A edge cloud computing access control and scheduling method and system based on Internet of Things applications

Abstract

The present invention discloses an edge cloud computing access control and scheduling method based on an Internet of Things application, comprising the following steps: an Internet of Things application terminal submits a task request to a task evaluator; the task evaluator receives the task request and evaluates whether the task meets the latency requirement; the task scheduler receives the task from the task evaluator and schedules the task to be executed on an edge data center or cloud according to the scheduling strategy and user service quality requirements; the resource manager receives the task request and allocates relevant resources to the task; the resource manager or cloud data center returns the scheduling result of the task execution to the Internet of Things application terminal. The present invention also discloses an edge cloud computing access control and scheduling system based on an Internet of Things application. The edge cloud computing access control and scheduling method and system based on an Internet of Things application provided by the present invention can meet the requirements of Internet of Things application tasks and is more suitable for an edge-cloud collaborative environment.

Description

Technical Field

[0001] The present invention relates to an edge cloud computing admission control and scheduling method and system based on Internet of Things applications, belonging to the technical field of edge cloud computing. Background Art

[0002] IoT applications leverage modern information and communication technologies to enable sensors in devices such as smart home appliances, mobile phones, and surveillance cameras to perceive their surroundings in real time. These devices connect to the internet via cellular and wireless networks. These IoT devices periodically collect data and upload it locally or to cloud servers for analysis and processing, providing knowledge and intelligent services. The increasing variety and functionality of these IoT applications place increasing demands on computing power and response latency. These applications require real-time processing of collected video information and feedback to IoT users, placing high demands on real-time performance. Therefore, latency has become a key metric for IoT applications. Currently, data processing for these latency-sensitive IoT applications primarily utilizes cloud computing. However, the long-distance communication between these and remote cloud computing platforms results in significant latency for end-user applications, making them unsuitable for latency-sensitive IoT applications.

[0003] On the other hand, the cloud computing model itself has certain flaws, namely, excessive resource centralization, typically concentrated in a few large data centers. This creates insurmountable latency barriers. Edge computing, on the other hand, provides services in a distributed resource model. In this model, resources are dispersed and geographically close to users. This proximity and resource distribution can provide high-bandwidth and low-latency services. However, edge computing lacks significant computing and storage capabilities, and cannot provide sufficiently powerful computing or storage services. Therefore, edge-cloud collaboration presents a promising solution. Edge computing can address the shortcomings of traditional centralized cloud computing. Cloud computing's powerful computing and storage capabilities can compensate for these shortcomings of edge computing. The edge and cloud form a complementary end-to-end system, achieving a coordinated architecture of centralized cloud and distributed edge computing, thereby providing more flexible and lower-latency services for IoT applications. In existing technologies, the urgent time requirements of IoT application tasks make traditional scheduling methods difficult to meet, and they cannot effectively guarantee task assessment and control of latency. Furthermore, due to the complexity of edge-cloud collaboration environments, it is difficult to adapt to edge-cloud collaboration. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the defects of the existing technology and provide an edge cloud computing access control and scheduling method and system based on IoT applications that can meet the needs of IoT application tasks and is more suitable for edge cloud collaborative environment.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A method for edge cloud computing admission control and scheduling based on Internet of Things applications includes the following steps:

[0007] The IoT application submits a task request to the task evaluator;

[0008] The task evaluator receives the task request and evaluates whether the task meets the delay requirement. If the task meets the delay requirement, it will hand the task over to the task scheduler. If the task does not meet the delay requirement, it will return the result of not meeting the delay requirement to the IoT application and notify the IoT application to resubmit the task with a new deadline.

[0009] The task scheduler receives tasks from the task evaluator and schedules them to the edge data center or cloud for execution based on the scheduling policy and user service quality requirements. When a task is suitable for execution on the cloud, the task scheduler sends the task to the cloud data center for execution. When a task is suitable for execution in the edge data center, the task scheduler sends the task to the resource manager.

