Mixing auction-based sidecar cooperative computing resource allocation method

Abstract

The invention discloses a hybrid auction-based sidecar cooperative computing resource allocation method, which comprises the following steps that: a sidecar cooperative computing service provider obtains parameter data in a current sidecar cooperative computing system, including user data, intelligent car data, the number of edge servers and associated data, and then performs bidirectional auction on resources on an intelligent car; the method comprises the following steps: firstly acquiring resources of the intelligent vehicle, then performing one-way auction on the resources on the edge server for the situation that the resources of the intelligent vehicle are not acquired, finally calculating the actual payment price of the user and the reward of the intelligent vehicle according to the two-way auction and one-way auction resource allocation result, and after the user pays the reward to the intelligent vehicle according to the actual payment price, the edge vehicle cooperative calculation service provider pays the reward to the intelligent vehicle. And the user uploads the task to the intelligent vehicle or the edge server for remote execution according to the allocation scheme. According to the invention, through effective resource allocation and pricing, the earnings of intelligent vehicles and edge service providers are increased, and the satisfaction degree of users is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of side vehicle collaboration, and more specifically, relates to a method for allocating computing resources for side vehicle collaboration based on a hybrid auction. Background Art

[0002] Edge computing sets up servers at the edge of the network to provide users with nearby computing and storage services. The advantages of edge computing are its powerful computing capabilities and low service costs, while the disadvantages are: ① Edge servers need to be pre-deployed and cannot be dynamically changed according to changes in user needs; ② Edge servers have no sensing devices and cannot collect sensing data of the surrounding environment.

[0003] Vehicle computing refers to an intelligent vehicle with computing capabilities and sensing devices as a vehicle computing platform to provide users with diverse services. The advantages of vehicle computing are: ① The computing power has the ability of dynamic deployment: the computing power changes with the number of intelligent vehicles; ② It has rich sensing devices and can collect sensing data of the surrounding environment in real time to serve tasks such as environmental detection. The disadvantage of vehicle computing is that the computing cost of intelligent vehicles is relatively high.

[0004] Therefore, side vehicle collaborative computing combines edge computing and vehicle computing to provide users with various types of services. However, the types of resources provided by edge computing and vehicle computing are different, and there are significant differences in service costs. Facing the complex situation of resource allocation in side vehicle collaborative computing, an efficient resource allocation and pricing method needs to be sought. Summary of the Invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for allocating computing resources for side vehicle collaboration based on a hybrid auction, which can increase the revenues of intelligent vehicles and edge service providers and improve user satisfaction through effective resource allocation and pricing.

[0006] To achieve the above object of the invention, the method for allocating computing resources for side vehicle collaboration based on a hybrid auction of the present invention includes the following steps:

[0007] S1: The side vehicle collaborative computing service provider obtains the participant data in the current side vehicle collaborative computing system, including user data, intelligent vehicle data, edge server data, and association data, where:

[0008] The user data includes: the demand vector of each user i for computing resources d ir represents the demand quantity of user i for computing resource r, i ∈ N, N represents the set of users in the side vehicle collaborative computing system, r ∈ R COM R COM represents the set of computing resource types in the side vehicle collaborative computing system; the demand vector of each user i for sensing resources k ir′ represents the demand quantity of user i for sensing resource r′, where r′ ∈ R SEN , R SEN represents the set of sensing resource types in the edge-cloud collaborative computing system; the maximum bid b of each user i i ;

[0009] The intelligent vehicle data includes: the computing resource supply vector of each intelligent vehicle s h sr represents the quantity of computing resource r that intelligent vehicle s can provide, where s ∈ S, and S represents the set of intelligent vehicles in the edge-cloud collaborative computing system; the sensing resource supply vector of each intelligent vehicle s w sr′ represents the quantity of sensing resource r′ that intelligent vehicle s can provide; the unit resource cost a of each intelligent vehicle s s ;

[0010] The edge server data includes: the computing resource supply vector of each edge server m c mr represents the quantity of computing resource r that edge server m can provide, where m ∈ M, and M represents the set of edge servers in the edge-cloud collaborative computing system;

[0011] The association data includes: the edge server coverage identifier α im , if user i and edge server m are within the same network coverage area and user i does not need sensing resources, then α im = 1, otherwise α im = 0; the intelligent vehicle coverage identifier β is , if user i and intelligent vehicle s are within the same network coverage area, then β is = 1, otherwise β is = 0;

[0012] The edge-cloud collaborative computing service provider initializes the resource allocation identifier of the user, and sets the user-intelligent vehicle allocation identifier y of each user is = 0, and the user-edge server allocation identifier x im = 0;

[0013] S2: The edge-cloud collaborative computing service provider conducts a two-sided auction on the resources on the intelligent vehicle. The specific method is as follows:

[0014] S2.1: Initialize the computing resource supply vector of each intelligent vehicle s The sensing resource supply vector

[0015] Initialize the unit bid ρ of each user i i , and the calculation formula is:

[0016]

[0017] S2.2: Sort all users in descending order according to the unit bid ρ i to obtain the user queue List user , and denote the original serial number of the u-th user as i u , where u = 1, 2, …, |N|, and |N| represents the number of users in the user set N;

[0018] S2.3: Sort all intelligent vehicles s in ascending order according to the unit cost a s to obtain the intelligent vehicle queue List vehicle , and denote the original serial number of the v-th intelligent vehicle as s v , where v = 1, 2, …, |S|, and |S| represents the number of intelligent vehicles in the intelligent vehicle set S;

[0019] S2.4: Screen the key users and key intelligent vehicles from the sorted users and intelligent vehicles. Denote the serial number of the key user in the user queue List user as u_key, and the serial number of the key intelligent vehicle in the intelligent vehicle queue List vehicle as v_key. The conditions satisfied by the key user u_key and the key intelligent vehicle v_key are as follows:

[0020] (ρ iu_key ≥ a sv_key ) & (ρ iu_key+1 ≥ a sv_key+1 )

[0021] or

[0022] where, i u_key represents the original serial number of the key user u_key, and s v_key represents the original serial number of the key intelligent vehicle v_key;

