Cloud computing platform task scheduling system based on edge computing network

Abstract

The invention discloses a cloud computing platform task scheduling system based on an edge computing network, which belongs to the field of computer science and technology and comprises a user layer, an edge layer, a coordination layer and a cloud layer which are connected through electric signals. The user layer marks metadata, a CPU core number, a memory size and a software library set corresponding to a system task, and submits the metadata, the CPU core number, the memory size and the software library set to the coordination layer; the edge layer receives an edge task scheduling instruction of the coordination layer, adjusts and executes an edge node task queue, processes terminal equipment data, obtains edge node resources through an edge resource manager ERM and reports the edge node resources to the coordination layer; the coordination layer classifies system tasks, combines edge node resources and cloud load indexes, and generates edge and cloud task scheduling instructions through an intelligent scheduling algorithm; and the cloud layer obtains a cloud load index, reports the cloud load index to the coordination layer, receives a cloud task scheduling instruction, and adjusts and executes a cloud layer task queue. The system realizes efficient scheduling of tasks and optimal configuration of resources.

Description

Technical Field

[0001] The present invention belongs to the field of computer science and technology, and particularly relates to a task scheduling system for a cloud computing platform based on an edge computing network. Background Technique

[0002] With the rapid development of the Internet of Things, industrial Internet, and 5G technology, a large number of terminal devices are connected to the cloud computing platform, giving rise to an urgent need for real-time and low-latency data processing. The traditional cloud computing platform adopts a centralized architecture, uploading all tasks to the cloud for processing, resulting in high network transmission latency and large bandwidth occupancy, making it difficult to meet the requirements of real-time response scenarios. Edge computing realizes local data processing by deploying edge nodes at the network edge, effectively reducing latency. However, there are still technical bottlenecks in existing edge-cloud collaborative systems.

[0003] Existing edge-cloud collaborative systems lack refined classification of task characteristics and cannot reasonably allocate tasks to edge nodes or the cloud according to the specific requirements of tasks. The status of hardware and software resources of edge nodes cannot be synchronized to the scheduling center in real time, and there is a lack of a dynamic evaluation mechanism for cloud load indicators, resulting in a disconnection between task scheduling and actual resource capabilities, causing overloading or idling of edge node resources and unreasonable allocation of cloud resources. At the same time, traditional scheduling algorithms do not combine task metadata with edge-cloud resource status for joint optimization, making it difficult to achieve dynamic and balanced task allocation. Therefore, how to achieve accurate task classification, real-time perception of resource status, and cross-layer collaborative task scheduling to solve the problems of high latency, low resource utilization, and rigid scheduling strategies in existing systems has become an urgent problem to be solved. For this reason, a task scheduling system for a cloud computing platform based on an edge computing network is proposed. Summary of the Invention

[0004] The purpose of the present invention is to provide a task scheduling system for a cloud computing platform based on an edge computing network to solve the problems raised in the above background technique.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] A task scheduling system for a cloud computing platform based on an edge computing network includes a user layer, an edge layer, a coordination layer, and a cloud layer. The user layer, edge layer, coordination layer, and cloud layer are connected by electrical signals;

[0007] The edge layer receives the edge task scheduling instruction from the coordination layer, adjusts and executes the edge node task queue according to the edge task scheduling instruction, and obtains terminal device data through the edge node, performs edge computing processing on the terminal device data. At the same time, the hardware and software resources of the edge node are obtained in real time through the Edge Resource Manager (ERM) and reported to the coordination layer;

[0008] The coordination layer classifies the obtained system tasks to obtain edge-priority processing tasks and cloud processing tasks, and performs task scheduling on the hardware and software resources of edge nodes, the load metrics of the cloud, edge-priority processing tasks, and cloud processing tasks, thereby obtaining edge and cloud task scheduling instructions and sending them to the coordination layer and the cloud layer;

[0009] The cloud layer obtains the load metrics of the cloud and reports them to the coordination layer. At the same time, it receives the cloud task scheduling instructions from the coordination layer and adjusts and executes the cloud layer task queue according to the cloud task scheduling instructions;

[0010] The user layer labels the metadata, number of CPU cores, memory size, and software library set corresponding to the system tasks, and submits the system tasks and the corresponding metadata, number of CPU cores, memory size, and software library set of the system tasks to the coordination layer.