[0010] The resource manager receives task requests and allocates relevant resources to the tasks. At the same time, the resource manager monitors the resource status of the edge data center. The monitoring information includes the execution status of the task and the usage of resources. The resource manager regularly updates the monitoring information and feeds it back to the task evaluator to support the task evaluator's evaluation.

[0011] The resource manager or cloud data center returns the scheduling results of task execution to the IoT application.

[0012] In the IoT application side, the scheduling time window is TS, the number of task requests submitted by users in the time window is N, and the IoT application task set in the current time window is T = {T1, T2, ..., T N}, the i-th IoT application task T i Described as a sextet <T i,id ,T i,len, T i,com ,T i,bw ,T i,sto ,T i,dl >, where T i,id Indicates the task number submitted by the IoT application, T i,len Indicates the task length, T i,com ,T i,bw ,T i,sto They represent the computing, bandwidth and storage requirements of IoT application tasks, respectively. i,dlIndicates the deadline requirement of the IoT application end; the number of available edge resources in the resource manager is X×Y, where X represents the number of edge data centers, Y represents the number of resources in each edge data center, and the available edge resource set is R e ={R1,R2,…,R X×Y}, the edge resource R e The jth resource R j,e Described as a quad <R j,id ,R j,com ,R j,bw ,R j,sto >; The number of available cloud resources in the cloud data center is D, and the set of available cloud resources is R c ={R1,R2,…,R D}, the cloud resource R c The kth resource R k,c Described as a four-tuple <R k,id ,R k,com ,R k,bw ,R k,sto >, where R *,id Indicates the resource number, R *,com ,R *,bw , R *,sto Represents the computing, bandwidth, and storage capabilities of the resource respectively.

[0013] During the task scheduling process, the task scheduler needs to meet the following constraints: 1) The edge resources of the resource manager or the cloud resources of the cloud data center can only process one task at a time, but all resources can be executed in parallel; 2) All tasks need to be assigned to the edge resources of the resource manager or the cloud resources of the cloud data center for execution, and each task can only be executed on one edge resource or cloud resource.

[0014] The resource monitor in the resource manager is configured with a delay model, which is used to calculate the average delay d of the edge system. The specific calculation method is as follows:

[0015] Input known tasks and resource status T i ,R j ,N,x,X,y,Y and the resource processing capacity of data center x h x ;

[0016] Initialize the number of requests currently accessing edge data center x and the number of available resources f′ currently in edge data center x x =0,g′ x =0;

[0017] If the number of requests received by edge data center x during TS is f x (t)≠0 or TS

[0018] The number of requests that have not been served by the edge data center q x,y (t)≠0; then calculate

[0019]

[0020]

[0021] f′ x =f′ x +f x,y ;

[0022] g′ x =g′ x +g x,y ;

[0023] Among them, T i,dl Represents the deadline of the IoT application task, t x,y For task T i The average round-trip time to resource y in edge data center x, h x is the resource processing capacity of data center x;

[0024] Otherwise f x,y =0;g x,y =0;

[0025] Calculate the average delay d of the edge computing system

[0026]

[0027] In the Resource Manager, apply task T i The execution time of edge resource j includes two parts: transmission time and computing time, namely t j,e =t j,com +t j,tran

[0028] The transmission time t j,com and calculation time t j,tran The calculation formula is as follows:

[0029]

[0030]

[0031] Among them, T i,len represents the task length;

[0032] In cloud data centers, IoT application tasks T i Execution time t on cloud resources j,c Including calculation time t j,com , transmission time t k,tranand edge-to-cloud migration time t mig , t k,c =t k,com +t k,tran +t mig

[0033] where t k,com and t k,tran The calculation formula is as follows.

[0034]

[0035]

[0036]

[0037] Among them, T i,bw Represents the mission bandwidth requirement.

