Calculation power arrangement method, device and equipment for edge calculation, medium and program product

Abstract

The embodiment of the invention discloses a computing power arrangement method, device and equipment for edge computing, a medium and a program product. Comprising the following steps: acquiring to-be-processed service flow information of each to-be-processed service flow; mapping each to-be-processed service flow into a corresponding real-time service sub-queue and a corresponding non-real-time service sub-queue according to service flow characteristics in each piece of to-be-processed service flow information; determining a real-time computing power resource reservation amount according to the to-be-processed data amount of each real-time service sub-queue and the computing power reservation error correction coefficient, and determining a real-time computing power arrangement scheme according to the real-time computing power resource reservation amount; determining a non-real-time computing power arrangement scheme according to the to-be-processed data volume, the current residual computing power resource and the real-time computing power resource reservation volume of each non-real-time service sub-queue; wherein the computing power reservation error correction coefficient is determined according to the previous actual computing power resource occupation amount and the previous real-time computing power resource reservation amount in the previous service period corresponding to the current service period. And the computing power resource utilization rate of the edge computing equipment is improved.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things technology, and in particular to a computing power orchestration method, device, equipment, medium and program product for edge computing. Background Art

[0002] The Internet of Things is an expanded application and network extension of communication networks and the Internet. It uses sensing technology and intelligent equipment to perceive and identify the physical world, and performs calculations, processing, and knowledge mining through network transmission and interconnection. It realizes information interaction and seamless connection between people and objects, and between objects, and achieves the purpose of controlling, accurately managing, and making scientific decisions about the physical world. It has been widely used in the power sector to form the power Internet of Things.

[0003] As a new computing mode, edge computing can enable data to be processed promptly and effectively at the edge network near the source. With the rapid development of the Internet of Things, the explosive growth of data volume far exceeds the capacity of network bandwidth, and the emergence of a large number of new intelligent applications has put forward higher requirements for system latency performance. In the edge computing mode, the terminal offloads tasks to nearby edge servers for execution, effectively alleviating the network congestion and high latency problems caused by transmitting all data to the data center in traditional cloud computing.

[0004] However, the computing, communication and storage resources of edge computing devices are limited. In order to achieve efficient and flexible business processing and decision-making, the terminals, data and services under the new business ecology of the power Internet of Things are constantly increasing. Massive power data needs to be analyzed and calculated on edge computing devices, which will make the resources of edge computing devices tend to be saturated, resulting in a serious decline in service quality. At the same time, due to the dynamic characteristics of the new power system business, the edge computing resource scheduling method has also changed from a fixed configuration mode based on a certain business object to a flexible configuration mode based on a dynamic business object. Therefore, the contradiction between the growing business processing needs and limited computing resources has become increasingly prominent, reducing the business processing capabilities of the power Internet of Things network. Summary of the invention

[0005] The present invention provides a computing power orchestration method, device, equipment, medium and program product for edge computing. According to the different types of business flows, computing power orchestration plans are formulated for real-time business and non-real-time business in different computing power orchestration methods, so that for both real-time business and non-real-time business, dynamically changing business needs and actual business processing status can be fully considered, and a flexible and efficient computing power orchestration plan suitable for edge computing devices is obtained, thereby improving the computing power resource utilization rate of edge computing devices.

[0006] In a first aspect, an embodiment of the present invention provides a computing power arrangement method for edge computing, including:

[0007] Obtaining the pending business flow information of each pending business flow in the current business cycle;

[0008] According to the service flow characteristics in the information of each service flow to be processed, each service flow to be processed is mapped to a corresponding real-time service sub-queue and a non-real-time service sub-queue;

[0009] Determine the real-time computing resource reservation amount according to the amount of data to be processed of each real-time service sub-queue and the computing power reservation error correction coefficient, and determine the real-time computing power scheduling plan according to the real-time computing power resource reservation amount;

[0010] Determine the non-real-time computing power scheduling plan based on the amount of data to be processed in each non-real-time business sub-queue, the current remaining computing power resources, and the reserved amount of real-time computing power resources;

[0011] Among them, the computing power reservation error correction coefficient is determined based on the previous actual computing power resource occupancy and the previous real-time computing power resource reservation in the previous business cycle corresponding to the current business cycle.

[0012] In a second aspect, an embodiment of the present invention further provides a computing power orchestration device for edge computing, including:

[0013] An information acquisition module is used to acquire the pending business flow information of each pending business flow in the current business cycle;

[0014] A queue mapping module, used for mapping each to-be-processed service flow into a corresponding real-time service sub-queue and a non-real-time service sub-queue according to the service flow characteristics in the information of each to-be-processed service flow;

[0015] The first scheme determination module is used to determine the real-time computing power resource reservation amount according to the amount of data to be processed of each real-time service sub-queue and the computing power reservation error correction coefficient, and determine the real-time computing power scheduling scheme according to the real-time computing power resource reservation amount;

[0016] The second scheme determination module is used to determine the non-real-time computing power scheduling scheme according to the amount of data to be processed of each non-real-time service sub-queue, the current remaining computing power resources and the reserved amount of real-time computing power resources;

[0017] Among them, the computing power reservation error correction coefficient is determined based on the previous actual computing power resource occupancy and the previous real-time computing power resource reservation in the previous business cycle corresponding to the current business cycle.

[0018] In a third aspect, an embodiment of the present invention further provides a computing power orchestration device for edge computing, the computing power orchestration device for edge computing comprising:

[0019] at least one processor; and a memory communicatively coupled to the at least one processor;

[0020] Among them, the memory stores a computer program that can be executed by at least one processor, and the computer program is executed by at least one processor so that the at least one processor can implement the computing power orchestration method for edge computing of any embodiment of the present invention.

[0021] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to execute the computing power orchestration method for edge computing of any embodiment of the present invention.

[0022] In a fifth aspect, an embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, is used to execute the computing power orchestration method for edge computing of any embodiment of the present invention.