[0011] Preferably, in the edge layer, the process of receiving the task scheduling instructions from the coordination layer and adjusting and executing the edge node task queue:

[0012] The edge node obtains the edge task scheduling instructions issued by the coordination layer. The edge task scheduling instructions include execution node information, edge task priority, and task hardware and software resource requirements. The execution node information is used to determine the edge node that executes the task. The edge task priority is used to determine the order of task execution. Tasks with higher task priorities will be processed first. The task hardware and software resource requirements are used to specify the resource conditions required for task operation. The Edge Resource Manager ERM responsible for edge node resource management and coordination monitors the resource status of the edge node in real time according to the task hardware and software resource requirements in the edge task scheduling instructions, allocates hardware and software resources to the edge node uniquely determined by the execution node information. At the same time, according to the edge task priority, through the priority queue algorithm, the execution order of tasks is arranged according to the task priority of the tasks, thereby adjusting the edge node task queue of the edge node. The edge node task queue is a key data structure for the edge node to manage the task execution order and resource allocation;

[0013] According to the edge node task queue, for the edge nodes that have been allocated hardware and software resources, the edge computing framework deployed on the edge device is used to call the corresponding software library to perform edge computing processing on the obtained terminal device data. The edge computing framework can call the appropriate software library according to the task requirements to complete the edge computing processing.

[0014] Preferably, in the edge layer, the process of obtaining terminal device data through the edge node and performing edge computing processing on the terminal device data:

[0015] Pre - deploy an edge computing framework, an embedded operating system, and a software library on the edge node. The embedded operating system is an operating system specifically designed for embedded devices, which can ensure the stable and efficient operation of the edge node. The software library provides specific implementation methods for data analysis, feature extraction, and rule judgment. The edge computing framework, the embedded operating system, and the software library cooperate with each other to jointly lay a foundation for the edge computing processing of terminal device data;

[0016] The edge node collects terminal device data from the terminal device, and for the obtained terminal device data, through the edge computing framework deployed on the edge node, it calls the corresponding software library to perform data analysis, feature extraction, or rule judgment on the terminal device data, and obtains the terminal device data processed by edge computing. Data analysis, feature extraction, and rule judgment are edge computing methods for the edge node to perform edge computing on the obtained terminal device data locally. Data analysis is to process terminal device data using statistical methods or data mining algorithms. Feature extraction extracts data features from terminal device data. Rule judgment is to judge terminal device data according to pre - set rules.

[0017] Preferably, in the edge layer, the process of the Edge Resource Manager (ERM) obtaining the hardware and software resources of the edge node in real - time and reporting them to the coordination layer:

[0018] The Edge Resource Manager (ERM) obtains the hardware resources of the edge node, including the number of CPU cores , the memory size and network bandwidth by interacting with the embedded operating system deployed on the edge node. At the same time, the Edge Resource Manager (ERM) obtains the software library set of the edge node , that is, the software resources of the edge node, through the task manager provided by the embedded operating system deployed on the edge node. The task manager can list all software libraries and programs installed in the current embedded operating system;

[0019] The Edge Resource Manager (ERM) sends the obtained hardware resource and software resource information of the edge node to the coordination layer in the JSON format of lightweight data exchange.

[0020] Preferably, in the coordination layer, the process of classifying the obtained system tasks to obtain edge - priority processing tasks and cloud - processing tasks:

[0021] The coordination layer generates a system task category label through the task metadata rule engine for the received system tasks and the metadata corresponding to the system tasks , and classifies the system tasks according to the system task category label .

[0022] If the system task category label is 1, it indicates that the system task is an edge - priority task;

[0023] If the system task category label is 0, it indicates that the system task is a cloud - processing task;

[0024] The task metadata rule engine is as follows:

[0025] ;

[0026] Among them, is the system task category label, is the real - time index of the system task, is the data volume of the system task, is the computational complexity level of the system task, is the network dependence of the system task.