[0038] The task evaluator's task evaluation specifically includes:

[0039] Input known tasks, resource conditions and current delay T i ,R j ,R k ,N,X,Y,d;

[0040] The average delay d in getting scheduled updates from the resource manager;

[0041] Calculate IoT application tasks T i Estimated time t at edge, cloud, and edge cloud j,e , t k,x and t i , t i =ω e ×t j,e +ω c ×t k,c

[0042] The 0-1 control variable ω e and ω c The value of is calculated as follows:

[0043]

[0044]

[0045] ω e ≠ω c ;

[0046] The task is allowed to be admitted based on the estimated execution time and delay of the task in the edge cloud. If t i <T i,dl or T i,dl <d, then T iAdmission, otherwise T i block.

[0047] The task scheduler specifically schedules tasks including:

[0048] Input IoT application tasks and resource related parameters, task request number T i , R j ,R k ,q x,y (t),g x,y ;

[0049] Calculate the average resource requirement R of each scheduling cycle task avg (t);

[0050]

[0051] Among them, P max It is the maximum number of tasks that can be accommodated in each scheduling cycle. Its calculation formula is as follows

[0052] P max ≥N;

[0053] In each scheduling cycle, a threshold θ is set for edge resources. θ is dynamically adjustable and is set to queue q in each cycle. x,t (t) Number of allocated resources g x,y (t) time average expectation, that is: θ = g x,y ;

[0054] Task scheduler based on R avg (t) and θ values to judge, if R avg (t)>θ, indicating that the current edge data center load is too large, then the task is scheduled to the cloud data center for execution; otherwise, the task is scheduled to the resource manager for execution. In terms of scheduling in the edge data center, the resource manager adopts the strategy of scheduling all unserved queues q in the edge data center. x,y (t) Sort the current task T i Dispatch to q x,y (t) Smallest data center.

[0055] An edge cloud computing admission control and scheduling system based on Internet of Things applications includes: an Internet of Things application end, a task evaluator, a task scheduler, a resource manager and a cloud data center, wherein the Internet of Things application end is used to submit a task request to the task evaluator; the task evaluator is used to receive the task request, evaluate whether the task meets the delay requirement, and if the task meets the delay requirement, the task is handed over to the task scheduler; if the task does not meet the delay requirement, the result of not meeting the delay requirement is returned to the Internet of Things application end, and the Internet of Things application end is notified to resubmit the task with a new deadline; the task scheduler is used to receive the task from the task evaluator, and schedule the task to the edge data center according to the scheduling strategy and user service quality requirements. When a task is suitable for execution on the cloud, the task scheduler transmits the task to the cloud data center for execution; when a task is suitable for execution on the edge data center, the task scheduler transmits the task to the resource manager; the resource manager is used to receive task requests and allocate relevant resources to tasks, and return the scheduling results of task execution to the Internet of Things application end. At the same time, the resource manager monitors the resource status of the edge data center, and the monitoring information includes the execution status of the task and the usage of resources, and regularly updates the monitoring information and feeds it back to the task evaluator as support for the evaluation of the task evaluator; the cloud data center is used to return the scheduling results of task execution to the Internet of Things application end.

[0056] Beneficial effects of the present invention: The present invention provides an edge cloud computing access control and scheduling method and system based on Internet of Things applications, which can automatically evaluate whether the control task is allowed according to the Internet of Things application task parameters and resource conditions, avoid system congestion, ensure the task's delay requirements, and ensure the service quality of the Internet of Things application. At the same time, it can automatically calculate and update the delay according to the resource status parameters, save scheduling time, and improve the scheduling efficiency of Internet of Things application tasks in the edge-cloud collaborative environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a flow chart of a method for edge cloud computing admission control and scheduling based on Internet of Things applications in the present invention. DETAILED DESCRIPTION

[0058] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention.

[0059] The present invention discloses a method and system for edge cloud computing access control and scheduling based on Internet of Things applications. The key to this method is to dynamically consider the resource utilization in real time based on parameters such as the size of the Internet of Things application task, the requirements for computing, storage and bandwidth, and the computing, storage and bandwidth capabilities of the edge cloud resources themselves, evaluate whether the current new task is allowed, and schedule the allowed tasks to solve the delay requirement problem of Internet of Things application tasks.

