An end-to-end coordination method

Abstract

The application relates to an end-edge coordination method, belonging to the field of wireless communication, and comprising the following steps: S1: according to data exchanged by terminal equipment, a system model combining a physical layer and a social layer is established, wherein the physical layer represents physical constraints and communication requirements among equipment, and the social layer represents social links among the equipment; S2: based on a task scene, an end-edge coordination computing system taking minimum time delay as an objective is established; and S3: a VCG auction mechanism is used to solve the selfishness of resource sharing among the equipment, and an auction winner combination and pricing are determined. Under the precondition of optimizing time delay, the application improves the effective utilization rate of resources, and guarantees individual rationality and incentive compatibility of the mechanism.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of wireless communication, and relates to an end-edge collaborative method. BACKGROUND

[0002] Under the influence of global industrial reform, countries implement the development of industry, and the development trend of modern factories is gradually advancing towards high automation, which marks the successful transformation and upgrading of traditional industry, and creates a new situation of industrial equipment interconnection. With the continuous emergence of a large number of mobile devices such as industrial monitors, industrial robots and industrial sensors, the huge traffic generated by these devices brings great pressure to wireless access networks. As a promising solution, edge computing sinks cloud computing to the network edge to pursue less computing latency while maintaining low transmission latency. However, the communication capability and computing resources of edge servers are limited, and it is difficult to meet the current complex task computing demand by using only edge computing mode, and the close cooperation of devices and edges is a promising solution to expand communication coverage and solve network congestion problems.

[0003] When these industrial devices are interconnected through D2D technology, resource-constrained devices can directly offload their computationally intensive tasks to nearby resource-rich devices to achieve high transmission rate, low latency and high resource utilization, and other high-quality experiences. Although resource sharing has a bright application prospect in industrial internet, it also faces challenging challenges such as profit-driven incentives. Specifically, SRs and SPs belong to different interest groups, and considering the energy and resource constraints of devices, SPs may not be willing to provide resources to SRs in the absence of profit. Another factor to consider in resource sharing is social awareness in the process of industrial device collaboration. A large number of devices have a considerable part carried by humans or installed on manned machines, and in the process of communicating with each other through D2D links, they inevitably exhibit social relationships (such as friendship, kinship and colleague relationships) and regularity. Even without human involvement, devices tend to share information with trusted devices, and according to the widely used access control framework, establishing new access requires additional costs.

[0004] Therefore, providing services through end-edge collaboration in complex industrial internet scenarios can exert the efficiency of terminal devices and improve resource utilization. However, how to design a reasonable incentive mechanism to encourage resource idle devices to share their resources while optimizing task latency is worth studying. SUMMARY

[0005] In view of this, the purpose of this invention is to address the problems of limited computing power of edge servers and low resource utilization due to device selfishness in resource sharing in complex industrial Internet scenarios. This invention provides an end-edge collaboration method that considers the collaborative computing tasks between terminal devices and edge servers. Through VCG auctions, collaborative devices are incentivized to share their resources to improve resource utilization.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] An edge-end collaboration method includes the following steps:

[0008] S1: Based on the data exchanged between terminal devices, establish a system model that combines the physical layer and the social layer, where the physical layer represents the physical constraints and communication requirements between devices, and the social layer represents the social links between devices.

[0009] S2: Based on the task scenario, establish an edge-end collaborative computing system with the goal of minimizing latency;

[0010] S3: Based on the VCG auction mechanism, solve the selfishness problem of resource sharing among devices, and determine the auction winner combination and pricing.

[0011] Furthermore, step S1 specifically includes the following steps:

[0012] S11: In the system model combining the physical and social layers, the social relationship topology is modeled as a weighted undirected graph G(V,E,W), where... It is a set of industrial equipment, E = {e x,y |x,y∈V} represents the set of social edges, and E represents whether there is a social relationship between devices. If there is a social relationship between two devices, they are connected by an edge. Each edge is assigned a scalar value in the range (0,1), which is the strength or weight of the social connection.

[0013] S12: The improved cosine similarity in the task-content-based collaborative filtering algorithm is selected as the indicator to measure the strength of social relationships; the numerical differences between different dimension vectors are corrected by subtracting the mean user ratings for the task; the social relationship between device i and device j is:

[0014]

[0015] Where A represents the set of tasks jointly scored by device i and device j; R i,a and R j,a Let i and j represent the scores given by device i and device j for task a, respectively. and Let i and j represent the average scores of devices i and j for the task, respectively.