[0023] S2.5: Obtain the user queue

[0024] Obtain the set of alternative intelligent vehicles corresponding to each user i

[0025] S2.6: Select the first user i in the current user queue and delete it from the user queue ;

[0026] S2.7: Select the intelligent vehicle s in the intelligent vehicle set S i one by one and delete it from the intelligent vehicle set S iDelete. Determine whether the intelligent vehicle s meets the task requirements of user i, that is, whether it meets and If not, no allocation is made, and then the next intelligent vehicle is selected from the set of intelligent vehicles until the set of intelligent vehicles S i is empty; once an intelligent vehicle meets the requirements, update the computing resource supply on the intelligent vehicle s and the sensing resource supply and let the user-intelligent vehicle allocation identifier y is = 1, and the resource allocation for user i is completed;

[0027] S2.8: Determine whether the user queue is empty. If it is not empty, it means the allocation is not completed yet. Return to step S2.6. Otherwise, the resource allocation on the intelligent vehicle is completed, and enter step S2.9;

[0028] S2.9: Determine the intelligent vehicle resource allocation plan according to the user-intelligent vehicle allocation identifier: that is, y is = 1 means that the task of user i is uploaded to the intelligent vehicle s for execution, and y is = 0 means that the task of user i is not uploaded to the intelligent vehicle s for execution;

[0029] S3: The side vehicle collaborative computing service provider conducts a one-way auction on the resources on the edge server. The specific method is as follows:

[0030] S3.1: Obtain the user set Sort the users i in the user set in descending order according to the unit bid ρ i to obtain the user queue List u ′ ser ;

[0031] S3.2: Initialize the computing resource supply provided by the edge server m Initialize the resource matching degree ω u ′ ser between each user i in the user queue List im , m ∈ M. The calculation formula is as follows:

[0032]

[0033] S3.3: Select the first user i in the user queue List u ′ ser and delete it from the user queue List u ′ ser ;

[0034] S3.4: Obtain the set of edge servers of user i For each edge server in the edge server set θ i sort them in ascending order according to the resource matching degree ω im to obtain the initial edge server queue List of user i sever,i ;

[0035] S3.5: Select the edge server m in the edge server queue List sever,i one by one, and determine whether the edge server m meets the task requirements of user i, that is, whether it meets If not, no allocation is made, and then the next edge server is selected from the edge server queue List sever,i until the edge server queue List sever,i is empty; once an edge server m meets the requirements, update the available resource amount on the edge server m and set the user-edge server allocation flag x im = 1, and the resource allocation of user i is completed;

[0036] S3.6: Determine whether the user queue List u ′ ser is empty. If it is not empty, it means the allocation is not completed yet. Return to step S3.3. Otherwise, the allocation is completed and go to step S3.7;

[0037] S3.7: Determine the resource allocation scheme according to the user-edge server allocation flag: that is, when x im = 1, it means that the task of user i is uploaded to the edge server m for execution. When x im = 0, it means that the task of user i is not uploaded to the edge server m for execution;

[0038] S4: The sidecar collaborative computing service provider calculates the actual payment price of the user and the reward of the intelligent vehicle. The specific method is as follows:

[0039] S4.1: For each user i, initialize its actual payment price pay i = 0;

[0040] S4.2: Initialize the set of users who obtain resources on the intelligent vehicle

[0041] S4.3: The user i in the user set N vehicle obtains resources from the intelligent vehicle, and the actual payment price pay of user i i is calculated using the following formula:

[0042]

[0043] S4.4: For the intelligent vehicle s in the set of intelligent vehicles S, calculate the reward pay of the intelligent vehicle s using the following formula s :

[0044]

[0045] S4.5: Initialize the set of users who obtain resources on the edge server

[0046] S4.6: Select the first user i in the set of users N mec and delete it from the user queue N mec to obtain the set of users

[0047] S4.7: Let the set of users denote the computing resource supply vector of the edge server m

[0048] S4.8: Select each user j in the set of users in turn and delete its user set and allocate edge server resources to this user j using the same method as in steps S3.3 - S3.6. Whenever a user j obtains resources from the edge server m, only update the available resource amount of the edge server m Do not set x jm = 1, and then determine whether the available resource amounts on all edge servers that meet the network coverage can meet the requirements of user i:

[0049] If for any one of the edge servers m where all α im = 1, there is at least one type of computing resource such that then the available computing resources on all edge servers that meet the network coverage do not meet the requirements of user i, and this user j is taken as the new critical user j * , the search for critical users ends, otherwise select the next user in the set of users until the set of users is empty;

[0050] S4.9: Determine whether a critical user is found in step S4.8. If not, let the actual payment price pay of user i i = 0, otherwise calculate the actual payment price pay of user i using the following formula i :

[0051]

[0052] S4.10: Determine the set of users N mecWhether it is empty. If it is not empty, it means that the calculation of the actual payment price has not been completed, and return to step S4.6; otherwise, the calculation is completed.

[0053] S5: The user pays the price to the sidecar collaborative computing service provider according to the actual payment price, and the sidecar collaborative computing service provider pays the remuneration to the intelligent vehicle; the user uploads the task to the intelligent vehicle or the edge server for remote execution according to the allocation scheme.

[0054] The sidecar collaborative computing resource allocation method based on hybrid auction of the present invention. The sidecar collaborative computing service provider obtains the parameters or data in the current sidecar collaborative computing system, including user data, intelligent vehicle data, the number of edge servers and associated data, then conducts a two-way auction on the resources of the intelligent vehicle, and then conducts a one-way auction on the resources of the edge server for those who do not obtain the resources of the intelligent vehicle. Finally, according to the resource allocation results of the two-way auction and the one-way auction, calculate the actual payment price of the user and the remuneration of the intelligent vehicle. After the user pays according to the actual payment price, the sidecar collaborative computing service provider pays the remuneration to the intelligent vehicle, and the user uploads the task to the intelligent vehicle or the edge server for remote execution according to the allocation scheme.