[0060] like Figure 1 As shown, the present invention provides an edge cloud computing access control and scheduling method based on Internet of Things applications, which specifically includes the following steps: the Internet of Things application end submits a task request to the task evaluator, evaluates whether the task meets the delay requirements, and if the delay requirements are met, hands the task to the task scheduler, which schedules the task, and the resource manager regularly feeds back the load and usage of the edge data center resources to the task evaluator and task scheduler.

[0061] The task evaluator evaluates the task and decides whether to accept or reject the request. This decision is based on QoS (Quality of Service) and latency guarantees, meaning that the task must be completed before the deadline. If the task cannot be completed on time, i.e., violates QoS, the task is rejected, a result is returned to the user, and the user is notified to resubmit the task with a new deadline. If the task is accepted, the task evaluator transmits it to the task scheduler. The task scheduler schedules the task to an edge data center or a cloud data center based on the scheduling policy and user QoS requirements. If the task is suitable for cloud execution, the task scheduler transmits it to the cloud data center; otherwise, the task scheduler transmits it to the resource manager. The resource manager is responsible for receiving task requests and allocating relevant resources to the task. It also monitors the resource status of the edge data center, including task execution status and resource usage, and regularly updates this monitoring information to the task evaluator to support its evaluation. Finally, the task manager or cloud data center returns the task execution results to the IoT application.

[0062] In this paper, tasks submitted by IoT applications are divided into time slots (TS) of equal length, set to Δt seconds. These slots are run in time slices, and the user request scheduling method is determined at the start and end time of each TS. Assume that the edge-cloud collaborative environment here includes X edge data centers and one cloud data center, denoted as sets X and D, respectively, where the xth data center contains Y resources.

[0063] Assume that the scheduling time window is TS, the number of task requests submitted by users in the time window is N, the number of available edge resources is X×Y, the number of available cloud resources is D, and the IoT application task set in the current time window is T={T1,T2,…,TN}, the available edge resource set is R e ={R1,R2,…,R X×Y}, the available cloud resource set is R c ={R1,R2,…,R D}, the IoT application tasks and resources are described as follows.

[0064] 1) IoT application task model: The i-th IoT application task T i Described as a sextet <T i,id ,T i,len ,T i,com ,T i,bw ,T i,sto ,T i,dl >, where T i,id Indicates the task number submitted by the IoT application, T i,len Indicates the task length, T i,com ,T i,bw ,T i,sto They represent the computing, bandwidth and storage requirements of IoT application tasks, respectively. i,dl Indicates the deadline requirement of the IoT application.

[0065] 2) Resource model: Edge resources R e The jth resource R j,e Described as a quad <R j,id ,R j,com ,R j,bw ,R j,sto >, cloud resources R c The kth resource R k,c Described as a quad <R k,id ,R k,com ,R k,bw ,R k,sto >, where R *,id Indicates the resource number, R *,com ,R *,bw ,R *,sto Specifically, it represents the computing, bandwidth, and storage capabilities of the resource.

[0066] During the scheduling process, the following constraints need to be met:

[0067] 1) Resource limitations in the resource pool: The computing resources and storage space required for IoT application tasks must be within the scope of the resources available in edge and cloud data centers. Furthermore, the computing, storage, and bandwidth performance of edge and cloud resources vary. Furthermore, each resource can only process one task at a time, but all resources can execute in parallel.

[0068] 2) IoT application task execution restrictions: All tasks need to be assigned to specific edge or cloud resources for execution, and each task can only be executed on one edge cloud resource.

[0069] In the present invention, the delay model is deployed in the resource monitor of the resource manager, regularly updated and fed back to the task request evaluator, and each task request arriving at the edge cloud data center is evaluated in the task request evaluator to evaluate whether the task is accessed or blocked.