[0016] Furthermore, step S2 specifically includes the following steps:

[0017] S21: Collaborating devices (Service Providers, SPs) process tasks from resource-poor requesting devices (Service Requesters, SRs) and offload some of the tasks to edge services;

[0018] S22: Taking the total latency of all devices as the optimization objective, the optimization problem is decomposed into two sub-problems. The first sub-problem is to minimize the transmission latency between the requesting device and the collaborating device, and the other is to minimize the computation latency between the collaborating device and the edge server.

[0019] S23: Use Cauchy's inequality to find the optimal time slot allocation and design a closed-form solution for the optimal communication resource allocation strategy;

[0020] S24: By analyzing the monotonicity between latency and task offloading ratio, the optimal offloading ratio λ is determined. * ;

[0021] S25: Based on the optimal task unloading ratio λ * The problem of minimizing computational delay is transformed into a convex problem, and a closed-form solution for optimal allocation of computational resources is obtained.

[0022] Furthermore, step S21 includes the following four types of delays: transmission delay for each requesting device. The computational latency of each collaborative device is Transmission latency between collaborative device SP and edge server edge server computation latency Collaborative device SP j Processing equipment SR i The total latency of the task is expressed as:

[0023]

[0024] Furthermore, in step S21, SR i With SP j The transmission delay between them is expressed as: In collaborative device SP j The computational delay on the above is expressed as: The transmission latency between the collaborative device and the edge server is expressed as Edge computing latency is Among them, L j,i Indicates the size of the input data used to process the task, C j,i This represents the number of CPU cycles required to compute 1 bit of data for this task, where T represents the length of a TDMA frame, and R represents the number of CPU cycles required. j,iThis is the expected channel capacity, assuming fixed TDMA scheduling, i.e., the time slot resource τ for each device. j,i It is fixed in each time frame, λ j,i The task split ratio, which is expressed in SP j The proportion of data processed in the edge computing task, and the number of CPU cycles required to successfully process the edge computing portion of the data, are denoted as λ. j,i L j,i C j,i The computing resources of collaborative devices are defined as follows: Edge servers are assigned to SPs j Computing resources are W j Indicates the transmission rate.

[0025] Furthermore, the optimization problem in step S22 is:

[0026]

[0027] stC1:

[0028] C2:

[0029] C3:

[0030] C4:0≤λ j,i ≤1,

[0031] Average transmission delay of device SR Based solely on the time slot length τ j,i It is determined that constraint C1 is independent of other optimization variables and is not coupled with other constraints, therefore... The problem is broken down into two sub-problems. The first is to minimize the transmission latency between the device and the cooperating device, expressed as:

[0032]

[0033] stC1

[0034] Another is to minimize the computation latency of devices (SPs) and edge servers:

[0035]

[0036] Furthermore, step S23 specifically includes: communication resource allocation strategy and transmission delay. Related, through solving The problem yields the optimal communication resource allocation strategy;

[0037] Assumption Where δ∈(0,T), according to the formula Transmission delay along with As δ increases and decreases, when allocated to any device, the corresponding transmission delay... The latency of other devices remains unchanged, while the latency of transmission in other devices decreases, resulting in Decrease, result and The optimal solution requires a contradiction, therefore the optimal time slot allocation... satisfy According to Cauchy's inequality:

[0038]

[0039] The condition for the equation to hold is: Substitute the conditions into From The final calculation yields the optimal communication resource allocation result:

[0040]

[0041] Furthermore, step S24 specifically includes: middle With λ j,i The increase is monotonically increasing, while and With λ j,i Increases while monotonically decreasing, when hour, The optimal task unloading ratio corresponding to reaching the minimum value is:

[0042]

[0043] Furthermore, step S25 specifically includes: placing Substitution In have to:

[0044]

[0045] therefore Problem transformed into

[0046]

[0047] stC2,C3

[0048] By solving the KKT conditions, the optimal computational resource allocation strategy is derived:

[0049]

[0050] Where, (x) + =max{x,0}, and θ * It is the optimal value of the Lagrange multiplier when the equality of resource constraints C2 and C3 is satisfied.