[0023] The embodiments of the present invention provide a computing power orchestration method, device, equipment, medium and program product for edge computing, which obtains the pending business flow information of each pending business flow in the current business cycle; maps each pending business flow to a corresponding real-time business sub-queue and a non-real-time business sub-queue according to the business flow characteristics in the pending business flow information; determines the real-time computing power resource reservation amount according to the amount of pending data of each real-time business sub-queue and the computing power reservation error correction coefficient, and determines the real-time computing power orchestration plan according to the real-time computing power resource reservation amount; determines the non-real-time computing power orchestration plan according to the amount of pending data of each non-real-time business sub-queue, the current remaining computing power resources and the real-time computing power resource reservation amount; wherein the computing power reservation error correction coefficient is determined based on the previous actual computing power resource occupancy and the previous real-time computing power resource reservation amount in the previous business cycle corresponding to the current business cycle. By adopting the above technical solution, when it is necessary to provide corresponding computing resources for each pending business flow in the current business cycle, firstly, the characteristics of different pending business flows are used to distinguish whether they belong to real-time business and map the business queues, and different computing power orchestration methods are used to determine the computing power orchestration plan for real-time business and non-real-time business. Among them, in order to ensure that real-time business can have sufficient computing power resources for execution, when determining the computing power orchestration plan, the real-time computing power orchestration plan will be determined first, so as to give priority to completing the reservation of the real-time computing power resources required for real-time business in the remaining computing power resources in the edge computing device according to the determined real-time computing power orchestration plan; and for non-real-time business, it will be determined based on the currently remaining available computing power resources of the edge computing device and the amount of real-time computing power reserved resources, so that the remaining computing power resources in the edge computing device can be fully utilized while ensuring that the real-time tasks can be executed in a timely manner. At the same time, the computing power reservation error correction coefficient determined according to the actual business execution situation in the previous business cycle is used to assist in the determination of the real-time computing power orchestration plan, so that the final real-time computing power orchestration plan can fully consider the dynamically changing business needs and the actual business processing status, and then the non-real-time computing power orchestration plan determined based on the real-time computing power resource reservation can also fully consider the dynamically changing business needs and the actual business processing status, and finally obtain a more flexible and efficient computing power orchestration plan suitable for edge computing devices, thereby improving the computing power resource utilization rate of edge computing devices.

[0024] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0026] Figure 1 A flowchart of a computing power arrangement method for edge computing provided in Embodiment 1 of the present invention;

[0027] Figure 2 A flowchart of a computing power arrangement method for edge computing provided in Embodiment 2 of the present invention;

[0028] Figure 3 A schematic diagram of the structure of a computing power orchestration device for edge computing provided in Embodiment 3 of the present invention;

[0029] Figure 4 A schematic diagram of the structure of an edge computing computing power orchestration device provided in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] Embodiment 1

[0033] Figure 1A flowchart of a computing power orchestration method for edge computing provided in the first embodiment of the present invention. The embodiment of the present invention can be applied to the situation of planning and allocating computing power resources for edge computing devices in the power Internet of Things. The method can be executed by a computing power orchestration device for edge computing. The computing power orchestration device for edge computing can be implemented by software and / or hardware. The computing power orchestration device for edge computing can be configured in the computing power orchestration device for edge computing. Optionally, the computing power orchestration device for edge computing can be an electronic device, which can be a notebook, a desktop computer, a smart tablet, an edge computing device in the power Internet of Things, etc. The embodiment of the present invention does not limit this.

[0034] like Figure 1 As shown, an edge computing computing power arrangement method provided by an embodiment of the present invention specifically includes the following steps:

[0035] S101. Obtaining the pending business flow information of each pending business flow in the current business cycle.

[0036] In this embodiment, the business cycle can be specifically understood as the time period required for a business to be started and completed, or one of the time periods when the complete time required for a business to be started and completed is divided into multiple time periods according to certain rules. The current business cycle can be specifically understood as the business cycle at the current moment. The business flow to be processed can be specifically understood as business data that is input to the edge computing device via the business interface and needs to be processed using the computing power resources in the edge computing device. The business flow information to be processed can be specifically understood as information used to characterize the relevant characteristics of the business flow to be processed.

[0037] Specifically, in order to fully utilize the computing power resources of the edge computing device during the edge computing process, the corresponding computing power resources are allocated to the business to be processed before the business flow to be executed is input into the edge computing device. Multiple pending business flows connected to the edge computing device through various business interfaces during the current business cycle can be counted. Each pending business flow can be generated by a power terminal device connected to the edge computing device, and then when the pending business flow is obtained, the pending business flow information used to characterize the relevant characteristics of the pending business flow is obtained.

[0038] S102: Map each to-be-processed service flow to a corresponding real-time service sub-queue and non-real-time service sub-queue according to service flow characteristics in the information of each to-be-processed service flow.

[0039] In this embodiment, the business flow characteristics can be specifically understood as specific or potential meaningful information used to characterize the business flow in a certain aspect. Optionally, for the business flow in the power Internet of Things, the corresponding business flow characteristics can be represented by multiple characteristic parameters. Exemplarily, the characteristic parameters may include the maximum packet, the minimum packet, the average packet size, the average arrival time, the mean interval time, the business flow size, the business flow duration and the protocol flag, etc. The embodiment of the present invention does not limit this. The real-time business subqueue can be specifically understood as a business queue containing real-time type business flows to be processed. The non-real-time business subqueue can be specifically understood as a business queue containing non-real-time type business flows to be processed. It can be understood that the multiple application software contained in the edge computing device can be divided into real-time application software and non-real-time application software according to the type of business that can be processed. Each application software can correspond to a business sub-queue, and the business flow mapped to the business sub-queue is the business that the application software needs to process; that is, the business sub-queue corresponding to the real-time application software is the real-time business sub-queue, and the business sub-queue corresponding to the non-real-time application software is the non-real-time business sub-queue; depending on the number of application software, the edge computing device can simultaneously contain multiple real-time business sub-queues and multiple non-real-time business sub-queues; since multiple to-be-processed business flows can be processed by the same application software in the edge computing device, multiple to-be-processed business flows can be mapped to the same real-time business sub-queue or non-real-time business sub-queue, and the order of each to-be-processed business flow in the corresponding business sub-queue is determined according to the mapping order.

[0040] Specifically, the business flow characteristics corresponding to each business flow to be processed are extracted from the logistics information of each business flow to be processed. Based on the business flow characteristics, it can be comprehensively determined whether the business flow to be processed is a business flow that needs to be processed in real time or a business flow that does not need to be processed in real time, and which software in the edge computing device should process the business flow to be processed. After clarifying the software to which the business flow to be processed should be assigned, the business flow to be processed can be mapped to the business sub-queue corresponding to the software, and the business sub-queue corresponding to the software that belongs to real-time processing can be determined as a real-time business sub-queue, and the business sub-queue corresponding to the software that belongs to non-real-time processing can be determined as a non-real-time business sub-queue.