[0070] For a data center x, task requests are stored in a cache queue and resources are allocated according to a certain scheduling policy. For each task request, its maximum delay requirement is Tdl. If the delay exceeds Tdl, it violates its QoS requirement. In a data center x, let its resource processing capacity be h x , that is, each TS can process h x Requests. Let the number of requests received by a data center x during TS be f x (t), obviously:

[0071] f x (t)≤N(1)

[0072] All incoming requests will be sent to an available resource Y, that is,

[0073]

[0074] For a certain edge data center x, let q x,y (t) represents the number of requests that have not been served by the edge data center at TS. x,y (t) is expressed as q x,y (t) The number of resources allocated. Then

[0075]

[0076] Use g x (t) represents the total available resources of an edge data center x at TS, that is,

[0077] g x (t)=∑ y g x,y (t)(4)

[0078] ∑ y f x,y ≤N(5)

[0079] ∑ x g x,y ≤Y(6)

[0080] f x,y ,g x,y ,gx They are a i,j (t), f x,y (t), g x (t) is the time average expectation.

[0081]

[0082]

[0083]

[0084] Use t x,y Represents task T i The average round-trip time for resource y to the xth edge data center is calculated as follows

[0085]

[0086] t x,y Normalize to get

[0087]

[0088] Then we get f x,y and g x,y The following relationship

[0089]

[0090]

[0091] By f x,y and g x,y The formula for the average delay of the entire edge center can be obtained as follows:

[0092]

[0093] The specific calculation method of average delay is as follows

[0094] Step 1: Input known tasks and resource status T i ,R j ,N,xX,y,Y and the resource processing capacity of data center xh x ;

[0095] Step 2: Initialize the number of requests currently accessing edge data center x and the number of available resources f′ currently in edge data center x x =0,g′ x =0;

[0096] Step 3: If the number of requests received by edge data center x during TS is f x (t)≠0 or the number of requests that have not been served by the edge data center when TSx,y (t)≠0; then calculate

[0097]

[0098]

[0099] f′ x =f′ x +f x,y ;

[0100] g′ x =g′ x +g x,y ;

[0101] otherwise

[0102] f x,y =0;

[0103] g x,y =0;

[0104] Step 4: Calculate the average latency d of the edge computing system

[0105]

[0106] The average delay d of the edge system can be obtained, which provides support for subsequent scheduling and admission evaluation and control of new IoT application tasks.

[0107] The execution time estimation model of IoT application tasks in the edge cloud data center in this invention is as follows:

[0108] (1) Execution time estimation model for IoT application tasks in edge data centers

[0109] IoT Application Tasks i At the edge resource j (where j is different from y, The execution time includes two parts: transmission time and calculation time.

[0110] t j,e =t j,com +t j,tran (15)

[0111] where t j,com and t j,tran The calculation formula is as follows.

[0112]

[0113]

[0114] (2) Execution time estimation model for IoT application tasks in cloud data centers

[0115] IoT Application Tasks i Execution time t on cloud resources k,c This includes computing time, transmission time, and edge-to-cloud migration time.

[0116] t k,c =t k,com +t k,tran +t mig (18)

[0117] where t k,com and t k,tran The calculation formula is as follows.

[0118]

[0119]

[0120] Edge-to-cloud migration time t mig is the time to migrate from the edge data center to the cloud, which takes into account the current available bandwidth and the bandwidth required for task migration T i,bw Therefore, the maximum bandwidth is used as the migration bandwidth at this time.

[0121]

[0122] The new IoT application task request admission evaluation is deployed in the task request evaluator of the system model. For the incoming task request, the execution time t of the task on the edge and cloud is first calculated using formulas (15) and (18). j,e and t k,c , and use formula (22) and 0-1 control variable ω e and ω c Estimated execution time t of the task request i and compare it with task T i The deadline T i,dl Compare it with the average delay d of the edge system to determine whether the task request is admitted or blocked.

[0123] t i =ω e ×t j,e +ω c ×t k,c (twenty two)

[0124] The 0-1 control variable ω e and ω c The value of is calculated as follows.