[0051] Furthermore, step S3 specifically includes the following steps:

[0052] S31: Model the interaction between devices. Resources of SPs are traded as commodities. SPs and SRs are considered sellers and buyers, respectively. Each bidder in the VCG auction submits the required resource quantity and the corresponding price they are willing to pay. Bids submitted by bidders are sealed and cannot be disclosed before the auction ends. If device SR... i and equipment SP j If a direct D2D link exists, the bid is represented by the following formula:

[0053]

[0054] In the VCG mechanism, a bidder's dominant strategy is to ensure that their bid price reflects the true valuation, i.e., the bidder's true valuation v. j,i It refers to the bid price, which measures v. j,i It includes two factors: task processing latency T j,i social relations w j,i ;

[0055] During the auction phase, the goal is to maximize the efficiency of equipment resource allocation, i.e., to maximize social welfare. The optimization function is expressed as:

[0056]

[0057] S32: The winner selection process is equivalent to the knapsack problem, where the seller SP's resource quantity represents the maximum weight the knapsack can bear, the buyer SR's resource requirement represents the weight of the item, and the buyer's bid price is the corresponding value of the item; under the constraint of limited resources, the optimal combination of winners in each physical community is calculated using the knapsack algorithm to maximize social welfare; during the process, if the buyer SR unloads its task from a seller, then SR can no longer participate in the auction.

[0058] S33: In the payment price determination phase, according to the VCG pricing mechanism, the price paid by the winner represents the marginal harm to other buyers, expressed as:

[0059]

[0060] The first summation term is the device SR. i The greatest social welfare for other bidders in the absence of the other bidders Indicates device SRi The optimal allocation solution in the absence of the device SR, the second summation term represents the optimal allocation solution when the device SR is not present. i The social welfare values ​​of other bidders in the presence of the parties, among which Indicates consideration of SR i The optimal allocation solution under the given conditions.

[0061] The beneficial effects of this invention are: it improves the effective utilization rate of resources while optimizing latency, and at the same time ensures the individual rationality and incentive compatibility of the mechanism.

[0062] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0063] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0064] Fig. 1 This is a system architecture diagram of the edge-end collaboration method described in this invention;

[0065] Fig. 2 A diagram illustrating the price paid to the winner of this invention;

[0066] Fig. 3 This is a flowchart of the edge-end collaboration method described in this invention. Detailed Implementation

[0067] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0068] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0069] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0070] Please see Figs. 1-3 The edge-end collaboration method described in this invention specifically includes the following steps:

[0071] Step 1: Based on the data exchanged between terminal devices, establish a system model combining the physical and social layers. The physical layer represents the physical constraints and communication requirements between devices, while the social layer represents the social connections between devices. In this system model combining the physical and social layers, the social relationship topology is modeled as a weighted undirected graph G(V,E,W), where... It is a set of industrial equipment, E = {e x,y |x,y∈V} represents the set of social edges. E indicates whether there are social relationships between devices; if two devices have a social relationship, they are connected by an edge. To quantitatively analyze the closeness of social relationships, each edge is assigned a scalar value in the range (0,1), which is the strength or weight of the social connection.

[0072] We choose an improved cosine similarity algorithm based on task content as the metric for measuring the strength of social relationships. Since cosine similarity is difficult to measure the numerical differences between vectors of different dimensions, this bias can be corrected by subtracting the mean user ratings of the tasks. Let A represent the set of tasks jointly rated by device i and device j; R... i,a and R j,a Let i and j represent the scores given by device i and device j for task a, respectively. and Let represent the average scores given by devices i and j for the task, respectively. The social relationship between devices i and j is as follows:

[0073]

[0074] Step 2: Based on the task scenario, establish an edge-end collaborative computing system with the goal of minimizing latency, which specifically includes the following steps:

[0075] Step 2.1: Collaborating devices (SPs) process tasks from resource-constrained requesting devices (SRs), partially offloading the tasks to edge services. The entire process considers four types of latency: transmission latency for each requesting device. The computational latency of each collaborative device is Transmission latency between collaborative device SP and edge server edge server computation latency To determine the overall latency of the SR device, the following reasonable assumptions are made: To better implement the task splitting strategy and ensure the accuracy of the splitting results, each cooperating SP device cannot split a task before receiving it. Furthermore, by limiting the task splitting and transmission time and the order in which tasks are processed, the reliability and accuracy of the tasks can be guaranteed. Based on these assumptions, the cooperating SP device... j Processing equipment SR i The total latency of the task can be expressed as:

[0076]

[0077] SR i With SP j The transmission delay between them is expressed as: In collaborative device SP j The computational delay on the above is expressed as: The transmission latency between the collaborative device and the edge server is expressed as Edge computing latency is

[0078] Among them, L j,i Indicates the size of the input data used to process the task, C j,i This represents the number of CPU cycles required to compute 1 bit of data for this task, where T represents the length of a TDMA frame, and R represents the number of CPU cycles required. j,i This is the expected channel capacity, assuming fixed TDMA scheduling, i.e., the time slot resource τ for each device. j,i It is fixed in each time frame, λ j,i The task split ratio, which is expressed in SP j The proportion of data processed in edge computing tasks, and the number of CPU cycles required to successfully process the edge computing portion of the data, can be expressed as λ. j,i L j, i C j,i The computing resources of collaborative devices are defined as follows: Edge servers are assigned to SPs j Computing resources are W j This indicates the transmission rate. Therefore, the cooperating device SP j Processing equipment SR iThe total latency of the task can be expressed as:

[0079]

[0080] Step 2.2: Using the total latency of all devices as the optimization objective, the optimization problem is expressed as:

[0081]

[0082] stC1:

[0083] C2:

[0084] C3:

[0085] C4:0≤λ j,i ≤1,

[0086] Average transmission delay of device SR Based solely on the time slot length τ j,i It is determined that constraint C1 is independent of other optimization variables and is not coupled with other constraints; therefore, The problem can be broken down into two sub-problems. The first is to minimize the transmission latency between the device and the cooperating device, expressed as:

[0087]

[0088] stC1

[0089] Another is to minimize the computation latency of devices (SPs) and edge servers:

[0090]

[0091] stC2,C3,C4

[0092] Step 2.3: Use Cauchy's inequality to find the optimal time slot allocation and design a closed-form solution for the optimal communication resource allocation strategy. Communication resource allocation strategy and transmission delay. Related, therefore, by solving The problem is to find the optimal communication resource allocation strategy. Assume... Where δ∈(0,T), according to the formula Transmission delay along with As δ increases and decreases, when allocated to any device, the corresponding transmission delay... The latency of other devices remains unchanged, while the latency of transmission in other devices decreases, resulting in Decrease, result and The optimal solution requires a contradiction, therefore the optimal time slot allocation... satisfy According to Cauchy's inequality:

[0093]

[0094] The condition for the equation to hold is: Substitute the conditions into From The final calculation yields the optimal communication resource allocation result:

[0095]

[0096] Step 2.4: By analyzing the monotonicity between the calculation latency and the task offloading ratio, the optimal offloading ratio λ is determined. * . middle With λ j,i The increase is monotonically increasing, while and With λ j,i It increases and monotonically decreases, therefore, when hour, The minimum value can be achieved, and the corresponding optimal task unloading ratio is:

[0097]

[0098] Step 2.5: Based on the optimal task unloading ratio λ * The problem of minimizing computational delay is transformed into a convex problem, and a closed-form solution for optimal computational resource allocation is then obtained. Substitution In have to:

[0099]

[0100] therefore Problem transformed into

[0101]

[0102] stC2,C3

[0103] To verify Is the problem convex? Calculate Hessian matrix:

[0104]

[0105]

[0106]

[0107] According to linear algebra theory, if all principal minors of a matrix are greater than 0, then it can be determined that... The Hessian matrix is ​​a positive definite matrix. It is a convex function. It is a series The sum of, therefore This is a convex optimization problem; by solving the KKT conditions, the optimal computational resource allocation strategy is derived:

[0108]

[0109] Where, (x) + =max{x,0}, and θ * It is the optimal value of the Lagrange multiplier when the equality of resource constraints C2 and C3 is satisfied.

[0110] Step 3: Address the selfishness issue in resource sharing among devices based on the VCG auction mechanism, and determine the winning combination and pricing. This includes the following steps:

[0111] Step 3.1: Model the interaction between devices. Resources of SPs are traded as commodities. SPs and SRs are considered sellers and buyers, respectively. Each bidder in the VCG auction submits the required resource quantity and the corresponding price they are willing to pay. Bids submitted by bidders are sealed and cannot be disclosed before the auction ends. If device SR... i and equipment SP j If a direct D2D link exists between two parties, the bid can be represented by the following formula:

[0112]

[0113] In the VCG mechanism, a bidder's dominant strategy is to ensure that their bid price reflects the true valuation, i.e., the bidder's true valuation v. j,i It refers to the bid price, which measures v. j,i It includes two factors: task processing latency T j,i social relations w j,i .

