A task allocation and scheduling method based on multi-round auction

Abstract

The application relates to the field of communication technology, in particular to a task allocation and scheduling method based on multi-round auction, aiming at the scene that a terminal device simultaneously and in parallel processes multiple tasks, constructing an auction framework based on a D2D cooperative MEC architecture, regarding a busy terminal as a buyer, regarding an idle terminal as a seller, regarding a trusted third party as an auctioneer, jointly improving system utility through edge cooperative computing unloading and task scheduling based on an incentive mechanism, specifically adopting a resource provider selection strategy based on a bid performance ratio and a task scheduling sequence strategy based on a greedy profit-emergency ratio, realizing matching of optimizers and optimal task processing sequences. Simulation results show that the application increases the number of effective task processing and the task processing time delay, fully utilizes the advantages of edge cooperative task unloading and flexible adjustment of task processing sequences, and improves the resource utilization rate.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of communication technology, in particular to a task allocation and scheduling method based on multi-round auction. BACKGROUND

[0002] With the explosive growth of intelligent terminal devices and new applications, the traditional cloud computing architecture cannot meet the delay requirements of these new applications due to long propagation delay. In order to solve these problems, multi-access edge computing (MEC) has emerged. MEC deploys limited computing resources at the edge of the wireless access network and helps terminal devices to process tasks through offloading technology, aiming to achieve low-latency task requests. The latest trend of multi-access edge computing (MEC) is collaborative offloading, that is, the computing tasks of busy terminals (such as smartphones, computers, etc.) can not only be executed by MEC servers, but also can be offloaded to other available terminal devices (referred to as idle terminals) nearby through D2D communication technology. However, there are some important challenges in implementing collaborative offloading in MEC networks. Executing computing tasks requires consuming computing resources, and idle terminals are selfish and unwilling to contribute their limited computing resources.

[0003] In order to attract idle terminals and edge servers to participate in collaborative computing, it is crucial to introduce an incentive mechanism in the MEC system, but the existing resource allocation and scheduling methods introduce an incentive mechanism, but do not consider the collaborative offloading between different terminal devices, and do not consider the reasonable use of the available resources of the terminal devices; also do not consider the influence of task scheduling order on task completion delay, which is limited to the sorting problem of processing only a single task at the same time in the real situation, and does not explore the scene of parallel processing of multiple tasks by idle terminals at the same time, which wastes computing resources and does not achieve the goal of maximizing system utility. SUMMARY

[0004] The purpose of the present application is to provide a task allocation and scheduling method based on multi-round auction, which is aimed at the scene of parallel processing of multiple tasks by terminal devices at the same time, minimizes the delay, energy consumption and penalty cost, and increases the number of effectively processed tasks, and improves the utilization rate of resources.

[0005] In order to achieve the above purpose, the present application provides a task allocation and scheduling method based on multi-round auction, comprising the following steps:

[0006] Step 1: constructing an auction framework comprising an auctioneer, a buyer and a seller;

[0007] Step 2: after receiving the task information published by the auctioneer, the resource provider submits a task bid to the auctioneer;

[0008] Step 3: The auctioneer selects the winner corresponding to each task based on the bidding performance ratio of each task;

[0009] Step 4: Task scheduling order scheme based on urgency-profit ratio;

[0010] Step 5: Send the optimal result obtained in step 4 to the task performer;

[0011] Step 6: After the completion of the task, the resource provider charges the corresponding buyer and returns the calculation result.

[0012] Optionally, the auctioneer in step 1 is a trusted third party, i.e. a central cloud, an edge cloud or a professional auction agency, the buyer is a busy terminal, the seller is an idle terminal, all participants are honest in the auction process, and the bid satisfies p m,j (k) = c m,j .