[0125]

[0126]

[0127] ω e ≠ω c (25)

[0128] The specific IoT application task request admission evaluation and control methods are as follows:

[0129] Step 1: Input known tasks, resource conditions and current delay T i ,R j ,R k ,N,X,Y,d;

[0130] Step 2: Obtain the regularly updated average delay value d from the resource manager;

[0131] Step 3: Calculate the IoT application task T i Estimated time t at edge, cloud, and edge cloud j,e , t k,c and t i ;

[0132] Step 4: Determine whether the task is allowed to be accepted based on the estimated execution time and delay of the task in the edge cloud: If t i <T i,dl or T i,dl <d, then T i Admission, otherwise T i block.

[0133] Approved IoT application tasks are transmitted to the task scheduler, which determines whether to schedule the task to the cloud or the edge. If scheduling to the edge, it also determines which edge data center to schedule the task to. Regarding edge-cloud scheduling, the strategy employed is to schedule tasks to the cloud when edge resources are insufficient or edge data centers are overloaded, otherwise executing them on edge resources.

[0134] For IoT application tasks T i The judgment of whether the edge resource capacity is sufficient is based on the assumption that one task occupies one resource in the resource task model. Therefore, a simple judgment can be made using the storage and computing capabilities of the task and resource. That is, if T i,com >R j,com or T i,sto >R j,sto or T i,bw >R j,bw It is considered as marginal capacity deficiency, and the task T i Dispatch to the cloud.

[0135] The edge data center load is mainly judged based on the average resource requirements of tasks in each cycle.

[0136] Step 1: Enter IoT application task and resource related parameters, task request number T i , R j ,R k ,q x,y (t),g x,y ;

[0137] Step 2: Calculate the average resource requirement R of each scheduling period task avg (t).

[0138]

[0139] Among them, P max It is the maximum number of tasks that can be accommodated in each scheduling cycle. Its calculation formula is as follows

[0140]

[0141] P max ≥N(28)

[0142] Step 3: Set a threshold θ for edge resources in each scheduling cycle. θ is dynamically adjustable and is set to queue q in each cycle. x,y (t) Number of allocated resources g x,y (t) is the time average expectation, that is:

[0143] θ=g x,y (29)

[0144] Step 4: Task Scheduler based on R avg (t) and θ values to judge, if R avg (t)>θ, indicating that the current edge data center load is too large, the task is scheduled to the cloud for execution; otherwise, the task is scheduled to the edge for execution.

[0145] Step 5: In terms of scheduling of edge data centers, the strategy adopted is to schedule all unserved queues q in all edge data centers. x,y (t) to sort the current task T i Dispatch to q x,y (t) Smallest data center.

[0146] The present invention discloses an edge cloud computing admission control and scheduling system based on Internet of Things applications, comprising: an Internet of Things application end, a task evaluator, a task scheduler, a resource manager and a cloud data center, wherein the Internet of Things application end is used to submit a task request to the task evaluator; the task evaluator is used to receive the task request, evaluate whether the task meets the delay requirement, and if the task meets the delay requirement, hand the task to the task scheduler; if the task does not meet the delay requirement, the result of not meeting the delay requirement is returned to the Internet of Things application end, and the Internet of Things application end is notified to resubmit the task with a new deadline; the task scheduler is used to receive the task from the task evaluator, and schedule the task according to the scheduling strategy and user service quality requirements. The task is sent to the edge data center or the cloud for execution. When the task is suitable for execution on the cloud, the task scheduler transmits the task to the cloud data center for execution. When the task is suitable for execution in the edge data center, the task scheduler transmits the task to the resource manager; the resource manager is used to receive task requests and allocate relevant resources to tasks, and return the scheduling results of task execution to the IoT application end. At the same time, the resource manager monitors the resource status of the edge data center. The monitoring information includes the execution status of tasks and the usage of resources, and regularly updates the monitoring information and feeds it back to the task evaluator as a support for the task evaluator's evaluation; the cloud data center is used to return the scheduling results of task execution to the IoT application end.