[0027] Preferably, in the coordination layer, the process of task scheduling for the hardware and software resources of edge nodes, the load metrics of the cloud, edge - priority processing tasks, and cloud - processing tasks, and obtaining edge and cloud task scheduling instructions is as follows:

[0028] According to the hardware resources and software resources of edge nodes, the load metrics of the cloud, edge - priority processing tasks, and cloud - processing tasks, respectively construct an edge - node set , a cloud - load - metric set , an edge - priority - processing - task set and a cloud - processing - task set , and perform task scheduling on the edge - node set, cloud - load - metric set, and task set through an intelligent scheduling algorithm to obtain edge and cloud task scheduling instructions;

[0029] The edge - node set , each edge node in the edge - node set includes the number of CPU cores , the memory size , the network bandwidth and the software - library set , the cloud - load - metric set , for the edge - node set , obtain the resource availability of each edge node in the edge - node set through the edge - node resource quantization formula, and for the cloud - load - metric set , obtain the cloud - resource pressure value through the cloud - resource evaluation formula;

[0030] The edge node resource quantization formula is as follows:

[0031] ;

[0032] Among them, , and are the total number of CPU cores, the total memory size, and the total network bandwidth of the edge node respectively. , and are the number of CPU cores required by the edge node , the memory size and the network bandwidth respectively. , and are weight coefficients, and , is the resource availability of the edge node .

[0033] The cloud resource evaluation formula is as follows:

[0034] ;

[0035] Among them, is the cloud resource pressure value, and are the cloud CPU utilization rate, memory usage rate, and task queue length respectively. , and are weight coefficients, and ;

[0036] Traverse the edge priority processing task set . Each task in the edge priority processing task set contains the number of CPU cores corresponding to the task , the memory size and the software library set . For each task , filter the edge node from the edge node set . belongs to and the system task, and the system task includes ;

[0037] If the number of CPU cores of the edge node , the memory size and software library set , satisfy , and , and , then add the edge node to the task corresponding edge node set ;

[0038] If the CPU core count of the edge node , memory size and software library set do not satisfy , and , and , then filter new edge nodes;

[0039] For the edge node set corresponding to the task , calculate the fitness of each edge node in the edge node set for the task using the fitness formula , select the edge node with the highest fitness to assign the task , and generate an edge task scheduling instruction including execution node information , edge task priority, and task hardware and software resource requirements;

[0040] The fitness formula is:

[0041] ;

[0042] Wherein, is the fitness of the task , is the system task priority corresponding to the task , is the resource availability of the edge node ;

[0043] If the edge node set corresponding to the task is an empty set, then mark the task as a task to be reallocated and put it into the pending queue;

[0044] Traverse the cloud processing task set , each task in the cloud processing task set includes the CPU core count corresponding to the task and memory size , for each task , by comparing with the set cloud resource pressure threshold for comparison belonging to and system tasks, and the system tasks include ;

[0045] If is less than the set cloud resource pressure threshold at this time, the cloud CPU core number and cloud memory size required for the task are obtained through the cloud CPU core number formula and the cloud memory formula the cloud CPU core number that the task should be allocated and the cloud memory size , and accordingly, a cloud task scheduling instruction corresponding to the task including the cloud CPU core number, the cloud memory size, and the cloud task priority is generated ;

[0046] The cloud CPU core number formula is:

[0047] ;

[0048] Among them, is the cloud CPU core number that the task should be allocated , is the CPU core number corresponding to the task , is the total sum of the CPU core numbers required for all tasks allocated to cloud processing is the cloud CPU utilization rate is the total number of cloud CPU cores;

[0049] The cloud memory formula is:

[0050] ;

[0051] Among them, is the cloud memory size that the task should be allocated , is the memory size corresponding to the task , is the total sum of the memory sizes required for all tasks allocated to cloud processing is the memory utilization rate is the total cloud memory size;

[0052] When is greater than or equal to the set cloud resource pressure threshold at this time, the task is placed in the pending queue;

[0053] Accordingly, the process of implementing the intelligent scheduling algorithm for task scheduling is realized;

[0054] Tasks in the queue to be processed and , according to the time interval , re - perform task scheduling through the intelligent scheduling algorithm until the queue to be processed is empty;

[0055] The queue to be processed is a buffer area for temporarily storing edge - priority processing tasks and cloud - processing tasks that cannot be allocated execution resources for the time being, to ensure that edge - priority processing tasks and cloud - processing tasks are not missed.