[0055] The present invention has the following technical effects:

[0056] 1) The present invention is suitable for complex resource allocation scenarios based on heterogeneous resources (computing resources and sensing resources) in sidecar collaborative computing;

[0057] 2) The resource allocation and pricing method based on hybrid auction proposed by the present invention effectively solves the problem of heterogeneous costs of intelligent vehicles (the costs of each intelligent vehicle may be different);

[0058] 3) The method proposed by the present invention can effectively improve social welfare and meet the trusted attributes, and has practical application value. Brief Description of the Drawings

[0059] Figure 1 is the schematic diagram of the sidecar collaborative computing resource allocation based on hybrid auction of the present invention;

[0060] Figure 2 is the flowchart of the specific implementation manner of the sidecar collaborative computing resource allocation method based on hybrid auction of the present invention;

[0061] Figure 3 is the flowchart of the two-way auction of the computing resources and sensing resources of the intelligent vehicle of the present invention;

[0062] Figure 4 is the flowchart of the one-way auction of the computing resources of the edge server of the present invention;

[0063] Figure 5 is the flowchart of the calculation of the actual payment price of the user and the remuneration of the intelligent vehicle of the present invention. Detailed implementation manners

[0064] The following describes the detailed implementation manners of the present invention in conjunction with the accompanying drawings, so that those skilled in the art can better understand the present invention. It should be particularly noted that in the following description, when the detailed description of known functions and designs may obscure the main content of the present invention, these descriptions will be omitted here.

[0065] To better illustrate the technical solution of the present invention, the principle of the present invention will be briefly described first. Figure 1 is a schematic diagram of sidecar collaborative computing resource allocation based on hybrid auction of the present invention. Figure 1 Two network access points are set in : Network access point 1 and network access point 2. The dotted line represents the network coverage range of the network access point. User 1, User 2, Edge server 1, green intelligent vehicle, and black intelligent vehicle are within the network coverage range of network access point 1. User 2, User 3, Edge server 2, black vehicle, and red vehicle are within the network coverage range of network access point 2. Only edge servers or intelligent vehicles within the same network coverage range can provide services to users. Therefore, edge server 1 and the green intelligent vehicle can only provide services to User 1 and User 2; edge server 2 and the red intelligent vehicle can only provide services to User 2 and User 3; the black intelligent vehicle is within the coverage of both networks and can provide services to all users simultaneously.

[0066] The process of sidecar collaborative computing resource allocation based on hybrid auction of the present invention includes five steps: First, the sidecar collaborative computing service provider collects information of users, intelligent vehicles, and edge servers; Second, the sidecar collaborative computing service provider conducts a two-way auction on the computing resources and sensing resources of the intelligent vehicles; Third, the sidecar collaborative computing service provider conducts a one-way auction on the computing resources of the edge servers; Fourth, the sidecar collaborative computing service provider calculates the actual payment price of the users and the remuneration of the intelligent vehicles; Fifth, the users who obtain remote resources upload tasks to the intelligent vehicles or edge servers for remote execution and pay the price to the sidecar collaborative computing service provider, and the sidecar collaborative computing service provider pays the remuneration to the intelligent vehicles.

[0067] To meet the heterogeneous allocation of computing resources and sensing resources in sidecar collaborative computing and the heterogeneous cost constraints of intelligent vehicles, the present invention proposes the following linear programming problem for resource allocation in sidecar collaborative computing:

[0068]

[0069] Among them, x im represents the user-edge server allocation identifier, x im =1 indicates that user i obtains resources from edge server m, x im= 0 indicates that user i has not obtained resources from edge server m, where i ∈ N, N represents the set of users in the edge-cloud collaborative computing system, and m ∈ M, M represents the set of edge servers in the edge-cloud collaborative computing system.

[0070] y is represents the user-vehicle allocation identifier, y is = 1 indicates that user i has obtained resources from vehicle s, y is = 0 indicates that user i has not obtained resources from vehicle s, where s ∈ S, S represents the set of vehicles in the edge-cloud collaborative computing system.

[0071] a s represents the unit resource cost of vehicle s, b i represents the maximum bid of each user i, d ir represents the required quantity of computing resource r by user i, R COM represents the set of computing resource types in the edge-cloud collaborative computing system, k ir′ represents the required quantity of sensing resource r' by user i, R SEN represents the set of sensing resource types in the edge-cloud collaborative computing system, c mr represents the quantity of computing resource r that edge server m can provide, h sr represents the quantity of computing resource r that vehicle s can provide. α im represents the edge server coverage identifier. If user i and edge server m are within the same network coverage range and user i does not require sensing resources, then α im = 1, otherwise α im = 0. β is represents the vehicle coverage identifier. If user i and vehicle s are within the same network coverage range, then β is = 1, otherwise β is = 0.

[0072] Objective condition (1) represents maximizing social welfare; constraint condition (1a) indicates that the total computing resources allocated to users on the edge server cannot exceed the total available computing resources; constraint condition (1b) indicates that the total computing resources allocated to users on the vehicle cannot exceed the available computing resources; constraint condition (1c) indicates that the total sensing resources allocated to users on the vehicle cannot exceed the available sensing resources; constraint condition (1d) indicates that each user can obtain resources from at most one edge server or one vehicle; constraint condition (1e) indicates that users can only obtain resources from edge servers within the same network coverage range; constraint condition (1f) indicates that users can only obtain resources from vehicles within the same network coverage range; constraint condition (1g) is a constraint condition on the allocation variables.

[0073] Based on the above analysis, the present invention designs a sidecar collaborative resource allocation method based on hybrid auction. Figure 2 It is the flowchart of the specific implementation manner of the sidecar collaborative computing resource allocation method based on hybrid auction in the present invention. As Figure 2 shown, the specific steps of the sidecar collaborative computing resource allocation method based on hybrid auction in the present invention include:

[0074] S201: Obtain the participant data in the current system:

[0075] In the present invention, first, the sidecar collaborative computing service provider needs to obtain the participant data in the current sidecar collaborative computing system, including user data, intelligent vehicle data, edge server data, and association data. Next, each type of data will be described in detail.