[0013] Optionally, during the execution of step 2, the edge server sends the information (d m ,C m ,t m ,τ m ) of each task to the resource provider for auction, where d m is the data volume of the computing task , C m is the computing resource required for processing the task, τ m is the maximum allowed time delay of the task , t m is the time when the task is generated on the busy terminal, denotes the set of busy terminals, denotes the set of all idle terminals;

[0014] After receiving the task information, the resource provider submits a task bid to the auctioneer, where p m,j (k) represents the bidding price of the resource provider for the task in the kth round of auction, f j represents the provided computing capacity, and denotes the set of resource providers of all tasks in the system (index 0 represents the edge server).

[0015] Optionally, the bidding performance ratio expression in step 3 is as follows:

[0016]

[0017] where δ ∈ [0, 1] is a control coefficient.

[0018] Optionally, each computing task in step 4 First, match with the resource provider with the lowest bid performance ratio, for There are two cases:

[0019] Resource provider The number of matched tasks is less than or equal to the maximum number of tasks that can be processed simultaneously;

[0020] Resource provider The number of matched tasks is greater than the maximum number of tasks that can be processed simultaneously n j .

[0021] Optionally, when the number of matched tasks is less than or equal to the maximum number of tasks that the resource provider can process simultaneously, the utility of the computing system to complete the task is calculated, and the formula is as follows:

[0022]

[0023] Where v m is the true value of the task, p e,j , p t,j represent the economic cost of unit energy consumption and time delay, respectively;

[0024] For tasks with U m,j < 0, discard them, and for tasks with U m,j ≥ 0, add them to the matched task set , then update the waiting time delay and bid price on the resource provider for use in the next round of auction, then the of this round (the kth round of auction) will be updated as follows:

[0025]

[0026] Where represents the set of tasks allocated in the kth round of auction, and then the bid price is updated according to , then the corresponding bid performance ratio γ m,j value will also be updated to reflect the changes in waiting time delay and bid price.

[0027] Optionally, when the number of matched tasks is greater than the maximum number of tasks that the resource provider can process simultaneously, the tasks are arranged in descending order of profit-urgency ratio λ m,j , and the resource provider will select the first n j tasks, and if there is a task with a profit-urgency ratio λ j less than 0 among the first n m,j tasks, it will be discarded directly;

[0028] Profit-urgency ratio λ m,j The system utility is the ratio of system utility and time urgency, and is expressed as follows:

[0029]

[0030] The application provides a multi-round auction-based task allocation and scheduling method. For the scenario of simultaneous and parallel processing of multiple tasks by a terminal device, a multi-task scheduling sequence model is constructed based on a D2D cooperative MEC architecture. Busy terminals are regarded as buyers, idle terminals are regarded as sellers, a third-party central cloud, an edge cloud, or a professional auction agency is regarded as an auctioneer. The edge cooperative computing offloading and task scheduling are combined to improve the system utility based on an incentive mechanism. Specifically, a resource provider selection strategy based on a bid performance ratio and a task scheduling sequence strategy based on a greedy profit-urgency ratio are adopted to realize matching of optimizers and optimal task processing sequences. Simulation results show that the application increases the number of effectively processed tasks and the system utility, fully mobilizes the advantages of edge cooperative task offloading and flexible adjustment of task processing sequences, and improves resource utilization. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0032] Figure 1 is a step flowchart of a multi-round auction-based task allocation and scheduling method of the present application.

[0033] Figure 2 is a multi-task scheduling sequence model structure diagram of a specific embodiment of the present application.

[0034] Figure 3 is a task scheduling diagram in a specific embodiment of the present application.

[0035] Figure 4 is a time-delay-based penalty function relationship diagram in a specific embodiment of the present application.

[0036] Figure 5 is an execution time v.s. task quantity diagram in a specific embodiment of the present application.

[0037] Figure 6 is a system utility v.s. task quantity diagram in a specific embodiment of the present application.

[0038] Figure 7is a schematic diagram of the effective processing task number v.s. the task number in the embodiment of the present application.

[0039] Figure 8 is a schematic diagram of the effective processing task number v.s. the task number in the embodiment of the present application.

[0040] Figure 9 is a schematic diagram of the effective processing task number v.s. the task number in the embodiment of the present application. DETAILED DESCRIPTION

[0041] Embodiments of the present application are described in detail below with reference to examples illustrated in the accompanying drawings, in which the same or similar components are denoted by the same or similar reference numerals throughout. The embodiments described below by reference to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as limiting the present application.