[0056] Preferably, in the cloud layer, obtain the load metrics of the cloud and report them to the coordination layer:

[0057] The cloud layer collects the load metrics of the cloud with the help of monitoring tools provided by cloud service providers. At the same time, the cloud layer sends the load metric data of the cloud to the coordination layer in the JSON format of lightweight data exchange;

[0058] The load metrics of the cloud include CPU utilization , memory usage and the length of the task queue .

[0059] Preferably, in the cloud layer, the process of accepting the cloud task scheduling instruction from the coordination layer and adjusting and executing the cloud layer task queue:

[0060] The cloud layer obtains the cloud task scheduling instruction issued by the coordination layer. The cloud task scheduling instruction includes the number of cloud CPU cores, the size of cloud memory, and the priority of cloud tasks. The cloud layer parses the cloud task scheduling instruction to obtain the number of cloud CPU cores, the size of cloud memory, and the priority of cloud tasks. The cloud layer adjusts the cloud task queue through the dynamic priority scheduling algorithm according to the number of cloud CPU cores, the size of cloud memory, and the priority of cloud tasks. The dynamic priority scheduling algorithm comprehensively considers the real - time state of tasks and the system resource situation, and dynamically changes the task priority. Based on the adjusted cloud layer task queue, cloud computing is performed. Cloud computing is a computing model based on the Internet. The cloud layer can utilize a large number of computing resources provided by cloud service providers to provide users with powerful computing capabilities. The cloud layer task queue is a key data structure for cloud management of task execution order and resource allocation.

[0061] Preferably, in the user layer, the process of annotating the metadata, CPU core number, memory size, and software library set corresponding to the system task and submitting the system task and the corresponding metadata, CPU core number, memory size, and software library set to the coordination layer:

[0062] According to the specific situation of the system tasks, the user manually annotates the metadata corresponding to the system tasks, the number of CPU cores , the memory size and the software library set , and sends the system tasks and the metadata corresponding to the system tasks, the number of CPU cores , the memory size and the software library set to the coordination layer in JSON format;

[0063] The metadata corresponding to the system tasks includes real-time metrics , data volume , computational complexity level , network dependency and system task priority . The real-time metric represents the sensitivity of the task to latency, and the data volume represents the size of the data scale required to process the task. The computational complexity level reflects the computing resources and difficulty required for task execution. The range is from level 1 to level 5. Level 1 represents the simplest calculation, and level 5 represents the most complex calculation. The network dependency represents whether the task depends on cloud resources for data interaction, which is represented by 0 or 1. 0 means only local data is required, and 1 means data interaction with the cloud is required. The system task priority is set by the user according to the importance and urgency of the task. The range is from level 1 to level 5. Level 1 is the lowest priority, and level 5 is the highest priority.

[0064] Due to the adoption of the above technical solution, the technical progress achieved by the present invention compared with the prior art is:

[0065] 1. The present invention realizes precise classification and hierarchical collaborative scheduling of tasks. Through the metadata of the system tasks annotated by the user layer and using the task metadata rule engine of the coordination layer, automatic classification of system tasks is realized, distinguishing between edge-priority processing tasks and cloud-processing tasks. Compared with the rough task allocation mode in the prior art, the present invention can preferentially deploy system tasks with high real-time requirements and lightweight calculations to edge nodes, reduce data transmission latency, and allocate computationally intensive and large-data-volume system tasks to the cloud, giving full play to the edge-cloud collaborative advantage and improving the overall processing efficiency of the system.