[0076] The user data includes: the demand vector of each user i for computing resources d ir represents the demand quantity of user i for computing resource r, i ∈ N, N represents the set of users in the sidecar collaborative computing system, r ∈ R COM , R COM represents the set of computing resource types in the sidecar collaborative computing system; the demand vector of each user i for sensing resources k ir′ represents the demand quantity of user i for sensing resource r′, r′ ∈ R SEN , R SEN represents the set of sensing resource types in the sidecar collaborative computing system; the maximum bid b i of each user i, that is, the highest price that user i is willing to pay when the task of user i obtains the corresponding resources. Obviously, if user i has no demand for computing resources or sensing resources, then d i = 0 or k i = 0.

[0077] In the present invention, the intelligent vehicle can provide both computing resources and sensing resources. Therefore, the intelligent vehicle data includes: the computing resource supply vector of each intelligent vehicle s h sr represents the quantity of computing resource r that intelligent vehicle s can provide, s ∈ S, S represents the set of intelligent vehicles in the sidecar collaborative computing system; the sensing resource supply vector of each intelligent vehicle s w sr′ represents the quantity of sensing resource r′ that intelligent vehicle s can provide; the unit resource cost a s of each intelligent vehicle s.

[0078] In the present invention, the edge server only provides computing resources. Therefore, the edge server data includes: the computing resource supply vector of each edge server m cmr denotes the amount of computing resources \(r\) that the edge server \(m\) can provide, where \(m\in M\), and \(M\) represents the set of edge servers in the edge-cloud collaborative computing system.

[0079] The associated data includes: the edge server coverage identifier \(\alpha\) im , if the user \(i\) and the edge server \(m\) are within the same network coverage and the user \(i\) does not require sensing resources (\(k\) ir′ = 0, ) then \(\alpha\) im = 1, otherwise \(\alpha\) im = 0; the intelligent vehicle coverage identifier \(\beta\) is , if the user \(i\) and the intelligent vehicle \(s\) are within the same network coverage then \(\beta\) is = 1, otherwise \(\beta\) is = 0.

[0080] In addition, the edge-cloud collaborative computing service provider initializes the resource allocation identifier of the user, and sets the user-intelligent vehicle allocation identifier \(y\) of each user is = 0, and the user-edge server allocation identifier \(x\) im = 0.

[0081] S202: Conduct a two-sided auction for the resources on the intelligent vehicle:

[0082] The edge-cloud collaborative computing service provider first conducts a two-sided auction for the resources on the intelligent vehicle. Figure 3 This is the flowchart of the two-sided auction for the computing resources and sensing resources of the intelligent vehicle in the present invention. As Figure 3 shown, the specific steps of the two-sided auction for the computing resources and sensing resources of the intelligent vehicle in the present invention include:

[0083] S301: Initialize parameters:

[0084] Initialize the computing resource supply vector of each intelligent vehicle \(s\) and the sensing resource supply vector

[0085] Initialize the unit bid \(\rho\) of each user \(i\) i , and the calculation formula is:

[0086]

[0087] S302: Sort users:

[0088] Sort all users according to the unit bid \(\rho\) i from largest to smallest to obtain the user queue List user , and denote the original serial number of the \(u\)-th user as \(i\) u , where \(u = 1, 2, \ldots, |N|\), and \(|N|\) represents the number of users in the user set \(N\).

[0089] S303: Sorting of intelligent vehicles:

[0090] Sort all intelligent vehicles s in ascending order according to the unit cost a s to obtain the intelligent vehicle queue List vehicle , and denote the original serial number of the v-th intelligent vehicle as s v , where v = 1, 2, …, |S|, and |S| represents the number of intelligent vehicles in the intelligent vehicle set S.

[0091] S304: Screening of key users and key intelligent vehicles:

[0092] Screen key users and key intelligent vehicles from the sorted users and intelligent vehicles. Denote the serial number of the key user in the user queue List user as u_key, and the serial number of the key intelligent vehicle in the intelligent vehicle queue List vehicle as v_key. The conditions satisfied by the key user u_key and the key intelligent vehicle v_key are as follows:

[0093]

[0094] or

[0095] where i u_key represents the original serial number of the key user u_key, and s v_key represents the original serial number of the key intelligent vehicle v_key.

[0096] S305: Updating the user set and the intelligent vehicle set:

[0097] Obtain the user queue that is, delete the users whose unit bids are less than the unit bid of the key user i u_key and the key user from the user queue.

[0098] Obtain the set of alternative intelligent vehicles corresponding to each user i that is, take the intelligent vehicles that are within the same network coverage range as user i and whose unit costs are less than the unit cost of the key intelligent vehicle s v_key except the key intelligent vehicle s v_key as the alternative intelligent vehicles of user i.

[0099] Next, only conduct two-way auctions for the users in the user queue and the intelligent vehicles in the set of alternative intelligent vehicles S i . That is to say, the users not in the user queue will not be able to obtain resources from the intelligent vehicles, and the intelligent vehicles not in the set of alternative intelligent vehicles S i will not be able to provide resources to user i. And the key user i u_keyResources cannot be obtained from the intelligent vehicle, and the key intelligent vehicle s v_key cannot provide services to users. The key users s v_key The bid of v_key is used as the unit bid for the actual payment price of other users, and the key intelligent vehicle s

[0100] S306: Select a user:

[0101] Select the current user queue and delete the first user i from the user queue

[0102] S307: Match an intelligent vehicle:

[0103] Select the intelligent vehicle s in the intelligent vehicle set S one by one i and delete it from the intelligent vehicle set S i Determine whether the intelligent vehicle s meets the task requirements of the user i, that is, whether it meets and If not, no allocation is made, and then the next intelligent vehicle is selected from the intelligent vehicle set until the intelligent vehicle set S i is empty. Once an intelligent vehicle meets the requirements, update the computing resource supply and sensing resource supply on the intelligent vehicle s and set the user allocation flag y is = 1, and the resource allocation for the user i is completed.

[0104] S308: Determine whether the user queue is empty. If it is not empty, it means the allocation is not completed yet. Return to step S306. Otherwise, the resource allocation on the intelligent vehicle is completed, and go to step S309.