[0041] S103: Determine the real-time computing power resource reservation amount according to the amount of data to be processed of each real-time service sub-queue and the computing power reservation error correction coefficient, and determine the real-time computing power scheduling plan according to the real-time computing power resource reservation amount.

[0042] Among them, the computing power reservation error correction coefficient is determined based on the previous actual computing power resource occupancy and the previous real-time computing power resource reservation in the previous business cycle corresponding to the current business cycle.

[0043] In this embodiment, the amount of data to be processed can be specifically understood as the amount of data that is located in the same business sub-queue and needs to be processed by the corresponding software in the edge computing device, that is, the amount of data that needs to be processed using the computing power resources in the edge computing device. The computing power reservation error correction coefficient can be specifically understood as a correction value determined based on the computing power occupancy during the real-time type business processing in the previous business cycle, which is used to adjust the amount of computing power resources required for the current cycle for the real-time type business estimate to be reserved according to the actual processing status. The amount of real-time computing power resources reserved can be specifically understood as the amount of computing power resources that the edge computing device needs to reserve for the pending business flows that need to be processed in real time in the current business cycle. The real-time computing power orchestration plan can be specifically understood as a plan for orchestrating and allocating the required computing power resources for the software corresponding to each real-time business sub-queue in the current business cycle.

[0044] In this embodiment, the previous business cycle can be specifically understood as the business cycle that is one business cycle before the current business cycle. It can be understood that the length of the previous business cycle may be consistent with the length of the current business cycle, or may be inconsistent, and the embodiment of the present invention does not limit this. The previous actual computing power resource occupancy can be specifically understood as the amount of computing power resources actually consumed by real-time type software or each real-time business sub-queue in the process of business processing during the previous business cycle. The previous real-time computing power resource reservation amount can be specifically understood as the real-time computing power resource reservation amount determined during the processing of the previous business cycle.

[0045] Specifically, the amount of data to be processed in each real-time service sub-queue is comprehensively counted and converted into the amount of computing power resources that the edge computing device needs to provide to process the above data volume, which is determined based on the amount of data contained in the business flow to be processed contained in each real-time service sub-queue. Then, the error correction of the above-determined amount of computing power resources is performed based on the computing power reservation error correction coefficient, and finally the real-time computing power resource reservation amount that the edge computing device needs to reserve for real-time type services in the current business cycle is obtained. Then, the real-time computing power resource reservation amount is divided based on the demand for computing power resources of each real-time service sub-queue, and the real-time computing power scheduling plan for the software corresponding to each real-time service sub-queue is obtained.

[0046] S104: Determine a non-real-time computing power scheduling plan based on the amount of data to be processed in each non-real-time service sub-queue, the current remaining computing power resources, and the reserved amount of real-time computing power resources.

[0047] In this embodiment, the current remaining computing power resources can be specifically understood as the amount of computing power resources remaining in the edge computing device when entering the current business cycle. The non-real-time computing power scheduling plan can be specifically understood as a plan for allocating the required computing power resources for the software corresponding to each non-real-time business sub-queue in the current business cycle.

[0048] Specifically, the amount of computing power resources that can be supplied to non-real-time services for processing in the current business cycle can be determined based on the current remaining computing power resources and the reserved amount of real-time computing power resources. Furthermore, the amount of computing power resources required for non-real-time business flows in the current business cycle can be determined based on the amount of data to be processed in each non-real-time business subqueue. Since the execution timeliness requirements for non-real-time services are not high, the amount of computing power resources required for non-real-time business flows may not be provided all at once. At this time, the processing of the computing power resource data required for non-real-time business flows in the current business cycle can be completed in batches based on the amount of computing power resources that can be supplied to non-real-time services for processing in the current business cycle, and the corresponding non-real-time computing power scheduling plan can be obtained.

[0049] The technical solution of this embodiment is to obtain the pending business flow information of each pending business flow in the current business cycle; according to the business flow characteristics in each pending business flow information, map each pending business flow to the corresponding real-time business sub-queue and non-real-time business sub-queue; determine the real-time computing power resource reservation amount according to the amount of pending data of each real-time business sub-queue and the computing power reservation error correction coefficient, and determine the real-time computing power scheduling plan according to the real-time computing power resource reservation amount; determine the non-real-time computing power scheduling plan according to the amount of pending data, the current remaining computing power resources and the real-time computing power resource reservation amount of each non-real-time business sub-queue; wherein, the computing power reservation error correction coefficient is determined according to the last actual computing power resource occupancy and the last real-time computing power resource reservation amount in the previous business cycle corresponding to the current business cycle. By adopting the above technical solution, when it is necessary to provide corresponding computing power resources for each pending business flow in the current business cycle, firstly, distinguish whether it belongs to the real-time business and map the business queue according to the characteristics of different pending business flows, and use different computing power scheduling methods to determine the computing power scheduling plan for real-time and non-real-time businesses. Among them, in order to ensure that real-time services can have sufficient computing resources for execution, when determining the computing power scheduling plan, the real-time computing power scheduling plan will be determined first, so as to give priority to completing the reservation of the real-time computing power resources required for real-time services in the remaining computing resources in the edge computing device according to the determined real-time computing power scheduling plan; and for non-real-time services, it will be determined based on the current remaining available computing power resources of the edge computing device and the amount of real-time computing power reservation resources, while ensuring that the real-time tasks can be executed in time, so that the remaining computing power resources in the edge computing device can be fully utilized. At the same time, the computing power reservation error correction coefficient determined according to the actual business execution situation in the previous business cycle is used to assist in the determination of the real-time computing power scheduling plan, so that the final real-time computing power scheduling plan can fully consider the dynamically changing business needs and the actual processing status of the business, and then the non-real-time computing power scheduling plan determined based on the real-time computing power resource reservation can also fully consider the dynamically changing business needs and the actual processing status of the business, and finally obtain a more flexible and efficient computing power scheduling plan suitable for edge computing devices, which improves the computing power resource utilization rate of edge computing devices.