[0066] 2. The present invention realizes real-time perception and dynamic matching of resource status. The edge layer collects the hardware resources and software resources of edge nodes in real time through the Edge Resource Manager (ERM) and reports them to the coordination layer in JSON format; the cloud layer obtains load metrics through cloud service monitoring tools. The coordination layer dynamically calculates the resource availability and cloud pressure value based on the edge node resource quantization formula and the cloud resource evaluation formula, realizes the accurate matching of task requirements and resource status, avoids the idle or overloaded resources of edge nodes, and solves the problem that the resource status lags behind the scheduling decision in the prior art.

[0067] 3. The intelligent scheduling algorithm of the present invention improves the cross-layer resource utilization rate. The coordination layer, through the intelligent scheduling algorithm, selects the optimal edge node for edge-priority tasks according to the adaptation degree formula to ensure the rapid execution of high-priority tasks. At the same time, for cloud processing tasks, it dynamically allocates the number of CPU cores and memory resources according to the cloud resource pressure threshold, and adjusts the cloud task queue in combination with the dynamic priority scheduling algorithm, effectively balancing the edge-cloud load. Compared with the traditional fixed strategy scheduling, it can reduce the length of the cloud task queue and improve the resource utilization rate of edge nodes. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0069] Figure 1 It is a schematic diagram of the system function module process of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0070] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0071] Embodiment, as Figure 1 described, a task scheduling system for a cloud computing platform based on an edge computing network includes a user layer, an edge layer, a coordination layer, and a cloud layer. The user layer, the edge layer, the coordination layer, and the cloud layer are connected by electrical signals;

[0072] The edge layer is used to receive the edge task scheduling instructions from the coordination layer, adjust and execute the edge node task queue according to the edge task scheduling instructions, obtain terminal device data through the edge nodes, perform edge computing processing on the terminal device data, and at the same time, obtain the hardware and software resources of the edge nodes in real time through the Edge Resource Manager (ERM) and report them to the coordination layer;

[0073] The coordination layer is used to classify the obtained system tasks to obtain edge-priority processing tasks and cloud processing tasks, perform task scheduling processing on the hardware and software resources of the edge nodes, the load metrics of the cloud, the edge-priority processing tasks and the cloud processing tasks, obtain edge and cloud task scheduling instructions accordingly, and send them to the coordination layer and the cloud layer;

[0074] The cloud layer is used to obtain the load metrics of the cloud and report them to the coordination layer. At the same time, it receives the cloud task scheduling instructions from the coordination layer and adjusts and executes the cloud layer task queue according to the cloud task scheduling instructions;

[0075] The user layer is used to label the metadata, CPU core count, memory size, and software library set corresponding to the system tasks, and submit the system tasks and the corresponding metadata, CPU core count, memory size, and software library set to the coordination layer.

[0076] Further, the working principle of the present invention is illustrated by the following embodiments:

[0077] Taking the intelligent factory production management system as an example, this system includes a large number of sensors, intelligent devices, and a cloud server responsible for data analysis and decision-making. This system needs to process equipment status monitoring, production process optimization, and quality inspection, and has high requirements for the real-time performance and accuracy of task scheduling.

[0078] In the intelligent factory, various system tasks are continuously generated. According to the specific situation of the tasks, the user manually labels the metadata corresponding to the system tasks in the user layer. For example, for the equipment status monitoring task, its real-time index is set to 50 ms, the data volume is small, about 10 KB, the computing complexity level is level 1, the network dependence is 0, only local data is required, the system task priority is set to level 3. At the same time, the CPU core count, memory size, and software library set required for this task are labeled. After completion of the labeling, the system task and the corresponding metadata are submitted to the coordination layer in JSON format.