[0105] S309: Determine the intelligent vehicle resource allocation plan:

[0106] Determine the intelligent vehicle resource allocation plan according to the user-intelligent vehicle allocation flag: that is, y is = 1 indicates that the task of the user i is uploaded to the intelligent vehicle s for execution, and y is = 0 indicates that the task of the user i is not uploaded to the intelligent vehicle s for execution.

[0107] S203: Conduct a one-way auction for resources on the edge server:

[0108] The side vehicle collaborative computing service provider then conducts a one-way auction for the resources on the edge server. Figure 4 This is the flowchart of the one-way auction for the edge server computing resources of the present invention. As Figure 4 ​As shown in the figure, the specific steps of the one-way auction of computing resources for the edge server of the present invention include:

[0109] S401: Obtain the user set:

[0110] Obtain the user set That is, the users in the user set have not obtained resources from the intelligent vehicle and do not need sensing resources. If a user has not obtained resources from the intelligent vehicle but needs sensing resources, then this user cannot obtain resources from the edge server either, because the edge server does not provide sensing resources.

[0111] Sort the users i in the user set in descending order according to the unit bid ρ i to obtain the user queue List u ′ ser .

[0112] S402: Initialize parameters:

[0113] Initialize the computing resource supply provided by the edge server m

[0114] Initialize the resource matching degree ω u ′ ser between each user i in the user queue List im , m ∈ M, indicating the matching degree between the computing resources required by user i and the computing resources provided by edge server m. The resource matching degree ω im is calculated using the following formula:

[0115]

[0116] S403: Select a user:

[0117] Select the first user i in the user queue List u ′ ser and delete it from the user queue List u ′ ser . Next, perform resource allocation for this user on the edge server.

[0118] S404: Initialize the edge server queue:

[0119] Obtain the edge server set of user i i The edge server m in the edge server set θ

[0120] is within the same network coverage as user i, so the edge server m can provide services for user i. The edge server set θi Each edge server in im is sorted from small to large according to the resource matching degree ω to obtain the initial edge server queue List of user i sever,i . The smaller the resource matching degree, the more the computing resources required by user i match the computing resources provided by edge server m

[0121] S405: Match edge servers:

[0122] Select edge server m in the edge server queue List one by one, and determine whether edge server m meets the task requirements of user i, that is, whether it meets sever,i If it does not meet the requirements, no allocation is made, and then the next edge server is selected from the edge server queue List until the edge server queue List sever,i is empty. Once an edge server m meets the requirements, update the available resource amount on edge server m sever,i and set the user-edge server allocation flag x = 1, and the resource allocation of user i is completed im

[0123] S406: Determine whether the user queue List u ′ ser is empty. If it is not empty, it means that the allocation has not been completed, and return to step S403. Otherwise, the allocation is completed, and enter step S407

[0124] S407: Determine the edge server resource allocation plan:

[0125] Determine the resource allocation plan according to the user-edge server allocation flag: that is, when x im = 1, it means that the task of user i is uploaded to edge server m for execution. When x im = 0, it means that the task of user i is not uploaded to edge server m for execution

[0126] S204: Calculate the actual payment price and reward:

[0127] The sidecar collaborative computing service provider calculates the actual payment price of the user and the reward of the intelligent vehicle Figure 5 is the flowchart of the calculation of the actual payment price of the user and the reward of the intelligent vehicle in the present invention. As Figure 5 shown, the specific steps of the calculation of the actual payment price of the user and the reward of the intelligent vehicle in the present invention include:

[0128] S501: Initialize the payment price:

[0129] For each user i, initialize its actual payment price pay i = 0​

[0130] S502: Initialize the user set that obtains resources from the intelligent vehicle:

[0131] Initialize the user set that obtains resources from the intelligent vehicle

[0132] S503: Calculate the actual payment price of the users who obtain resources from the intelligent vehicle:

[0133] User set N vehicle The user i in obtains resources from the intelligent vehicle, and the actual payment price pay of user i i is calculated using the following formula:

[0134]

[0135] As can be seen from the above formula, the actual payment price of user i is based on the key user i u_key 's unit bid price for calculation to ensure that the actual payment price of the user is not less than the cost and trusted attributes of the intelligent vehicle.

[0136] S504: Calculate the reward of the intelligent vehicle:

[0137] For the intelligent vehicle s in the intelligent vehicle set S, the reward pay of the intelligent vehicle s is calculated using the following formula s :

[0138]

[0139] It can be seen that for the intelligent vehicles that do not provide services their reward pay s = 0. For the intelligent vehicle s that provide services, its reward is calculated according to the key intelligent vehicle s v_key 's unit cost for calculation to ensure that the reward obtained by the intelligent vehicle is not less than its cost and trusted attributes.

[0140] S505: Initialize the user queue of users who do not obtain resources from the intelligent vehicle:

[0141] Initialize the user set of users who obtain resources on the edge server

[0142] S506: Select users who do not obtain resources from the intelligent vehicle:

[0143] Select the first user i in the user set N mec and delete it from the user queue N mec to obtain the user set

[0144] S507: Update the user set:

[0145] Let the user set Denote the computing resource provision vector of edge server m

[0146] S508: Search for new critical users:

[0147] Select each user j in the user set in turn and delete it from its user set . Use the same method as in steps S403 - S406 to allocate edge server resources for this user j. Whenever a user j obtains resources from edge server m, only update the available resource amount of edge server m Do not set x jm = 1, and then determine whether the available resource amounts on all edge servers that meet the network coverage can meet the needs of user i:

[0148] If for any edge server m among all edge servers where α im = 1, there is at least one type of computing resource such that then the available computing resources on all edge servers that meet the network coverage do not meet the needs of user i, and this user j is taken as the new critical user j * , and the search for critical users ends. Otherwise, select the next user in the user set until the user set is empty.

[0149] S509: Calculate the actual payment price of the users who obtain resources from the edge server:

[0150] Determine whether a critical user is found in step S508. If not, let the actual payment price pay of user i i = 0. Otherwise, calculate the actual payment price pay of user i using the following formula i :

[0151]

[0152] S510: Determine whether the user set N mec is empty. If not, it means that the calculation of the actual payment price has not been completed, and return to step S506. Otherwise, the calculation is completed.