[0147] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for edge cloud computing admission control and scheduling based on Internet of Things applications, characterized by: The following steps are involved: The IoT application submits a task request to the task evaluator; The task evaluator receives the task request and evaluates whether the task meets the delay requirement. If the task meets the delay requirement, it will hand the task over to the task scheduler. If the task does not meet the delay requirement, it will return the result of not meeting the delay requirement to the IoT application and notify the IoT application to resubmit the task with a new deadline. Compute the execution time t of the task on the edge and cloud j,e and t k,c , t i =ω e ×t j,e +ω c ×t k,c , where 0-1 control variable ω e and ω c The value of is calculated as follows: oh e ≠ω c ; The specific IoT application task request admission evaluation and control methods are as follows: Step 1: Input known tasks, resource conditions and current delay T i ,R j ,R k ,N,X,Y,d, where R j and R k are the jth and kth resources in edge and cloud computing respectively, N is the number of task requests submitted by users in the time window, X represents the number of edge data centers, and Y represents the number of resources in each edge data center; Step 2: Obtain the regularly updated average delay value d from the resource manager; Step 3: Calculate the IoT application task T i Estimated time t at edge, cloud, and edge cloud j,e , t k,c and t i ; Step 4: Determine whether the task is admissible based on the estimated execution time and latency of the edge-cloud for the task: If t i <T i,dl or T i,dl <d, then T i is admissible; otherwise, T i is blocked, where T i,dl represents the deadline requirement of the IoT application side; The task scheduler receives tasks from the task evaluator and schedules them to the edge data center or cloud for execution based on the scheduling policy and user service quality requirements. When a task is suitable for execution on the cloud, the task scheduler sends the task to the cloud data center for execution. When a task is suitable for execution in the edge data center, the task scheduler sends the task to the resource manager. The resource manager receives task requests and allocates relevant resources to the tasks. At the same time, the resource manager monitors the resource status of the edge data center. The monitoring information includes the execution status of the task and the usage of resources. The resource manager regularly updates the monitoring information and feeds it back to the task evaluator to support the task evaluator's evaluation. The resource manager or cloud data center returns the scheduling results of task execution to the IoT application.

2. The edge cloud computing admission control and scheduling method based on Internet of Things applications according to claim 1 is characterized by: In the IoT application side, the scheduling time window is TS, and the IoT application task set in the current time window is T = {T1, T2, ..., T N }, the i-th IoT application task T i Described as a sextuple <T i,id ,T i,len ,T i,com ,T i,bw ,T i,sto ,T i,dl >, where T i,id Indicates the task number submitted by the IoT application, T i,len Indicates the task length, T i,com ,T i,bw ,T i,sto They represent the computing, bandwidth, and storage requirements of IoT application tasks for resources; the number of available edge resources in the resource manager is X×Y, and the set of available edge resources is R e ={R1,R2,…,R X×Y }, the edge resource R e The jth resource R j,e Described as a quad <R j,id ,R j,com ,R j,bw ,R j,sto >; The number of available cloud resources in the cloud data center is D, and the set of available cloud resources is R c ={R1,R2,…,R D }, the cloud resource R c The kth resource R k,c Described as a quad <R k,id ,R k,com ,R k,bw ,R k,sto >, where R *,id Indicates the resource number, R *,com ,R *,bw ,R *,sto Represents the computing, bandwidth, and storage capabilities of the resource respectively.

3. The edge cloud computing admission control and scheduling method based on Internet of Things applications according to claim 2 is characterized by: During the task scheduling process, the task scheduler needs to meet the following constraints: 1) The edge resources of the resource manager or the cloud resources of the cloud data center can only process one task at a time, but all resources can be executed in parallel; 2) All tasks need to be assigned to the edge resources of the resource manager or the cloud resources of the cloud data center for execution, and each task can only be executed on one edge resource or cloud resource.