[0042] Referring to Figure 1 , the present application provides a multi-round auction-based task allocation and scheduling method, comprising the following steps:

[0043] S1: constructing an auction framework comprising auctioneers, buyers and sellers;

[0044] S2: after receiving the task information published by the auctioneer, the resource provider submits a task bid to the auctioneer;

[0045] S3: the auctioneer selects the winner corresponding to the task based on the bid performance ratio of each task;

[0046] S4: task scheduling order scheme based on urgency-profit ratio;

[0047] S5: sending the optimal result obtained in step S4 to the task performer;

[0048] S6: after the task is completed, the resource provider charges the corresponding buyer and returns the calculation result.

[0049] The following is further described in combination with specific embodiments and execution steps:

[0050] As Figure 2 shown, the multi-round auction-based task allocation and scheduling method of the present application is specifically executed on an MEC system that can realize D2D offloading and flexibly adjust the task processing order. The corresponding system is composed of an edge cloud pool (edge server) and multiple terminal devices. The terminal devices include two types of busy terminal devices generating task requests and idle terminal devices (such as notebook computers, tablets, etc. intelligent devices) having available resources, so that represents a set of busy terminal devices, denotes the set of all idle terminals. The computing tasks generated by busy terminals are offloaded to either edge servers over cellular links or idle terminals over D2D links to be processed to meet their own task demands, and then the edge servers and idle terminals sort and process the tasks offloaded to them in parallel. Let denote the set of all resource providers of tasks in the system (index 0 denotes the edge server), then the available computing capacity of the resource providers is denotes that f j denotes the resource provider The CPU frequency (clock cycles per second) of the resource provider (the total of idle terminals and edge servers), n j denotes the maximum number of tasks that can be processed simultaneously.

[0051] Correspondingly, in the execution process of the present application, the system needs to go through the task offloading, task transmission and computing, and task scheduling sequence processes.

[0052] The present application assumes that busy terminals do not process tasks locally due to their limited computing capacity. To simplify the calculation, each terminal has one computing task to be offloaded for processing, and the tasks are all independent and indivisible, then the set of tasks to be processed in the system can be represented as The computing task generated by the busy terminal is represented by (d m , C m , τ m , t m ), where d m is the data volume (in bits) of the computing task , C m is the computing resource (in cpu cycles) required to process the task, τ m is the maximum allowed time delay of the task , and t m is the time when the task is generated on the busy terminal. For introduce the offloading decision variable where x m,j = 1 indicates that the task is offloaded to the resource provider for processing, otherwise x m,j = 0. At this time, there are the following constraint conditions, which indicate that the busy terminal can offload to at most one resource provider:

[0053]

[0054] ​Busy terminals offload computational tasks to edge servers or idle terminals for processing. The entire offloading process consists of three parts: first, the task is uploaded wirelessly to a resource provider (edge ​​server or idle terminal) to await processing; second, the resource provider determines the start time of task processing and allocates computing resources for processing; and finally, the result is returned to the user. Since the amount of data transmitted back is much smaller than the amount of data transmitted up, latency and energy consumption during result transmission are not considered in this invention. In summary, the computational task... latency Including upload latency Waiting delay and processing latency System energy consumption includes the energy consumption of computing tasks during data transfer. and energy consumption

[0055] for The upload latency from the unloading point to the resource provider can be expressed as:

[0056]

[0057] in, Indicates the uplink transmission rate, w m,j Indicates resource provider The transmission bandwidth allocated for computation tasks is such that each busy terminal uses the same amount of bandwidth for data transmission, transmitting only one task at a time, P. m,j H represents uplink transmission power. m,j For channel gain, σ 2 This represents the channel noise power. Based on the transmission delay, the corresponding transmission energy consumption is:

[0058]

[0059] When the task Uploaded to resource provider Then, the resource provider allocates corresponding computing resources to process it, and the processing latency of the task at this time is:

[0060]

[0061] The corresponding computational energy consumption can then be obtained, expressed as:

[0062]

[0063] Among them, κ j idle terminal The energy consumption coefficient depends on the effective switching capacitor of the chip architecture.