[0153] S205: Task execution and payment:

[0154] The user pays the price to the sidecar collaborative computing service provider according to the actual payment price, and the sidecar collaborative computing service provider pays the remuneration to the intelligent vehicle. The user uploads the task to the intelligent vehicle or the edge server for remote execution according to the allocation scheme.

[0155] In the resource pricing based on hybrid auctions, the credible attributes of the resource allocation method are crucial. Only when the allocation method satisfies the credible attributes can it be ensured that users and intelligent vehicles cannot submit false bids and costs, thus affecting the efficiency of the entire system. The credible attribute means that the utilities of users and intelligent vehicles are maximized only when they submit true bids and costs. If a user does not obtain resources, the utility is zero; if an intelligent vehicle does not provide services, the utility is zero. User utility = user bid - user actual payment price. Intelligent vehicle utility = reward - cost.

[0156] Theorem: The present invention satisfies the credible attributes in the two-way auction of intelligent vehicle resources

[0157] Proof: Assume that the true bid of user i is b i . Since the actual payment unit price is and therefore, the utility when the user submits the true bid is not less than zero. If the user submits a false bid and obtains resources from the intelligent vehicle, since the actual payment unit price is not determined by the bid of user i, but by the unit bid of the key user key user , the utility remains unchanged. If user i submits a false bid and does not obtain resources from the intelligent vehicle, then the user's utility is zero and not greater than the utility when submitting the true bid.

[0158] Assume that the true unit cost of intelligent vehicle s is a s . Since the unit price of the reward is and therefore, the utility of intelligent vehicle s is not less than zero. If intelligent vehicle s submits a false unit cost but since the unit price of the reward is not determined by intelligent vehicle s, but by intelligent vehicle v_key, the unit price of the reward remains unchanged, and its utility is not greater than the utility when the true cost is submitted.

[0159] Q.E.D.

[0160] Theorem: The present invention satisfies the credible attributes in the one-way auction of edge server resources

[0161] Proof: Assume that user i obtains resources from the edge server and its actual payment price is calculated based on the unit bid of user j. Since the actual payment price is not determined by the bid of user i, but by the unit bids of other users, the calculation of the actual payment price has nothing to do with the bid of user ii, so the actual payment price will not change and the utility will not increase.

[0162] Q.E.D.

[0163] To better illustrate the technical solution of the present invention, specific examples are used to conduct experimental verification on the present invention. In this embodiment, it is assumed that there are 5 users, 2 edge servers, and 3 intelligent vehicles, and they are all within the same network coverage area. There are two types of computing resources and one type of sensing resource. Table 1 is the resource demand list of users in this embodiment.

[0164] User Unit Bid Bid <![CDATA[d i1 > <![CDATA[d i2 > <![CDATA[k i1 > 1 2.50 10 1.00 2.00 1.00 2 2.05 9 2.00 1.00 1.40 3 2.00 8 2.00 2.00 0.00 4 1.75 7 1.00 3.00 0.00 5 1.25 5 2.00 2.00 0.00

[0165] Table 1 and Table 2 are the resource and cost lists provided by edge servers and intelligent vehicles in this embodiment.

[0166]

[0167] Table 2

[0168] First, a two-way auction is conducted on the resources of intelligent vehicles. The unit bids of User 1, User 2, User 3, User 4, and User 5 are 2.50, 2.05, 2.00, 1.75, and 1.25 respectively. After sorting the users in descending order of unit bid, they are User 1, User 2, User 3, User 4, and User 5. The unit costs of Intelligent Vehicle 1, Intelligent Vehicle 2, and Intelligent Vehicle 3 are 1.00, 1.50, and 1.90 respectively. After sorting the intelligent vehicles in ascending order of unit cost, they are Intelligent Vehicle 1, Intelligent Vehicle 2, and Intelligent Vehicle 3.

[0169] Next, the key user and the key intelligent vehicle are calculated. Since the unit bid of User 3, which is 2.00, is not less than the unit cost of Intelligent Vehicle 3, which is 1.90, and Intelligent Vehicle 3 is the last intelligent vehicle, the key user is User 3 and the key intelligent vehicle is Intelligent Vehicle 3. Since only the users with unit bids not less than the key user 3 are considered and the key user 3 is not allocated resources, the subsequent resource allocation only considers User 1 and User 2. Since only the intelligent vehicles with unit costs not less than the key intelligent vehicle 3 are considered and the key intelligent vehicle 3 does not provide services, the subsequent resource allocation only considers Intelligent Vehicle 1 and Intelligent Vehicle 2.

[0170] In resource allocation, users are selected in descending order of their unit bids. User set First, select the user set User 1 in it (the unit bid of User 1, which is 2.50, is greater than the unit bid of User 2, which is 2.05), and update the user set Intelligent vehicle set S1 = {Intelligent Vehicle 1, Intelligent Vehicle 2}. First, select Intelligent Vehicle 1 in the intelligent vehicle set S1. Since the available resource amount of Intelligent Vehicle 1 meets the needs of User 1, User 1 obtains resources from Intelligent Vehicle 1, and the allocation variable is set to y 11 = 1. Update the resource supply amount of Intelligent Vehicle 1: The resource allocation of User 1 is completed.

[0171] Next, select the user set and select user 2 in it to update the user set The intelligent vehicle set S2 = {Intelligent Vehicle 1, Intelligent Vehicle 2}. First, select Intelligent Vehicle 1 in the intelligent vehicle set S2. Since the available resource amount of Intelligent Vehicle 1 cannot meet the resource requirements of user 2, select the next intelligent vehicle, Intelligent Vehicle 2. Since the available resource amount of Intelligent Vehicle 2 can meet the requirements of user 2, user 2 obtains resources from Intelligent Vehicle 2, and let the allocation variable y 22 = 1. Update the resource supply amount of Intelligent Vehicle 2: The resource allocation for user 2 is completed.

[0172] The resource allocation on the intelligent vehicle is completed. Next, perform resource allocation on the edge server. Update the user set Next, only perform resource allocation for the users in the user set Sort the users in the user set in descending order according to the unit bid price to obtain the user queue List u ′ ser = {Intelligent Vehicle 3, Intelligent Vehicle 4, Intelligent Vehicle 5}.