[0079] After receiving the tasks and metadata, the coordination layer generates system task category labels through the task metadata rule engine. Since the real-time index of the equipment status monitoring task , according to the task metadata rule engine, whose category label is 1, is determined to be an edge-priority processing class task, and the coordination layer constructs a set of edge nodes HE1 cloud load metric set , calculate the resource availability of each edge node in the edge node set through the edge node resource quantification formula , assuming the edge node has 4 CPU cores , the memory size is 8GB, the network bandwidth is 100Mbps. When calculating the resource availability, set the weight coefficient to 0.5, to 0.3, to 0.2. The total number of CPU cores of the edge node is 10 cores, the total memory size is 20GB, and the total network bandwidth is 500Mbps, then is 0.4. For the device status monitoring tasks in the edge-priority processing class task set, screen the edge nodes that meet the hardware resource requirements from the edge node set, that is, the CPU cores, memory size, and software library set of the edge node need to meet the task requirements. If the edge node meets the CPU core and memory size requirements of the task and has the software library required by the task, add it to the edge node set corresponding to the task , calculate the fitness through the fitness formula. Assume the system task priority of the task is 3, is 0.4, then is 1.2. Compare the fitness of all suitable edge nodes, select the edge node with the highest fitness to assign the task, generate an edge task scheduling instruction, including execution node information, edge task priority, and task hardware and software resource requirements, etc., and send it to the edge layer.

[0080] After the edge nodes in the edge layer receive the edge task scheduling instruction, the edge resource manager ERM allocates hardware and software resources to the edge nodes according to the task hardware and software resource requirements in the instruction. According to the edge task priority, adjust the edge node task queue through the priority queue algorithm. The edge nodes, according to the adjusted task queue, call the corresponding software library through the edge computing framework deployed on the edge device, perform edge computing processing on the device status data collected from the terminal device, perform data analysis on the collected device temperature data, and judge whether it exceeds the preset threshold. If it exceeds, alarm processing is performed.

[0081] For some production data analysis tasks, the data volume is large and the computational complexity is high. The task metadata rule engine determines that they are cloud processing tasks. The cloud layer uses the monitoring tools provided by the cloud service provider to collect cloud load indicators, such as CPU utilization. , memory usage and the task queue length , and report it to the coordination layer. The coordination layer calculates the cloud resource pressure value according to the cloud resource evaluation formula , assuming the weight system is 0.4, is 0.3, is 0.3. If the current cloud CPU utilization is 60%, the memory utilization is 50%, and the task queue length is 20, we can calculate 7.9, setting the cloud resource pressure threshold is 8, due to , calculate the number of cloud CPU cores and cloud memory size to which the task should be assigned through the cloud CPU core number formula and the cloud memory formula, generate cloud task scheduling instructions, and send them to the cloud layer. After the cloud layer receives the instructions, it adjusts the cloud task queue through the dynamic priority scheduling algorithm according to the number of cloud CPU cores, cloud memory size and cloud task priority in the instructions, and then performs cloud computing processing, such as in-depth analysis of production data to explore potential problems and optimization points in the production process.

[0082] The edge resource manager ERM of the edge layer obtains the hardware and software resources of the edge nodes in real time and reports them to the coordination layer. The cloud layer also continuously reports the cloud load indicators. The coordination layer regularly re-evaluates the resource status of the edge nodes and the cloud based on these real-time information, and recalculates the resource availability and resource pressure values. If it is found that the resources of an edge node are tight, resulting in a delay in task execution, the coordination layer will reschedule the tasks and transfer some tasks to other edge nodes or clouds with sufficient resources for processing. For tasks in the queue to be processed, the time interval is calculated. , set to 1 minute, and reschedule tasks using the intelligent scheduling algorithm until the pending queue is empty.

[0083] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A task scheduling system for a cloud computing platform based on an edge computing network, comprising: A user layer, an edge layer, a coordination layer, and a cloud layer. The user layer, edge layer, coordination layer, and cloud layer are connected by electrical signals. It is characterized in that The edge layer receives the edge task scheduling instruction of the coordination layer, adjusts and executes the edge node task queue according to the edge task scheduling instruction, and obtains terminal device data through the edge node, performs edge computing processing on the terminal device data. At the same time, the hardware and software resources of the edge node are obtained in real time through the Edge Resource Manager (ERM) and reported to the coordination layer; The coordination layer classifies the obtained system tasks to obtain edge-priority processing tasks and cloud processing tasks, performs task scheduling processing on the hardware and software resources of the edge node, the load metrics of the cloud, the edge-priority processing tasks, and the cloud processing tasks, thereby obtaining edge and cloud task scheduling instructions and sending them to the edge layer and the cloud layer; The cloud layer obtains the load metrics of the cloud and reports them to the coordination layer. At the same time, it receives the cloud task scheduling instruction of the coordination layer and adjusts and executes the cloud layer task queue according to the cloud task scheduling instruction; The user layer annotates the metadata, number of CPU cores, memory size, and software library set corresponding to the system task, and submits the system task and the metadata, number of CPU cores, memory size, and software library set corresponding to the system task to the coordination layer.