[0064] Because resource providers have limited computing resources, they often cannot process all arriving tasks simultaneously. Later arriving tasks must wait in the task scheduling buffer (used to cache and schedule tasks that haven't been processed in time). For any resource provider... Introducing task scheduling decision variables To indicate the task to be unloaded from it. The scheduling strategy. Among them, Indicates task The start of processing This indicates the moment when task processing ends, and the following relation also exists:

[0065]

[0066] For in The tasks that begin processing at this time will only begin execution after all tasks initiated at this time have been completed. This is due to the resource provider. It can process a maximum of n at the same time. j There are a maximum number of tasks (maximum number of tasks to process), so the following constraints apply:

[0067]

[0068] in, Represents a set In addition to the task Other tasks besides It is an indicator function; if the dot is true, it takes the value 1, otherwise it takes the value 0.

[0069] Next, we will illustrate the specific process of task sorting with an example. For instance, there are five tasks in the system that need to be unloaded to a resource provider. The computational tasks are processed on the resource provider. A maximum of 2 tasks can be processed simultaneously. The known task scheduling strategy is... The specific scheduling diagram of the task is as follows: Figure 3 Example.

[0070] To maximize system utility, this invention also considers system utility from the perspectives of busy terminals and resource providers, and models and constrains the problem.

[0071] Busy terminals offload tasks to resource providers (edge ​​servers and idle terminals) for computation, thereby expanding computing power, meeting task latency requirements, and providing corresponding rewards. Tasks that time out and fail are penalized. The following sections provide specific definitions of user utility, resource provider utility, and system utility.

[0072] (1) Busy terminal utility

[0073] For busy terminals The utility of a task is defined as its value v m Subtracting the price p paid to the resource provider m,j And the transmission energy cost to offload the task Can be expressed as:

[0074]

[0075] (2) The utility of the resource provider

[0076] Considering that the computing resources available to the resource provider are limited and cannot provide satisfactory service for all computing tasks (the present application uses task delay as an indicator to measure the quality of service QoS), in order to motivate the resource provider to meet the delay requirements of the task to the greatest extent, the present application designs a delay-based penalty function, that is, when the task processing time exceeds its maximum delay, it will cause a loss of task value, which will be translated into a penalty to the resource provider, and the penalty function is expressed as follows:

[0077]

[0078] Wherein, The function represents the change of task value with delay, and the expression is as follows:

[0079]

[0080] Wherein, τ m is the maximum allowed delay of the task, v m is the true value of the task, and α is a parameter that controls the degree of decline of the curve when the delay exceeds τ m . Figure 4 The relationship between the penalty function And the delay .

[0081] In summary, the cost of the resource provider Processing computing tasks Can be expressed as:

[0082]

[0083] Wherein, And p e,j , p t,j respectively represent the economic cost of unit energy consumption and delay.

[0084] The utility of the resource provider Processing computing tasks Can be expressed as:

[0085]

[0086] (3) System utility

[0087] This invention defines system utility as a weighted sum of busy terminal utility and resource provider utility, which can be expressed as:

[0088]

[0089] Based on the above description, this invention maximizes system utility and provides users with satisfactory service quality within a D2D collaborative MEC system architecture. Therefore, the overall optimization problem can be formulated as the following P1 problem:

[0090]

[0091] P1 contains two types of decision variables: unloading decision variable X and task scheduling decision variable θ={t s ,t e Where C1 and C2 are offloading decision constraints, indicating that a task can only be offloaded to a resource provider; constraint C3 indicates that a task can only be processed after it is uploaded to a resource provider; constraint C4 indicates the processing latency of a task on a resource provider; and constraint C5 indicates that the number of tasks processed in parallel at the same time cannot exceed the maximum number of tasks that the resource provider can process at the same time.