[0173] Calculate the resource matching degrees of user 3, user 4, and user 5 with Edge Server 1 and Edge Server 2.

[0174] Table 3 is the resource matching degree list in this embodiment.

[0175] Resource Matching Degree Edge Server 1 Edge Server 2 User 3 0.00 0.50 User 4 2.00 1.50 User 5 0.00 0.50

[0176] Table 3

[0177] Select the first user, user 3, in the user queue List u ′ ser = {Intelligent Vehicle 3, Intelligent Vehicle 4, Intelligent Vehicle 5} and update the user queue List u ′ ser = {Intelligent Vehicle 4, Intelligent Vehicle 5}. The edge server set M3 = {Edge Server 1, Edge Server 2}. Sort the edge servers in the edge server set M3 in ascending order according to the resource matching degree with user 3 to obtain the edge server queue List sever,3 = {Edge Server 1, Edge Server 2}. First, select Edge Server 1 in the edge server queue List sever,3 . Since Edge Server 1 can meet the resource requirements of user 3, user 3 obtains resources from Edge Server 1, and let the allocation variable x 31 = 1. Update the available resource amount on Edge Server 1: The resource allocation for user 3 is completed.

[0178] Select the user queue List u ′ ser = the first user 4 in {intelligent vehicle 4, intelligent vehicle 5}, and update the user queue List u ′ ser = {intelligent vehicle 5}. The sorted edge server queue List sever,4 = {edge server 2, edge server 1}. First, select the edge server 2 in the edge server queue List sever,4 Since edge server 2 can meet the resource requirements of user 4, user 4 obtains resources from edge server 2, and let the allocation variable x 42 = 1. Update the available resource amount on edge server 2: The resource allocation for user 4 is completed.

[0179] Select the user queue List u ′ ser = the first user 5 in {intelligent vehicle 5}, and update the user queue The sorted edge server queue List sever,5 = {edge server 1, edge server 2}. First, select the edge server 1 in the edge server queue List sever,5 Since edge server 1 cannot meet the resource requirements of user 5, edge server 2 is selected. And since edge server 2 also cannot meet the resource requirements of user 5, user 5 cannot obtain resources. The resource allocation for user 5 is completed.

[0180] The users who obtain resources from the intelligent vehicles are user 1 and user 2. Since the key user is user 3, the actual payment prices of user 1 and user 2 are calculated based on the unit bid of user 3 as: 8.00 and 8.80 respectively. Intelligent vehicle 1 and intelligent vehicle 2 provide services. Since the key intelligent vehicle is intelligent vehicle 3, the rewards of intelligent vehicle 1 and intelligent vehicle 2 are calculated based on the unit cost of intelligent vehicle 3 as: 7.60 and 8.36 respectively. Intelligent vehicle 3 does not provide services, so the reward is 0.

[0181] The users who obtain resources from the edge servers are user 3 and user 4. First, calculate the actual payment price of user 3. If user 3 does not participate in the allocation, user 4 first obtains resources from edge server 2, and then user 5 obtains resources from edge server 1. The remaining resource amounts of edge server 1 and 2 cannot meet the needs of user 3. Therefore, the actual payment price of user 3 should be calculated according to the unit bid of user 5, and the actual payment price of user 3 is: 5. That is, if user 3 bids 5, its unit bid is not less than that of user 5. The calculation of the actual payment price of user 3 is completed.

[0182] Next, calculate the actual payment price of User 4. If User 4 does not participate in the allocation, User 3 first obtains resources from Edge Server 1, and then User 5 obtains resources from Edge Server 2. The remaining resource amounts of Edge Servers 1 and 2 cannot meet the needs of User 4. Therefore, the actual payment price of User 4 should be calculated based on the unit bid of User 5, and the actual payment price of User 4 is: 5. That is to say, if the bid of User 3 is 5, its unit bid is not less than that of User 5. The calculation of the actual payment price of User 4 is completed. Table 4 is the final allocation result table in this embodiment.

[0183]

[0184] Table 4

[0185] Although the above describes the illustrative specific embodiments of the present invention for the convenience of those skilled in the art to understand the present invention, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions and creations using the concept of the present invention are within the scope of protection.