2. The cloud computing platform task scheduling system based on an edge computing network according to claim 1, characterized in that In the edge layer, the process of receiving the task scheduling instruction of the coordination layer and adjusting and executing the edge node task queue: The edge node obtains the edge task scheduling instruction sent by the coordination layer. The edge task scheduling instruction includes execution node information, edge task priority, and task hardware and software resource requirements. The Edge Resource Manager (ERM) allocates hardware and software resources to the edge node uniquely determined by the execution node information according to the task hardware and software resource requirements in the edge task scheduling instruction. At the same time, according to the edge task priority, the edge node task queue of the edge node is adjusted through the priority queue algorithm; According to the edge node task queue, for the edge node with allocated hardware and software resources, the corresponding software library is called through the edge computing framework deployed on the edge device to perform edge computing processing on the obtained terminal device data; The Edge Resource Manager (ERM) is a tool responsible for resource management and coordination of edge nodes; The priority queue algorithm is a special queue data structure algorithm.

3. The cloud computing platform task scheduling system based on an edge computing network according to claim 2, characterized in that, In the edge layer, the process of obtaining terminal device data through the edge node and performing edge computing processing on the terminal device data: An edge computing framework, an embedded operating system, and a software library are pre-deployed on the edge node. The edge node collects terminal device data from the terminal device, and the obtained terminal device data is used to call the corresponding software library through the edge computing framework deployed on the edge node to perform data analysis, feature extraction, or rule judgment on the terminal device data, and obtain the terminal device data after edge computing processing; The data analysis, feature extraction, and rule judgment are an edge computing method for the edge node to perform edge computing on the obtained terminal device data locally; The edge computing framework, embedded operating system, and software library are important components in edge computing for supporting task processing and application operation.

4. The cloud computing platform task scheduling system based on an edge computing network according to claim 3, characterized in that, In the edge layer, the process of the Edge Resource Manager (ERM) obtaining the hardware and software resources of the edge node in real time and reporting them to the coordination layer: The Edge Resource Manager (ERM) obtains the hardware resources of the edge node, including the number of CPU cores , memory size and network bandwidth , by interacting with the embedded operating system deployed on the edge node. At the same time, the ERM obtains the software library set of the edge node , that is, the software resources of the edge node; The Edge Resource Manager (ERM) sends the obtained hardware resource and software resource information of the edge node to the coordination layer in JSON format; The JSON format is a lightweight data exchange format; The task manager is an important tool in the operating system for managing processes, applications, and system performance.

5. The cloud computing platform task scheduling system based on an edge computing network according to claim 4, characterized in that, In the coordination layer, the process of classifying the obtained system tasks to obtain edge-priority processing tasks and cloud processing tasks: The coordination layer generates system task category labels for the received system tasks and the metadata corresponding to the system tasks through the task metadata rule engine, and classifies the system tasks according to the system task category labels; If the system task category label is 1, it indicates that the system task is an edge-priority task; If the system task category label is 0, it indicates that the system task is a cloud processing task; The task metadata rule engine is a mathematical expression formula of classification rules.