[0050] Embodiment 2

[0051] Figure 2A flowchart of a computing power scheduling method for edge computing provided in the second embodiment of the present invention is further optimized on the basis of the above-mentioned optional technical solutions. After completing the mapping of each pending business flow to the corresponding business sub-queue, the pending data flow of each real-time business sub-queue is statistically converted to obtain the total real-time computing power demand, and the corresponding real-time computing power resource reservation is corrected by the computing power reservation error correction coefficient, and then the real-time computing power scheduling scheme is determined according to the proportion of the data volume corresponding to different real-time business sub-queues in the total real-time demand data volume. After the determination of the real-time computing power resource reservation, the non-real-time available computing power resources available for non-real-time services can be determined based on the current remaining computing power resources and the real-time computing power resource reservation. On the basis of ensuring that real-time services can have sufficient computing power resources to process them, according to the non-real-time delay threshold required for the execution of non-real-time services, combined with data aggregation processing, the non-real-time available computing power resources are used to the fullest extent, and the non-real-time computing power scheduling scheme is obtained. The current remaining computing power resources in the current business cycle can be used to the fullest extent. At the end of each business cycle, the computing power reservation error correction coefficient can be re-determined, and the determined computing power reservation error correction coefficient can be updated based on the latency when processing real-time type business within this cycle, so that the finally determined computing power reservation error correction coefficient can better reflect the dynamically changing business needs and actual business processing status, and thus make each computing power scheduling plan determined based on the computing power reservation error correction coefficient more suitable for the processing of edge computing devices, thereby improving the computing power resource utilization rate of edge computing devices.

[0052] like Figure 2 As shown, an edge computing computing power arrangement method provided by an embodiment of the present invention specifically includes the following steps:

[0053] S201. Obtain the pending business flow information of each pending business flow in the current business cycle.

[0054] S202. Input each pending business flow information into a pre-trained business type determination model, extract the business flow characteristics of each pending business logistics information through the business type determination model, and perform business classification operations based on each business flow characteristic to determine the business type corresponding to each pending business flow.

[0055] Among them, the business type determination model is a lightweight long short-term memory network (Naive Long Short-Term Memory, Naive-LSTM) model.

[0056] In this embodiment, the business type determination model can be specifically understood as a neural network model for classifying the business type corresponding to the business logistics information according to the business characteristics in the input business logistics information. Exemplarily, the business type may include a real-time business type and a non-real-time business type.

[0057] Specifically, the information of the business flow to be processed corresponding to each business flow to be processed is input into the pre-trained business type determination model, and the business type determination model first extracts features of the business flow information to be processed, and then comprehensively analyzes the business flow to be processed corresponding to the business flow information to be processed according to the extracted features, so as to classify the business flow to be processed based on whether it needs to be processed in real time, and classify each business flow to be processed into a real-time business type and a non-real-time business type. If the business flow to be processed is determined to be a real-time business type, step S203 is executed; if the business flow to be processed is determined to be a non-real-time business type, step S204 is executed.

[0058] Exemplarily, the business type determination model can be a Naive-LSTM model composed of an input layer, an LSTM layer and an output layer, wherein the input layer is used to input a variety of business flow features extracted from the business flow information to be processed, and the type of business flow features can be set to m types here; the LSTM layer can be composed of m*n Naive-LSTM units, that is, one LSTM layer has m types of feature values*n layers of Naive-LSTM units; the output layer is used to process the data output by the LSTM layer through a softmax classification function to obtain the business type corresponding to the business flow information to be processed.

[0059] Among them, since the LSTM layer simplifies the internal gate structure and internal parameters of the LSTM unit, the amount of computational data processing is reduced. The structural formula of Naive-LSTM is as follows:

[0060]

[0061] Among them, x k is the k-th input vector; y k is the k-th output vector; z k is the input signal; i k is the input gate; c k is the state unit; o k is the output gate; W z For z k The input weight matrix in U z , U i , U o They are z k ,i k , o k The recursive weight matrix in bz For z k The bias matrix; σ() is the sigmoid activation function; g() is the tanh activation function.

[0062] S203: Map the to-be-processed service flow whose service type is the real-time service type to the corresponding real-time service sub-queue, and execute S205.

[0063] Specifically, after completing the judgment of the business type corresponding to each business flow to be processed, if the business type of the business flow to be processed is a real-time business type, the software that can be used to process the business flow to be processed can be selected from the software that can process real-time business contained in the edge computing device, and the business flow to be processed can be written into the real-time business sub-queue corresponding to the software by mapping, and after all business flows to be processed have completed mapping, execute S205.

[0064] It can be understood that the pending business flows of each real-time business type will be mapped to the corresponding real-time business sub-queue according to the output order of the business type determination model, but since the pending business flows in each real-time business sub-queue of the same business cycle will be processed in the same batch, the mapping order has little effect on the processing of each pending business flow.

[0065] S204: Map the to-be-processed service flow whose service type is a non-real-time service type to a corresponding non-real-time service sub-queue, and execute S205.

[0066] Specifically, after completing the judgment of the business type corresponding to each business flow to be processed, if the business type of the business flow to be processed is a non-real-time business type, the software that can be used to process the business flow to be processed can be selected from the software that can process non-real-time businesses contained in the edge computing device, and the business flow to be processed can be written into the non-real-time business sub-queue corresponding to the software by mapping, and after all business flows to be processed have completed mapping, execute S205.

[0067] S205: Determine the total real-time required data volume according to the to-be-processed data volume of each real-time service sub-queue.

[0068] Specifically, for each real-time service sub-queue, the amount of data to be processed of the real-time service sub-queue can be determined based on its corresponding queue characteristic parameters such as queue length, queue capacity and queue processing speed, and then the amount of data to be processed of all real-time service sub-queues can be summed up to obtain the total amount of data required to be processed for real-time service type services in the current service cycle, and the total amount of data is determined as the total real-time demand data amount.

[0069] S206. Determine the total real-time computing power requirement as the product of the total real-time required data volume and the preset computing power conversion factor.

[0070] In this embodiment, the preset computing power conversion factor can be specifically understood as a parameter predetermined according to actual conditions for converting the amount of data corresponding to the required processing business flow and the computing power resources that the edge computing device can provide. The total real-time computing power demand can be specifically understood as the total amount of computing power resources that the edge computing device needs to provide for the pending business flow of the real-time business type in the current business cycle.

