicles; t h express: Where S represents the relative displacement between the calculable vehicle and the unloaded vehicle, ν h represents the speed of the computable vehicle, ν represents the speed of the unloaded vehicle; Unloading vehicles distinguish computable vehicles with different association times, energy consumption costs and capacitance coefficients by the type of computable vehicles, classify vehicles into multiple discrete types, and classify vehicle types β l express: Therefore, the vehicle type is β l The computable vehicle utility function is updated as: Among them, O l represents the updated computable vehicle utility model function, η l Indicates that after the update, uninstalling the vehicle gives the vehicle type to β l The unit communication association time reward of the vehicle can be calculated, f l Indicates that the updated vehicle type is β l Computational resources that can be shared by computing vehicles.

4. The vehicle collaborative task unloading method based on incentive mechanism according to claim 1 is characterized in that: The establishing of the unloading vehicle utility model comprises: The benefit of unloading a vehicle is related to the computational latency of the task. The computational time t0 required for the unloading vehicle to offload the task to the local execution is expressed as: Where f0 represents the computing power of the unloaded vehicle, d m represents the size of the mth task, k m Indicates the number of CPU cycles required to process a unit task; Given a task size, the computational delay of the task is In a linear relationship, the unloading vehicle will unload the task to the vehicle type β l The function representation of the unloading vehicle utility model of the computable vehicle is: Among them, U l represents the updated unloading vehicle utility model function, η l Indicates that the vehicle type is β l The time price reward of the computational vehicle, α represents the unit benefit factor of delay, f0 represents the computational power of the unloaded vehicle, and f l Indicates that the vehicle type is β l Computable vehicles can share computing resources.

5. The vehicle collaborative task unloading method based on incentive mechanism according to claim 1 is characterized in that: The vehicle private information estimation method based on the exploration and exploitation algorithm comprises: The private information of the vehicle represents the unit cost of energy consumption. At the initial stage, the maximum and minimum values ​​of the unit cost of energy consumption are known to the unloaded vehicle, and it is divided into L equal intervals, each of which corresponds to the initial energy consumption cost ε of a vehicle type l ; Unloading the vehicle will contract items (f l ,η l ) is broadcast to the computable vehicles. After receiving the feedback, the average unit cost of energy consumption of all computable vehicles in vehicle type l at time slot t is calculated. l,t , uncertainty measure A for vehicle type l,t , update the unit cost value of energy consumption ε l ,express: Among them, ψ l,t and A l,t Respectively represent: Among them, π l Indicates that the vehicle type is β l The vehicle probability, b l Indicates vehicle type β l The balance factor, |H(t)| represents the total number of computable vehicle sets at time slot t, Q l,t Indicates acceptance of the contract at time slot t (f l ,η l ), the vehicle type is β l The average energy consumption of the calculable vehicle set, ε l, h represents the unit cost of energy consumption of calculable vehicle h.

6. The vehicle collaborative task unloading method based on incentive mechanism according to claim 1 is characterized in that: The above mentioned constraints of individual rationality, incentive compatibility and monotonicity are satisfied to establish the optimization problem of maximizing the expected utility of unloading vehicles: Subject to .C1: C2： C3:β1<...<β l <...<β L ,1≤l≤L Among them, U represents the unloaded vehicle, C1 represents the IR constraint, C2 represents the IC constraint, C3 represents the constraint that the vehicle types are sorted in ascending order, L represents the size of the vehicle type set, l and k represent the index of the vehicle type set, f0 represents the computing power of the unloaded vehicle, and f l 、f k Indicates that the vehicle type is β l , β k The computing resources that can be shared by the computing vehicles, α represents the unit benefit factor of delay, η l , η k Indicates that the unloading vehicle is given a vehicle type of β l , β k The unit communication association time reward of the vehicle can be calculated.

7. The vehicle collaborative task unloading method based on incentive mechanism according to claim 1 is characterized in that: The optimization problem is updated by exploring the relationship between adjacent vehicle types to reduce the number of IR and IC constraints, including: Subject to .C1: C2： C3:β1<...<β l <...<β L ,1≤l≤L Among them, U represents the unloading vehicle, C1 represents the IR constraint, C2 represents the IC constraint, C3 represents the constraint that the vehicle types are sorted in ascending order, and β l represents the vehicle type, f0 represents the computing power of the unloaded vehicle, and f l Indicates that the vehicle type is β l The computing resources that can be shared by the computing vehicles, α represents the unit benefit factor of delay, η l Indicates that the unloading vehicle is given a vehicle type of β l The unit communication association time reward of the vehicle can be calculated, and L represents the size of the vehicle type set.

Images