[0092] In problem P1, task allocation and scheduling with the goal of maximizing system utility mainly focus on two sub-problems: 1) deciding where to unload tasks, and 2) determining the processing order on task resource providers.

[0093] Steps S2 to S6 detail the problem-solving process. In the multi-round auction process, busy terminals are buyers, idle terminals are sellers, and trusted third parties (central cloud, edge cloud, or professional auction houses) are auctioneers. The edge server stores information about each task in the system (d... m C m ,t m ,τ m The task information is sent to resource providers for auction. After receiving the task information, the resource providers submit their task bids to the auctioneer. These represent the resource providers at the k-th round of the auction. For the task The bid price p m,j (k) The computing power provided f j And the waiting time required for the task. This invention assumes that all participants are honest and that the bids satisfy p. m,j (k)=c m,j .

[0094] Step S2: Submit the bid

[0095] After receiving the task information, the resource provider submits a task bid to the auctioneer.

[0096] Step S3: Resource provider selection strategy based on bid performance ratio (BPR)

[0097] After receiving bids from all resource providers, the auctioneer calculates the bid performance ratio (BPR) for each task and then determines the winner for each task. resource providers The bid is defined as a weighted average of the bid price to the resource provider's processing capacity and the data size to the transmission rate. This invention uses parameter γ. m,j The bid performance ratio (BPR) is expressed as follows:

[0098]

[0099] Here, δ∈[0,1] is a control coefficient that busy terminals can adjust based on whether they require more powerful processing capabilities or more stable transmission. According to the definition of Bid Performance Ratio (BPR), busy terminals aim to achieve the greatest performance improvement at the lowest cost. Therefore, resource providers with a lower BPR should have a higher chance of winning bids to meet the service quality requirements of busy terminals.

[0100] Step S4: Task scheduling order scheme based on urgency-profit ratio

[0101] The matching process between tasks and resource providers operates in a round-robin fashion to determine the match between tasks and resource providers. The matching process in each round is as follows. Each computational task... First, match with the resource provider with the lowest bid performance ratio (BPR). For The following two situations may occur:

[0102] (1) Resource Provider If the number of matched tasks is less than or equal to the maximum number of tasks it can process simultaneously, the system's utility in completing the tasks is calculated using the following formula:

[0103]

[0104] For U m,j Tasks with a value less than 0 will be discarded, U m,j Tasks with a value ≥ 0 are added to the matching task set. Then update the resource provider. The waiting time and bid price are used for the next round of auction, then the current auction (the k-th auction) The update is as follows:

[0105]

[0106] wherein, denotes the set of tasks allocated at the kth round of auction, and then the bid price is updated according to The corresponding bid performance ratio γ m,j will also be updated to reflect the changes in the waiting time and the bid price.

[0107] (2) Resource provider If the number of matched tasks is greater than the maximum number of tasks n j that it can handle at the same time, in this case, the tasks will be arranged in descending order of the profit-urgency ratio λ m,j , and the resource provider will select the first n j tasks, if there is a task with a profit-urgency ratio λ j less than 0 in the first n m,j tasks will be discarded directly to maximize the system utility. For the profit-urgency ratio λ m,j , the present invention defines it as the ratio of the system utility to the time urgency, which is expressed as follows:

[0108]

[0109] The corresponding task will then be added to the set , and the resource provider 's waiting time , bid price and bid performance ratio γ value will be updated according to the updated m,j , and then all the remaining tasks matched with the resource provider in this round of auction will be moved to the set After all the resource providers are matched, if , the next round of auction will be carried out, and the auction will stop until each task is matched with a resource provider.

[0110] Step S5: Send the optimal result obtained in step S4 to the task performer;

[0111] Step S6: After the task is completed, the resource provider charges the corresponding buyer and returns the calculation result.

[0112] Further, the performance of the method is evaluated by simulation experiment and comparison with different algorithms, including the offloading algorithm of only offloading tasks to the edge server (edge cloud offloading algorithm), the offloading algorithm of random processing order (RS), and the offloading algorithm of first-in-first-out task processing order (FIFO). Hereinafter, the method of the present application is referred to as BMRA-TAS.