[0071] Specifically, in order to facilitate the determination of the amount of computing power resources provided by the edge computing device for the pending business flows of the real-time business type, after clarifying the total real-time demand data volume of the pending data flows of the real-time business type, it can be converted into a representation of the amount of computing power resources based on the pre-provided preset computing power conversion factor. At this time, the total real-time demand data volume can be multiplied by the preset quantity conversion factor, and the product is determined as the total real-time computing power demand.

[0072] S207. Determine the difference between the total real-time computing power demand and the computing power reservation error correction coefficient as the real-time computing power resource reservation amount.

[0073] Specifically, since the actual amount of computing power resources occupied will fluctuate due to various factors during the actual data processing process, in order to make full use of the computing power resources in the edge computing device, the total real-time computing power demand determined in the current business cycle can be adjusted by using the computing power reservation error correction coefficient determined based on the previous business cycle. In an embodiment of the present invention, the difference between the total real-time computing power demand and the computing power reservation error correction coefficient can be calculated, and the difference can be determined as the real-time computing power resource reservation amount that needs to be reserved and allocated for the real-time business type business flow in the current business cycle.

[0074] S208. Allocate the reserved amount of real-time computing power resources to the real-time business container corresponding to each real-time business sub-queue according to the proportion of the amount of data to be processed of each real-time business sub-queue in the total real-time demand data amount, and determine it as a real-time computing power scheduling plan.

[0075] Specifically, since edge computing devices usually have sufficient computing resources for real-time business types, the amount of real-time computing resources reserved should generally be less than or equal to the amount of computing resources available in the edge device during the current business cycle, that is, all pending business flows in each real-time business sub-queue can be allocated sufficient computing resources for processing. At this time, the amount of real-time computing resources reserved can be directly divided according to the proportion of the amount of data to be processed in each real-time business sub-queue in the total real-time demand data volume, and each allocated computing resource can be allocated to the real-time business container corresponding to each real-time business sub-queue to determine the real-time computing scheduling plan.

[0076] It can be understood that when the real-time computing power resource reservation amount is greater than the available computing power resources of the edge device in the current business cycle, the edge computing device will also give priority to the processing of pending business flows of real-time business types. At this time, the available computing power resources in the current business cycle can be used as the real-time computing power resource reservation amount, and the real-time computing power orchestration plan can be determined through the same division method as above.

[0077] S209: Determine the difference between the current remaining computing power resources and the reserved amount of real-time computing power resources as non-real-time available computing power resources.

[0078] Specifically, since edge computing devices need to give priority to real-time business processing during business processing, in the current business cycle, it is necessary to first reserve the computing resources required for real-time business, and then determine the remaining computing resources as computing resources available for executing non-real-time business, that is, the difference between the current remaining computing resources in the current business cycle and the reserved amount of real-time computing resources is calculated, and the difference is determined as non-real-time available computing resources.

[0079] S210. Determine the total non-real-time computing power requirement according to the amount of data to be processed of each non-real-time service sub-queue and a preset computing power conversion factor.

[0080] Specifically, for each non-real-time business sub-queue, the amount of data to be processed of the non-real-time business sub-queue can be determined based on the corresponding queue characteristic parameters such as queue length, queue capacity and queue processing speed, and then the amount of data to be processed of all non-real-time business sub-queues can be summed up to obtain the total amount of data required to be processed by non-real-time business type services in the current business cycle. In order to facilitate the determination of the amount of computing power resources that the edge computing device needs to provide for the non-real-time business type to be processed business flow, it can be converted into a representation of the amount of computing power resources based on a pre-provided preset computing power conversion factor, that is, the total amount of data required to be processed by the non-real-time business type services is multiplied by the preset computing power conversion factor, and the product is determined as the total non-real-time computing power demand.

[0081] S211. Whether the total non-real-time computing power demand is greater than the non-real-time available computing power resources, if so, execute S212; if not, execute S213.

[0082] Specifically, although non-real-time business type services do not require real-time execution, real-time execution is still the optimal execution method. Therefore, after clarifying the total non-real-time computing power demand, the total non-real-time computing power demand can be compared with the non-real-time available computing power resources. If the comparison result is that the total non-real-time computing power demand is less than or equal to the non-real-time available computing power resources, it can be considered that after the edge computing device provides computing power resources for real-time business type services, there are still enough computing power resources available for non-real-time business type services, which can also achieve real-time processing of non-real-time business type services, and S213 is executed at this time. If the comparison result is that the total non-real-time computing power demand is greater than the non-real-time available computing power resources, it can be considered that after the edge computing device provides computing power resources for real-time business type services, the remaining computing power resources are insufficient for real-time processing of non-real-time business type services, and S212 is executed at this time.

[0083] S212: Aggregate the non-real-time service sub-queues to generate an aggregated service block, and determine the non-real-time computing power scheduling plan according to the aggregated data volume of the aggregated service block, the non-real-time available computing power resources, and the preset non-real-time delay threshold.

[0084] In this embodiment, the aggregated service block can be specifically understood as a data block in which the data volume generated after aggregating the pending service flows in different non-real-time service sub-queues is the sum of the data volumes corresponding to the aggregated pending service flows. The aggregated data volume can be specifically understood as the data volume contained in the aggregate database. The preset non-real-time delay threshold can be specifically understood as a time threshold pre-set according to actual conditions to determine whether the non-real-time service type of service can be executed within the required time.

[0085] Specifically, when it is determined that the pending business flows corresponding to the non-real-time business sub-queues cannot be directly processed using the non-real-time available computing power resources, the pending business flows in the non-real-time business sub-queues to be processed can be aggregated to obtain aggregated business blocks, so that the pending business flows of each non-real-time business type can be allocated and processed as a whole. Then, based on the amount of aggregated data of the aggregated business block, the non-real-time available computing power resources in the current business cycle, and the preset non-real-time delay threshold, it can be determined whether the unified execution of the aggregated business block through the non-real-time available computing power resources can meet the execution time requirements of the non-real-time business type. If it cannot be met, the aggregated business block can be split or otherwise adjusted so that the execution time of a non-real-time business type can meet the requirements of the preset non-real-time delay threshold, and finally a non-real-time computing power scheduling plan is obtained.