6. The cloud computing platform task scheduling system based on an edge computing network according to claim 5, characterized in that, In the coordination layer, the process of performing task scheduling on the hardware and software resources of the edge node, the load metrics of the cloud, edge-priority processing tasks, and cloud processing tasks, and obtaining edge and cloud task scheduling instructions accordingly: Construct edge node sets, cloud load metric sets, edge-preferred processing task sets, and cloud processing task sets respectively according to the hardware and software resources of edge nodes, the load metrics of the cloud, edge-preferred processing tasks, and cloud processing tasks. The cloud load metric sets The edge-preferred processing task sets And the cloud processing task sets Perform task scheduling on the edge node sets, cloud load metric sets, and task sets through an intelligent scheduling algorithm to obtain edge and cloud task scheduling instructions. The process of performing task scheduling through the intelligent scheduling algorithm is as follows: For the set of edge nodes , the resource availability of each edge node in the set of edge nodes is obtained through the edge node resource quantization formula . For the set of cloud load metrics , the cloud resource pressure value is obtained through the cloud resource evaluation formula ; Traverse the edge-first processing task set For each task , , filter edge nodes from the edge node set ; If the number of CPU cores of the edge node , the memory size and the software library set meet , and , and , then add the edge node to the task corresponding edge node set ; If the number of CPU cores of the edge node , the memory size and the software library set do not meet , and , and , then filter new edge nodes; For the task The corresponding set of edge nodes Calculate the set of edge nodes through the fitness formula For each edge node in The fitness of the task , select the edge node with the highest fitness to assign the task , generate an edge task scheduling instruction including execution node information , edge task priority, and task hardware and software resource requirements; If the task corresponding set of edge nodes is an empty set, then mark the task as a task to be reallocated and put it into the processing queue; Traverse the cloud processing task set For each task , , compare with the set cloud resource pressure threshold ; If less than the set cloud resource pressure threshold then, obtain the number of cloud CPU cores and the cloud memory size that the task should be allocated through the cloud CPU core number formula and the cloud memory formula, and generate a task corresponding cloud task scheduling instruction that includes the number of cloud CPU cores, the cloud memory size, and the cloud task priority; When is greater than or equal to the set cloud resource pressure threshold the task is placed in the queue to be processed; Accordingly, the process of implementing task scheduling by the intelligent scheduling algorithm is achieved; Tasks in the processing queue and are rescheduled according to the time interval by the intelligent scheduling algorithm until the processing queue is empty.

7. The cloud computing platform task scheduling system based on an edge computing network according to claim 6, characterized in that, In the cloud layer, obtaining the load metrics of the cloud and reporting them to the coordination layer: The cloud layer collects the load metrics of the cloud with the help of the monitoring tools provided by the cloud service provider. At the same time, the cloud layer sends the load metric data of the cloud to the coordination layer in JSON format; The load metrics of the cloud include CPU utilization , memory usage , and task queue length .

8. The cloud computing platform task scheduling system based on an edge computing network according to claim 7, characterized in that, In the cloud layer, the process of receiving the cloud task scheduling instructions from the coordination layer and adjusting and executing the cloud layer task queue according to the cloud task scheduling instructions: The cloud layer obtains the cloud task scheduling instructions issued by the coordination layer. The cloud task scheduling instructions include the number of cloud CPU cores, the cloud memory size, and the cloud task priority. The cloud layer adjusts the cloud task queue through the dynamic priority scheduling algorithm according to the number of cloud CPU cores, the cloud memory size, and the cloud task priority. The cloud layer performs cloud computing based on the adjusted cloud layer task queue; The dynamic priority scheduling algorithm is an algorithm that dynamically adjusts the task priority according to the real-time state of the task and the system resource situation during the task scheduling process; Cloud computing is a computing model based on the Internet.

9. The cloud computing platform task scheduling system based on an edge computing network according to claim 8, characterized in that, In the user layer, the process of annotating the metadata, CPU core number, memory size, and software library set corresponding to the system task, and submitting the system task and the metadata, CPU core number, memory size, and software library set corresponding to the system task to the coordination layer: The user manually annotates the metadata, number of CPU cores corresponding to the system task, memory size and software library set according to the specific situation of the system task, and sends the system task, the metadata, number of CPU cores corresponding to the system task, memory size and software library set to the coordination layer in JSON format; The metadata corresponding to the system task includes a real-time index , data volume , computational complexity level , network dependence and system task priority .

Images