[0114] During the auction phase, the goal is to maximize the efficiency of equipment resource allocation, i.e., to maximize social welfare. The optimization function is expressed as:

[0115]

[0116] Step 3.2: The winner selection process can be equivalent to the knapsack problem, where the seller SP's resource quantity represents the maximum weight the knapsack can bear, the buyer SR's resource requirement represents the weight of the item, and the buyer's bid price is the corresponding value of the item. Therefore, the problem can be solved using a knapsack-based algorithm. Under the constraint of limited resources, the knapsack algorithm can calculate the optimal combination of winners for each physical community to maximize social welfare. During the process, once a bidder unloads their task from any seller, they can no longer participate in the auction. Under the constraint of limited resources, the knapsack algorithm can calculate the optimal combination of winners for each physical community to maximize social welfare.

[0117] Step 3.3: In the payment price determination stage, according to the VCG pricing mechanism, the price paid by the winner represents the marginal harm to other buyers, expressed as:

[0118]

[0119] The first summation term is the device SR. i The greatest social welfare for other bidders in the absence of the other bidders Indicates device SR i The optimal allocation solution in the absence of the device SR, the second summation term represents the optimal allocation solution when the device SR is not present. i The social welfare values ​​of other bidders in the presence of the parties, among which Indicates consideration of SR i The optimal allocation solution under the given conditions.

[0120] If SR i Absentee auction, exists Established, therefore If the equality sign is true, then otherwise, Therefore, the bidder's personal utility is greater than or equal to 0, and the mechanism is consistent with individual rationality (IR).

[0121] The intuitive reason why the incentive compatibility property of VCG auctions holds is that: if the true bid wins, it is obviously superior to the high bid; if the true bid cannot win and the high bid wins, this will inevitably result in a negative ex-post payoff. If the true bid cannot win, then the low bid cannot win either; if the true bid can win, then the low bid will reduce the probability of winning. Therefore, the true bid is the dominant strategy for bidders.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An end-to-end coordination method, characterized by: The method comprises the following steps: S1: according to the data of terminal device interaction, a system model combining physical layer and social layer is established, wherein the physical layer represents the physical constraints and communication requirements between devices, and the social layer represents the social links between devices; S2: based on the task scene, an end-edge collaborative computing system with the minimum time delay as the target is established; the step S2 specifically comprises the following steps: S21: the collaborative device SP processes the task of the request device SR with scarce resources, and offloads part of the task to the edge service; S22: taking the total time delay of all devices as the optimization target, the optimization problem is decomposed into two sub-problems, the first sub-problem is to minimize the transmission time delay between the request device and the collaborative device, and the other is to minimize the computing time delay between the collaborative device and the edge server; S23: the optimal time slot allocation is obtained by using Cauchy inequality, and a closed solution of the optimal communication resource allocation strategy is designed; S24: Obtain the optimal offloading ratio by analyzing the monotonicity between the time delay and the task offloading ratio ; S25: According to the optimal task offloading ratio The minimum calculation delay sub-problem is converted into a convex problem, and a closed solution of the optimal calculation resource allocation is obtained. S3: based on the VCG auction mechanism, the selfishness problem of resource sharing between devices is solved, and the winning combination and pricing of the auction are determined.

2. The end-to-end coordination method of claim 1, wherein: The step S1 specifically comprises the following steps: S11: In the system model combining the physical layer and the social layer, the social relationship topology is modeled as a weighted undirected graph wherein is a set of industrial devices, represents a set of social edges, indicates whether there is a social relationship between devices, if there is a social relationship between two devices, they are connected with an edge, and each edge specifies a scalar value in the range (0, 1), i.e. the strength or weight of the social connection; S12: Select the improved cosine similarity in the collaborative filtering algorithm based on the content of the task as the index to measure the strength of social relationship; modify the numerical difference of different dimension vectors by subtracting the average rating of the user on the task; the social relationship between the device and the device is: wherein denotes the passage of time through the device and the device a common score for the set of tasks; and denotes the score of the device and the device for a task , and denotes the average score of the device and for a task.