[0086] Optionally, a non-real-time computing power scheduling scheme is determined according to the aggregated data volume of the aggregated business block, the non-real-time available computing power resources, and the preset non-real-time delay threshold, which can be implemented in the following ways:

[0087] 1) The ratio of the amount of aggregated data to the non-real-time available computing resources is determined as the processing delay of the aggregated business block.

[0088] 2) If the processing delay is less than or equal to the preset non-real-time delay threshold, the aggregated business blocks are processed as a batch and determined as a non-real-time computing power scheduling solution.

[0089] 3) If the processing delay is greater than the preset non-real-time delay threshold, the aggregated service block is segmented, and the segmented aggregated service block is used as a new aggregated service block, and the process returns to step 1).

[0090] In this embodiment, the processing delay of the aggregated business block can be specifically understood as the time required to process the aggregated business block of the aggregated data volume through non-real-time available computing resources.

[0091] Specifically, when determining the non-real-time computing power scheduling plan, we can first calculate the ratio of the aggregated data volume to the non-real-time available computing power resources as the processing delay of the aggregated business block. This processing delay can be understood as the time required to process all current non-real-time business types' pending business flows through non-real-time available computing power resources. Then, the processing delay can be compared with the preset non-real-time delay threshold. If the processing delay is less than or equal to the preset non-real-time delay threshold, it can be considered that the aggregated business blocks are processed as a batch by the non-real-time available computing power resources in the edge computing device, which can also meet the business execution requirements. At this time, the aggregated business blocks can be directly processed as a batch to be determined as a non-real-time computing power scheduling solution; if the processing delay is greater than the preset non-real-time delay threshold, it can be considered that if the aggregated business blocks are processed as a batch by the non-real-time available computing power resources in the edge computing device, it is difficult to meet the business execution requirements. At this time, the aggregated business blocks can be split according to some preset rules, and the split aggregated business blocks are used as new aggregated business blocks. Each new aggregated business block is processed as a batch by the non-real-time available computing power resources in the edge computing device to determine whether each new aggregated business block can meet the business execution requirements. The above steps are repeated until all aggregated business blocks that meet the requirements are obtained, and each aggregated business block is processed as a batch to be determined as a non-real-time computing power scheduling solution.

[0092] For example, assuming D block is the aggregated data volume of the aggregated service block, C free is the non-real-time available computing power resource, T Limit is the preset non-real-time delay threshold, then the processing delay for the aggregation service block can be expressed as If T i ≤T Limit , the aggregated business blocks can be processed as a batch as a non-real-time computing power scheduling solution; if T i >T Limit, then the aggregation business block can be adjusted to D block =D block *0.5, and repeat the above steps until the condition T is met i ≤T Limit So far, the corresponding non-real-time computing power scheduling solution is obtained.

[0093] S213: Allocate the non-real-time available computing power resources according to the ratio between the amounts of data to be processed of each non-real-time service sub-queue to determine a non-real-time computing power scheduling plan.

[0094] Specifically, when the non-real-time available computing power resources can be directly used to process the pending business flows in each non-real-time business sub-queue, a method similar to the above S208 can be adopted to directly divide the non-real-time available computing power resources according to the ratio between the amount of pending data of each non-real-time business sub-queue, or according to the proportion of the amount of pending data of each non-real-time business sub-queue in the total amount of data corresponding to all non-real-time business sub-queues, and allocate each allocated computing power resource to the non-real-time business container corresponding to each non-real-time business sub-queue to obtain the corresponding non-real-time computing power orchestration plan.

[0095] Optionally, an embodiment of the present invention further provides a method for determining a computing power reservation error correction coefficient, including:

[0096] Substitute the last actual computing power resource occupancy corresponding to each moment in the previous business cycle corresponding to the current business cycle, and the last real-time computing power resource reservation corresponding to each moment into the preset correction coefficient determination formula to determine the computing power reservation error correction coefficient corresponding to the current business cycle;

[0097] Among them, the preset correction coefficient determination formula is:

[0098]

[0099] Among them, f C is the computing power reservation error correction coefficient corresponding to the current business cycle; T is the length of the previous business cycle; T i is the previous business cycle; T i+1 is the current business cycle; C real (t) is the actual computing power resource usage at the last moment t; C resv (t) is the previous real-time computing resource reservation amount at time t.

[0100] Specifically, in each business cycle, after the real-time computing power scheduling plan is determined, the pending business flows of each real-time business type will be processed according to the real-time computing power scheduling plan within the business cycle. Since the actual amount of computing power resources occupied by the real-time business sub-queue during the processing process within the business cycle will fluctuate, the computing power reservation error correction coefficient required for the next cycle can be re-determined within each business cycle based on the actual amount of computing power resources occupied within the cycle and the real-time computing power resource reservation amount. This ensures that the computing power reservation error correction coefficient used in each business cycle can fully consider the actual business processing situation in the business cycle that is most closely related to it, thereby improving the accuracy of determining the real-time computing power resource reservation amount for each business cycle.

[0101] Exemplarily, the above formula takes the previous business cycle of the current business cycle as an example to illustrate the method for determining the computing power reservation error correction coefficient in the current business cycle. The previous business cycle is divided into multiple moments, and the square of the difference between the previous real-time computing power resource reservation amount and the previous actual computing power resource occupancy corresponding to each moment is integrated. The square root of the integral value is minimized, and the computing power reservation error correction coefficient when the square root is the smallest is determined as the computing power reservation error correction coefficient corresponding to the current cycle.

[0102] It is understandable that in the current business cycle, after the real-time computing power scheduling plan is clarified, the same method as above will be used to determine the computing power reservation error correction coefficient required for the next business cycle of the current business cycle. The embodiment of the present invention will not elaborate on this.

[0103] Optionally, in order to improve the accuracy of the correction of the computing power reservation error correction coefficient to the total real-time computing power demand of the real-time business type, it is also necessary to consider the time demand for real-time business type business processing according to the real-time computing power scheduling plan in the previous business cycle. Therefore, after the computing power reservation error correction coefficient corresponding to the current business cycle is determined in the previous business cycle, it also includes:

[0104] Determine the previous processing delay of the previous business cycle based on the previous unfinished data volume corresponding to the previous business cycle, the previous real-time demand data volume corresponding to the previous real-time computing power resource reservation volume, and the previous real-time computing power resource reservation volume; if the previous processing delay is less than or equal to the preset real-time delay threshold, keep the computing power reservation error correction coefficient corresponding to the current business cycle unchanged; if the previous processing delay is greater than the preset real-time delay threshold, update the computing power reservation error correction coefficient corresponding to the current business cycle through the preset additional step size.