4. The edge cloud computing admission control and scheduling method based on Internet of Things applications according to claim 3 is characterized by: The resource monitor in the resource manager is set with a delay model, which is used to calculate the average delay d of the edge system. The specific calculation method is as follows: Input the known task and resource status T i ,R j ,N,x,X,y,Y and the resource processing capacity of data center x h x ; Initialize the number of requests currently accessing edge data center x and the number of available resources f′ currently in edge data center x x =0,g′ x =0; If the number of requests received by edge data center x during TS is f x (t)≠0 or TS The number of requests that have not been served by the edge data center q x,y (t)≠0; for t x,y Normalize to get Then calculate f′x=f′x+f x,y ; g′ x =g′ x +g x,y ; Among them, t x,y For task T i The average round-trip time to resource y in edge data center x; otherwise f x,y =0;g x,y =0; Calculate the average delay d of the edge computing system 5. The edge cloud computing admission control and scheduling method based on Internet of Things applications according to claim 4 is characterized by: In the Resource Manager, apply task T i The execution time of edge resource j includes two parts: transmission time and computing time, namely t j,e =t j,com +t j,tran The transmission time t j,com and calculation time t j,tran The calculation formula is as follows: In cloud data centers, IoT application tasks T i Execution time t on cloud resources k,c Including calculation time t k,com , transmission time t k,tran and edge-to-cloud migration time t mig , t k,c =t k,com +t k,tran +t mig where t k,com and t k,tran The calculation formula is as follows:

6. The edge cloud computing admission control and scheduling method based on Internet of Things applications according to claim 5 is characterized by: The task scheduler specifically schedules tasks including: Input IoT application tasks and resource related parameters, task request number T i ,R j ,R k ,q x,y (t),g x,y ; Calculate the average resource requirement R of each scheduling cycle task avg (t); Among them, P max It is the maximum number of tasks that can be accommodated in each scheduling cycle. Its calculation formula is as follows P max ≥N; In each scheduling cycle, a threshold θ is set for edge resources. θ is dynamically adjustable and is set to queue q in each cycle. x,y (t) Number of allocated resources g x,y (t) time average expectation, that is: θ = g x,y ; Task scheduler based on R avg (t) and θ values to judge, if R avg (t)>θ, indicating that the current edge data center load is too large, then the task is scheduled to the cloud data center for execution; otherwise, the task is scheduled to the resource manager for execution. In terms of scheduling in the edge data center, the resource manager adopts the strategy of scheduling all unserved queues q in the edge data center. x,y (t) to sort the current task T i Dispatch to q x,y (t) Smallest data center.

7. An edge cloud computing admission control and scheduling system based on Internet of Things applications, running the edge cloud computing admission control and scheduling method based on Internet of Things applications according to any one of claims 1 to 6, characterized in that: include: An Internet of Things application terminal, a task evaluator, a task scheduler, a resource manager, and a cloud data center, wherein the Internet of Things application terminal is used to submit a task request to the task evaluator; The task evaluator is used to receive task requests and evaluate whether the task meets the delay requirements. If the task meets the delay requirements, the task is submitted to the task scheduler. If the task does not meet the delay requirements, the result of not meeting the delay requirements is returned to the IoT application end, and the IoT application end is notified to resubmit the task with a new deadline. The task scheduler is used to receive tasks from the task evaluator and schedule tasks to the edge data center or the cloud for execution based on the scheduling strategy and user service quality requirements. When the task is suitable for execution on the cloud, the task scheduler transmits the task to the cloud data center for execution. When the task is suitable for execution in the edge data center, the task scheduler transmits the task to the resource manager. The resource manager is used to receive task requests and allocate relevant resources to the task, and return the scheduling result of the task execution to the IoT application end. At the same time, the resource manager monitors the resource status of the edge data center. The monitoring information includes the execution status of the task and the usage of the resources, and regularly updates the monitoring information and feeds it back to the task evaluator as support for the evaluation of the task evaluator. The cloud data center is used to return the scheduling result of the task execution to the IoT application end.

Images