Claims

1. A sidecar collaborative computing resource allocation method based on hybrid auction, characterized in that Including the following steps: S1: The sidecar collaborative computing service provider obtains the participant data in the current sidecar collaborative computing system, including user data, intelligent vehicle data, edge server data, and association data, where: The user data includes: the demand vector of each user i for computing resources d ir represents the demand quantity of user i for computing resource r, i ∈ N, where N represents the set of users in the sidecar collaborative computing system, and r ∈ R COM , R COM represents the set of computing resource types in the sidecar collaborative computing system; the demand vector of each user i for sensing resources k ir′ represents the demand quantity of user i for sensing resource r′, r′ ∈ R SEN , R SEN represents the set of sensing resource types in the sidecar collaborative computing system; the maximum bid b of each user i i ; The intelligent vehicle data includes: the computing resource supply vector of each intelligent vehicle s h sr represents the quantity of computing resources r that the intelligent vehicle s can provide, s ∈ S, where S represents the set of intelligent vehicles in the edge vehicle collaborative computing system; the sensing resource supply vector of each intelligent vehicle s w sr′ represents the quantity of sensing resources r′ that the intelligent vehicle s can provide; the unit resource cost a of each intelligent vehicle s s ; The edge server data includes: the computing resource supply vector of each edge server m c mr represents the quantity of computing resource r that edge server m can provide, where m ∈ M and M represents the set of edge servers in the sidecar collaborative computing system; The associated data includes: the edge server coverage identifier α im , if user i and edge server m are within the same network coverage range and user i does not require sensing resources, then α im = 1, otherwise α im = 0; the intelligent vehicle coverage identifier β is , if user i and intelligent vehicle s are within the same network coverage range, then β is = 1, otherwise β is = 0; The sidecar collaborative computing service provider initializes the resource allocation identifier of the user, and sets the user-intelligent vehicle allocation identifier y of each user is = 0, and the user-edge server allocation identifier x im = 0; S2: The sidecar collaborative computing service provider conducts a two-way auction on the resources of the intelligent vehicle. The specific method is as follows: S2.1: Initialize the computing resource supply vector of each intelligent vehicle s Sensing resource supply vector Initialize the unit bid ρ for each user i i , and the calculation formula is: S2.2: Sort all users according to the unit bid ρ i in descending order to obtain the user queue List user , and denote the original serial number of the u-th user as i u , where u = 1, 2, …, |N|, and |N| represents the number of users in the user set N; S2.3: Sort all the intelligent vehicles s in ascending order according to the unit cost a s to obtain the intelligent vehicle queue List vehicle , and denote the original serial number of the v-th intelligent vehicle as s v , where v = 1, 2, …, |S|, and |S| represents the number of intelligent vehicles in the intelligent vehicle set S; S2.4: Screen key users and key intelligent vehicles from the sorted users and intelligent vehicles, and record the serial number of the key user in the user queue List user as u_key, and record the serial number of the key intelligent vehicle in the intelligent vehicle queue List vehicle as v_key. The conditions satisfied by the key user u_key and the key intelligent vehicle v_key are as follows: or where i u_key represents the original serial number of the key user u_key, and s v_key represents the original serial number of the key intelligent vehicle v_key; S2.5: Obtain the user queue Obtain the set of alternative intelligent vehicles corresponding to each user i S2.6: Select the first user i in the current user queue and delete it from the user queue ; S2.7: Select the intelligent vehicles in the set S of intelligent vehicles one by one i and delete it from the set S of intelligent vehicles i . Determine whether the intelligent vehicle s meets the task requirements of user i, that is, whether it meets and . If not, no allocation is made, and then the next intelligent vehicle is selected from the set of intelligent vehicles until the set S of intelligent vehicles i is empty; once an intelligent vehicle meets the requirements, update the computing resource supply and sensing resource supply on the intelligent vehicle s and set the user-intelligent vehicle allocation flag y is = 1, and the resource allocation for user i is completed; S2.8: Determine the user queue Check if it is empty. If it is not empty, it means that the allocation is not yet complete, and return to step S2.

6. Otherwise, the resource allocation on the intelligent vehicle is completed, and proceed to step S2.9; S2.9: Determine the allocation scheme of intelligent vehicle resources according to the user-intelligent vehicle allocation identifier, i.e., y is = 1 indicates that the task of user i is uploaded to intelligent vehicle s for execution, y is = 0 indicates that the task of user i is not uploaded to intelligent vehicle s for execution; S3: The sidecar collaborative computing service provider conducts a one-way auction on the resources of the edge server. The specific method is as follows: S3.1: Obtain the user set For the user set where user i, sort the users in descending order according to the unit bid ρ i to obtain the user queue List u ′ ser ; S3.2: Initialize the computing resource supply provided by edge server m Initialize the user queue List u ′ ser The resource matching degree ω between each user i and edge server m in im , The calculation formula is as follows: S3.3: Select user queue List u ′ ser Select the first user i in it, and delete it from the user queue List u ′ ser ; S3.4: Obtain the edge server set of user i For the edge servers in the edge server set θ i sort them in ascending order according to the resource matching degree ω im to obtain the initial edge server queue List of user i sever,i ; S3.5: Select edge server queues List one by one sever,i The edge server m in it is judged to see if it meets the task requirements of user i, that is, whether it meets If not, no allocation is made, and then the next edge server is selected from the edge server queue List sever,i until the edge server queue List sever,i is empty; once an edge server m meets the requirements, update the available resource amount on the edge server m and set the user-edge server allocation flag x im = 1, and the resource allocation for user i is completed; S3.6: Determine whether the user queue List u ′ ser is empty. If it is not empty, it means that the allocation has not been completed, and return to step S3.

3. Otherwise, the allocation is completed, and proceed to step S3.7; S3.7: Determine the resource allocation plan according to the user-edge server allocation identifier: that is, when x im = 1, it means that the task of user i is uploaded to edge server m for execution, and when x im = 0, it means that the task of user i is not uploaded to edge server m for execution; S4: The sidecar collaborative computing service provider calculates the actual payment price of the user and the remuneration of the intelligent vehicle. The specific method is as follows: S4.1: For each user i, initialize their actual payment price pay i = 0; S4.2: Initialize the user set for obtaining resources on the intelligent vehicle S4.3: User set N vehicle User i in it obtains resources from the intelligent vehicle, and the actual payment price pay of user i i is calculated using the following formula: S4.4: For the intelligent vehicle s in the set S of intelligent vehicles, use the following formula to calculate the reward pay of the intelligent vehicle s s : S4.5: Initialize the user set that obtains resources on the edge server S4.6: Select the first user i from the user set N mec and delete it from the user queue N mec to obtain the user set S4.7: Let the user set denote the computing resource providing vector of the edge server m S4.8: Select each user j in the user set in turn and delete it from its user set Then, use the same method as in steps S3.3 - S3.6 to allocate edge server resources for this user j. Whenever a user j obtains resources from edge server m, only update the available resource amount of edge server m Do not set x jm = 1, and then determine whether the available resource amounts on all edge servers that meet the network coverage can meet the requirements of user i: If for all α im for any one of the edge servers m where im = 1, there is at least one type of computing resource such that then the available computing resources on all edge servers that meet the network coverage do not meet the requirements of user i, and this user j is taken as the new critical user j * , the search for critical users ends. Otherwise, select the next user in the user set , until the user set is empty; S4.9: Determine whether the key user is found in step S4.

8. If not, set the actual payment price pay of user i to 0; otherwise, calculate the actual payment price pay of user i using the following formula i : i : S4.10: Determine whether the user set N mec is empty. If it is not empty, it indicates that the calculation of the actual payment price has not been completed, and step S4.6 is returned; otherwise, the calculation is completed. S5: The user pays the price to the sidecar collaborative computing service provider according to the actual payment price, and the sidecar collaborative computing service provider pays the remuneration to the intelligent vehicle; the user uploads the task to the intelligent vehicle or the edge server for remote execution according to the allocation scheme.

Images