[0105] In this embodiment, the last unfinished data volume can be specifically understood as the data volume of the real-time business type business that has not been processed in the edge computing device when the last business cycle starts. The last processing delay can be specifically understood as the time consumed for processing the real-time business type business according to the real-time computing power scheduling plan in the last business cycle. The preset real-time delay threshold can be specifically understood as a time threshold that is pre-set according to actual conditions to determine whether the real-time business type business can be completed within the required time. The preset additional step size can be specifically understood as a time threshold that is pre-determined according to actual conditions to adjust the computing power reservation error correction coefficient so that the computing power reservation error correction coefficient is closer to the actual preset value.

[0106] Specifically, the previous unfinished data flow corresponding to the previous business cycle and the previous real-time demand data volume corresponding to the previous real-time computing power resource reservation amount can be summed to determine the total data volume that needs to be processed using the computing power resources of the previous real-time computing power resource reservation amount. Then, the total data volume is divided by the previous real-time computing power resource reservation amount to obtain the total time required to process the real-time business type of business in the previous business cycle, which can be used as the previous processing delay. Then, the previous processing delay can be compared with the preset real-time delay threshold. If the previous processing delay is less than or equal to the preset real-time delay threshold, it can be considered that the currently determined computing power reservation error correction coefficient is relatively consistent with the deviation caused in the actual operation process, and no adjustment is required, and the computing power reservation error correction coefficient corresponding to the current business cycle can be kept unchanged; if the previous processing delay is greater than the preset real-time delay threshold, it can be considered that the currently determined computing power reservation error correction coefficient is inconsistent with the deviation caused in the actual operation process. At this time, a preset additional step can be added to the computing power reservation error correction coefficient to complete the update of the computing power reservation error correction coefficient corresponding to the current business cycle, that is, the computing power reservation error correction coefficient used in the current business cycle should be the computing power reservation error correction coefficient with the increased preset additional step.

[0107] The technical solution of this embodiment, after completing the mapping of each pending business flow to the corresponding business sub-queue, statistically converts the pending data flow of each real-time business sub-queue to obtain the total real-time computing power demand, corrects it through the computing power reservation error correction coefficient to obtain the corresponding real-time computing power resource reservation, and then completes the determination of the real-time computing power scheduling plan according to the proportion of the data volume corresponding to different real-time business sub-queues in the total real-time demand data volume. After completing the determination of the real-time computing power resource reservation, the non-real-time available computing power resources available for non-real-time services can be determined based on the current remaining computing power resources and the real-time computing power resource reservation. On the basis of ensuring that real-time services can have sufficient computing power resources to process them, according to the non-real-time delay threshold required for the execution of non-real-time services, combined with data aggregation processing, the non-real-time available computing power resources are used to the fullest extent, and a non-real-time computing power scheduling plan is obtained. This enables the current remaining computing power resources in the current business cycle to be used to the fullest extent. At the end of each business cycle, the computing power reservation error correction coefficient can be re-determined, and the determined computing power reservation error correction coefficient can be updated based on the latency when processing real-time type business within this cycle, so that the finally determined computing power reservation error correction coefficient can better reflect the dynamically changing business needs and actual business processing status, and thus make each computing power scheduling plan determined based on the computing power reservation error correction coefficient more suitable for the processing of edge computing devices, thereby improving the computing power resource utilization rate of edge computing devices.

[0108] Embodiment 3

[0109] Figure 3 A schematic diagram of the structure of a computing power arrangement device for edge computing provided in Embodiment 3 of the present invention is shown in FIG. Figure 3 As shown, the computing power orchestration device for edge computing includes an information acquisition module 31, a queue mapping module 32, a first solution determination module 33 and a second solution determination module 34.

[0110] Among them, the information acquisition module 31 is used to obtain the pending business flow information of each pending business flow in the current business cycle; the queue mapping module 32 is used to map each pending business flow into a corresponding real-time business sub-queue and non-real-time business sub-queue according to the business flow characteristics in the pending business flow information; the first scheme determination module 33 is used to determine the real-time computing power resource reservation amount according to the amount of pending data of each real-time business sub-queue and the computing power reservation error correction coefficient, and determine the real-time computing power scheduling plan according to the real-time computing power resource reservation amount; the second scheme determination module 34 is used to determine the non-real-time computing power scheduling plan according to the amount of pending data of each non-real-time business sub-queue, the current remaining computing power resources and the real-time computing power resource reservation amount; wherein, the computing power reservation error correction coefficient is determined based on the previous actual computing power resource occupancy and the previous real-time computing power resource reservation amount in the previous business cycle corresponding to the current business cycle.

[0111] The technical solution of the embodiment of the present invention, when it is necessary to provide corresponding computing resources for each pending business flow in the current business cycle, firstly distinguish whether it belongs to real-time business and map the business queue according to the characteristics of different pending business flows, and use different computing power orchestration methods to determine the computing power orchestration scheme for real-time business and non-real-time business. Among them, in order to ensure that real-time business can have sufficient computing power resources for execution, when determining the computing power orchestration scheme, the real-time computing power orchestration scheme will be determined first, so as to give priority to completing the reservation of the real-time computing power resources required for real-time business among the remaining computing power resources in the edge computing device according to the determined real-time computing power orchestration scheme; and for non-real-time business, it will be determined based on the currently remaining available computing power resources of the edge computing device and the amount of real-time computing power reserved resources, so as to ensure that the real-time tasks can be executed in a timely manner, and at the same time, the remaining computing power resources in the edge computing device can be fully utilized. At the same time, the computing power reservation error correction coefficient determined according to the actual business execution situation in the previous business cycle is used to assist in the determination of the real-time computing power orchestration plan, so that the final real-time computing power orchestration plan can fully consider the dynamically changing business needs and the actual business processing status, and then the non-real-time computing power orchestration plan determined based on the real-time computing power resource reservation can also fully consider the dynamically changing business needs and the actual business processing status, and finally obtain a more flexible and efficient computing power orchestration plan suitable for edge computing devices, thereby improving the computing power resource utilization rate of edge computing devices.