[0113] The simulation scenario is set as follows:

[0114] 1 WiFi-based wireless access point and the corresponding edge computing node are set, the coverage radius is 350 m, and the number of busy terminals is set from 10 to 200, and the number of idle terminals is set from 30 to 90, which are uniformly distributed in the coverage range. The maximum transmission radius of D2D is R d2d ≤ 25 m. The computing capacity of the idle terminal is set as f j =[1.5, 3.5] GHz, the computing capacity of the edge node is f0=7 GHz, and the unit delay and unit energy consumption prices are set as [0.25, 0.5] $ and 0.3 $ respectively. The channel gain between the busy terminal and the resource provider is set as wherein, δ=4 represents the path loss factor, D m,j represents the distance between the busy terminal m and the resource provider j, and the transmission power is 1 W, the total transmission bandwidth B=40 MHZ, and the noise power spectral density is -174 dBm / Hz.

[0115] For the computing task generated by the busy terminal , the task evaluation in the simulation is set as v m =[0.1, 10] $, the data size is set as d m =[5, 10] MB, the maximum allowable delay τ m =[0.8, 1] s, and the required computing resource C m =[0.5, 1.2] Gcycles.

[0116] Figure 5 The relationship between the running time of the BMRA-TAS method and the number of busy terminals is shown. Obviously, with the increase of the number of tasks, the running time gradually increases, and when the number of tasks is large, the running time is only about 0.7 s, which will not have too much impact on the execution of the task and will not affect the real-time requirement of the offloaded task in the MEC system.

[0117] Figure 6 , 7 The trends of system utility and effective processing task number with the number of task requests and the number of resource providers are shown. Specifically, in Figure 6In the middle, the system utility increases with the increase of the number of tasks, but the utility at CoL=90 is significantly higher than that at CoL=70 and CoL=50. This shows that more idle terminals can improve the system processing efficiency, meet the delay requirements of more tasks, thereby reducing the penalty cost caused by the system completing the task due to timeout, and thus improving the overall utility of the system. However, when the number of tasks increases to a certain value, the system utility gradually tends to be flat, because when the offloaded tasks exceed the system processing capacity, the timeout penalty cost increases, the system utility increases slowly, and at the same time, the tasks with negative system utility are discarded, so the system utility gradually tends to be flat. In Figure 7 In the middle, the number of effective service task requests increases first and then tends to be flat with the increase of the number of busy terminals and idle terminals. This is because the system computing resources are limited, and as the number of task requests increases, the system will reach the maximum processing capacity. When CoL=70, the number of effective requests is significantly higher than that at CoL=50 and CoL=30, because more idle terminals can better cope with the requests generated by a large number of busy terminals, share the burden of processing tasks, and improve the service capacity of the overall system.

[0118] Figure 8 , Figure 9 By comparing with different algorithms, the performance of the BMRA-TAS algorithm proposed in the present application is evaluated (CoL=70 is set). Among them, from Figure 8 It can be seen that the BMRA-TAS method proposed in the present application is superior to the edge cloud offloading algorithm, the RS offloading algorithm and the FIFO offloading algorithm. In addition, the BMRA-TAS curve is always in the rising state, while for the RS offloading algorithm and the FIFO offloading algorithm, because both algorithms cannot guarantee the rationality of the participants, with the increase of the penalty cost caused by the task violating the delay requirement and the task processing cost, the curve will first rise and then rapidly decline. The curve of the edge cloud offloading algorithm rises and then quickly tends to be flat, because the edge cloud computing resources are limited, and with the increase of the number of tasks, the edge cloud quickly reaches the processing limit, and the tasks exceeding the processing capacity of the edge cloud are directly discarded, resulting in no profit for the system, so the curve tends to be flat. Figure 9 It can be seen in the middle that under the control of the BMRA-TAS method proposed in the present application, the number of effective task requests (i.e. the number of tasks completed within the delay allowed) is greater than that of the edge cloud offloading algorithm, the RS offloading algorithm and the FIFO offloading algorithm, thereby proving that under the control of the BMRA-TAS, the system can increase the number of effective task requests while improving the system utility by adjusting the execution order of the computing tasks.