3. The end-to-end coordination method of claim 1, wherein: The step S21 includes four time delays: a transmission time delay of each request device ; a calculation time delay of each cooperative device ; a transmission time delay of the cooperative device SP and the edge server ; a calculation time delay of the edge server ; a total time delay of the task of the cooperative device processing device is expressed as: 。 4. The end-to-end coordination method of claim 3, wherein: The transmission delay between the steps S21, The transmission delay between the steps S21, is expressed as: The computation delay on the cooperative device is expressed as: The transmission delay between the cooperative device and the edge server is expressed as The edge computation delay is expressed as Where, represents the input data size for processing the task, represents the number of CPU cycles required to compute 1 bit of data for the task, represents the length of a TDMA frame, is the expected channel capacity assuming a fixed TDMA schedule, i.e. a time slot resource for each device is fixed at each time frame, is the task split ratio, which represents the proportion of the task data processed on The number of CPU cycles required to successfully process the edge-computed part of the data is expressed as The computation resource of the cooperative device is defined as The computation resource allocated by the edge server to is , represents the transmission rate.

5. The end-to-end coordination method of claim 1, wherein: The optimization problem in the step S22 is: Average transmission latency of device SR Only by the length of the time slot Determination, regardless of other optimization variables, constraints Not coupled with other constraints, so that The problem is decomposed into two sub-problems, the first is to minimize the transmission latency between devices and cooperative devices, represented as: The other is to minimize the computing time delay of the device SP and the edge server: 。 6. The end-to-end coordination method of claim 1, wherein: The step S23 specifically comprises: the communication resource allocation strategy is associated with the transmission time delay The problem is solved by solving An optimal communication resource allocation strategy is obtained. Assume where , according to the formula , the transmission delay decreases with the increase of , when assigned to any device, the corresponding transmission delay decreases, while the transmission delay of other devices remains unchanged, resulting in decreases, resulting in contradiction with optimal solution requirements, so the optimal time slot allocation satisfies , according to the Cauchy inequality: Wherein, the equation is established condition , the condition is substituted into , and the following is obtained , and finally the optimal communication resource allocation result is calculated 。 7. The end-to-end synergy method of claim 1, wherein: The step S24 specifically includes: In , With the increase of the number of tasks, the first function monotonically increases, while the second function and the third function monotonically decrease with the increase of the number of tasks, and when , the first function reaches a minimum value, and the corresponding optimal task offloading ratio is: 。 8. The end-to-end synergy method of claim 1, wherein: The step S25 specifically includes: obtaining Substituting into , we get: Thus The problem translates into : Through KKT condition solving, the optimal computing resource allocation strategy is derived: where , and are the optimal values of the Lagrange multipliers when the equalities hold. and are the optimal values of the Lagrange multipliers when the equalities hold.

9. The end-to-end synergy method according to any of claims 1-8, characterized by: The step S3 specifically comprises the following steps: S31: Model the interaction between devices, the resources of SPs are traded as commodities, SPs and SRs are regarded as sellers and buyers respectively, each bidder in VCG auction submits the amount of required resources and the corresponding bid price they are willing to pay; the bid price submitted by the bidder is sealed and cannot be disclosed before the end of the auction; if there is a direct D2D link between the device and the device , the bid is represented by the following formula: In the VCG mechanism, the dominant strategy for a bidder is to bid his true valuation, i.e., the true valuation of the bidder is the bid price, which measures contains two factors: the task processing delay and the social relationship ; In the auction stage, the allocation efficiency of device resources is maximized, i.e. the social welfare is maximized, and the optimization function is represented as: S32: the winner selection process is equivalent to the knapsack problem, wherein the resource amount of the seller SP represents the maximum weight that the knapsack can bear, the resource demand of the buyer SR represents the weight of the goods, and the bid price of the buyer is the corresponding value of the goods; under the constraint of limited resources, the optimal combination of winners of each physical community is calculated by using the knapsack algorithm to realize the maximization of social welfare; in the process, the buyer SR unloads his own task from a seller, so the SR cannot participate in the auction again; S33: in the payment price determination stage, according to the VCG pricing mechanism, the price paid by the winner is the marginal harm to other buyers, which is represented as: where the first summation term is the device the maximum social welfare of the other bidders in the absence of the device denotes the device the optimal allocation solution in the absence of the device, the second summation term denotes the social welfare value of the other bidders when the device is present, where denotes the consideration of the optimal allocation solution in the presence of the device.

Images