[0119] The above disclosed is only a preferred embodiment of the present application, of course, cannot be limited by this, the person skilled in the art can understand that the whole or part of the above-mentioned embodiment is realized, and the equivalent change made according to the claims of the present application still belongs to the scope covered by the present application.

Claims

1. A method for task allocation and scheduling based on multi-round auction, characterized in that, Comprising the following steps: Step 1: Construct an auction framework containing auctioneers, buyers and sellers; The auctioneer in step 1 is a trusted third party, namely a central cloud, an edge cloud or a professional auction agency, the buyer is a busy terminal and the seller is an idle terminal, and all participants in the auction process are honest; Step 2: After receiving the task information published by the auctioneer, the resource provider submits a task bid to the auctioneer; During the execution of step 2, the auctioneer sends information (d m , C m , t m , τ m ) of each task to the resource providers for auction, where d m is the data volume of the computing task m∈M, C m is the computing resource required for processing the task, τ m is the maximum allowed time delay of the task m∈M, t m is the time when the task m∈M is generated on the busy terminal, M={1,2,...,M} represents the set of busy terminals, and C={1,2,...,j,...,C} represents the set of all idle terminals. The resource provider receives the task information and submits a task bid to the auctioneer p m,j (k) denotes the bid price of the resource provider j e J for the task m e M at the kth round of auction, f j for the provided computing power, the waiting latency required for the task, J = {C U 0} denotes the set of resource providers for all tasks in the system, and index 0 represents the edge server; Step 3: The auctioneer selects the winner corresponding to the task based on the bidding performance ratio of each task; The bidding performance ratio expression in step 3 is as follows: wherein, δ ∈ [0, 1] is a control coefficient, r m,j is the uplink transmission rate transmitted by task m to resource provider j; Step 4: Optimize the task scheduling sequence based on the urgency-profit ratio; Step 5: Send the optimal result obtained in step 4 to the task performer; Step 6: After the task is completed, the resource provider charges the corresponding buyer and returns the calculation result.

2. The multi-round auction-based task allocation and scheduling method according to claim 1, characterized in that, Each computing task m e M is first matched with the resource provider with the lowest bid performance ratio in Step 4, for The following two cases can occur: The number of matched tasks of a resource provider j e J is less than or equal to the maximum number of tasks n that it can handle simultaneously j ; The number of matched tasks of a resource provider j∈J is greater than the maximum number of tasks n that it can handle simultaneously j .

3. The multi-round auction-based task allocation and scheduling method according to claim 2, characterized in that, When the number of matched tasks is less than or equal to the maximum number of tasks that the resource provider can handle at the same time, the utility of the computing system to complete the task is calculated, and the formula is as follows: where v m is the true valuation of the task, p e,j , p t,j represent the economic cost of unit energy consumption and time delay, k j is the energy consumption coefficient of the resource provider j, is the penalty function based on time delay; For U m,j Tasks with U m,j ≥ 0 are added to the matching task set M assigned and the waiting delay and bid price on resource providers j e J are updated for the next round of auction, then the will be updated as follows: wherein, denotes the set of tasks allocated at the kth round of auction, then the corresponding bidding performance ratio γ is updated, the bidding performance ratio γ m,j will also be updated to reflect the changes in waiting time delay and bidding price.

4. The multi-round auction-based task allocation and scheduling method according to claim 3, characterized in that, When the number of tasks to be matched is greater than the maximum number of tasks that a resource provider can handle at the same time, the tasks are sorted in descending order of profit-urgency ratio λ m,j The resource provider j∈J will select the first n j tasks, if there is a task with profit-urgency ratio λ j less than 0 in the first n m,j tasks will be discarded directly; Profit-urgency ratio λ m,j The system utility-to-time-urgency ratio, denoted as λ, is given by: wherein, is the upload latency, is the processing latency.

Images