[0112] Optionally, the first solution determination module 33 is specifically configured to:

[0113] Determining the total real-time required data volume according to the to-be-processed data volume of each of the real-time service sub-queues;

[0114] The total real-time computing power requirement is determined by multiplying the total real-time computing power requirement by a preset computing power conversion factor;

[0115] Determine the difference between the total real-time computing power demand and the computing power reservation error correction coefficient as the real-time computing power resource reservation amount;

[0116] The real-time computing power resource reservation amount is allocated to the real-time business container corresponding to each real-time business sub-queue according to the proportion of the amount of data to be processed of each real-time business sub-queue in the total real-time demand data amount, and determined as a real-time computing power scheduling plan.

[0117] Optionally, the second solution determination module 34 is specifically configured to:

[0118] Determine the difference between the current remaining computing power resources and the reserved amount of the real-time computing power resources as the non-real-time available computing power resources;

[0119] Determine the total non-real-time computing power requirement according to the amount of data to be processed of each of the non-real-time service sub-queues and a preset computing power conversion factor;

[0120] If the total non-real-time computing power demand is less than or equal to the non-real-time available computing power resources, the non-real-time available computing power resources are allocated according to the ratio between the amounts of data to be processed of each of the non-real-time service sub-queues to determine a non-real-time computing power scheduling plan;

[0121] If the total non-real-time computing power demand is greater than the non-real-time available computing power resources, the non-real-time service sub-queues are aggregated to generate an aggregated service block, and the non-real-time computing power scheduling plan is determined based on the aggregated data volume of the aggregated service block, the non-real-time available computing power resources and the preset non-real-time delay threshold.

[0122] Optionally, determining a non-real-time computing power scheduling scheme according to the aggregated data volume of the aggregated service block, the non-real-time available computing power resources, and a preset non-real-time delay threshold includes:

[0123] Determine the ratio of the aggregated data volume to the non-real-time available computing resources as the processing delay of the aggregated business block;

[0124] If the processing delay is less than or equal to the preset non-real-time delay threshold, the aggregated business blocks are processed as a batch and determined as a non-real-time computing power scheduling solution;

[0125] If the processing delay is greater than the preset non-real-time delay threshold, the aggregated business block is split, and the split aggregated business block is used as a new aggregated business block, and the process returns to execute the step of determining the processing delay of the aggregated business block as the ratio of the aggregated data volume to the non-real-time available computing resources.

[0126] Optionally, the queue mapping module 32 is specifically used for:

[0127] Input each of the business flow information to be processed into a pre-trained business type determination model, so as to extract the business flow features of each of the business logistics information to be processed through the business type determination model, and perform business classification operations based on each of the business flow features to determine the business type corresponding to each of the business flows to be processed;

[0128] Mapping a to-be-processed service flow of a real-time service type to a corresponding real-time service sub-queue;

[0129] Mapping a to-be-processed service flow of a non-real-time service type to a corresponding non-real-time service sub-queue;

[0130] Among them, the business type determination model is a lightweight long short-term memory network model.

[0131] Optional methods for determining the computing power reservation error correction coefficient include:

[0132] Substitute the last actual computing power resource occupancy corresponding to each moment in the last business cycle corresponding to the current business cycle, and the last real-time computing power resource reservation corresponding to each moment into the preset correction coefficient determination formula to determine the computing power reservation error correction coefficient corresponding to the current business cycle;

[0133] Wherein, the preset correction coefficient determination formula is:

[0134]

[0135] Among them, the f C is the computing power reservation error correction coefficient corresponding to the current business cycle; T is the length of the previous business cycle; T i is the previous business cycle; i+1 is the current business cycle; real (t) is the actual computing resource usage at the last moment t; resv (t) is the previous real-time computing resource reservation amount at time t.

[0136] Optionally, the computing power orchestration device for edge computing also includes: a coefficient updating module.

[0137] Among them, the coefficient updating module is used to determine the previous processing delay of the previous business cycle according to the previous unfinished data volume corresponding to the previous business cycle, the previous real-time demand data volume corresponding to the previous real-time computing power resource reservation volume, and the previous real-time computing power resource reservation volume after determining the computing power reservation error correction coefficient corresponding to the current business cycle; if the previous processing delay is less than or equal to the preset real-time delay threshold, the computing power reservation error correction coefficient corresponding to the current business cycle is kept unchanged; if the previous processing delay is greater than the preset real-time delay threshold, the computing power reservation error correction coefficient corresponding to the current business cycle is updated by a preset additional step size.

[0138] The computing power orchestration device for edge computing provided by an embodiment of the present invention can execute the computing power orchestration method for edge computing provided by any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0139] Embodiment 4

[0140] Figure 4A schematic diagram of the structure of a computing power orchestration device for edge computing provided in Embodiment 4 of the present invention. The computing power orchestration device 40 for edge computing may be intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The computing power orchestration device 40 for edge computing may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present invention described and / or required herein.

[0141] like Figure 4 As shown, the computing power orchestration device 40 for edge computing includes at least one processor 41, and a memory connected to the at least one processor 41 in communication, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 41 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 42 or the computer program loaded from the storage unit 48 to the random access memory (RAM) 43. In RAM 43, various programs and data required for the operation of the computing power orchestration device 40 for edge computing can also be stored. The processor 41, ROM 42 and RAM 43 are connected to each other via a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0142] Multiple components in the edge computing computing power orchestration device 40 are connected to the I / O interface 45, including: an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the edge computing computing power orchestration device 40 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0143] The processor 41 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 41 executes the various methods and processes described above, such as the computing power orchestration method of edge computing.

[0144] In some embodiments, the edge computing power orchestration method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 48. In some embodiments, part or all of the computer program may be loaded and / or installed on the computing power orchestration device 40 via the ROM 42 and / or the communication unit 49. When the computer program is loaded into the RAM 43 and executed by the processor 41, one or more steps of the edge computing power orchestration method described above may be performed. Alternatively, in other embodiments, the processor 41 may be configured to perform the edge computing power orchestration method by any other appropriate means (e.g., by means of firmware).

[0145] Optionally, an embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, implements the edge computing computing power orchestration method provided by any embodiment of the present invention.

[0146] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0147] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0148] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0149] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a computing power orchestration device, which has: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball), through which a user can provide input to the computing power orchestration device. Other types of devices can also be used to provide interaction with a user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0150] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0151] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0152] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0153] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A computing power arrangement method for edge computing, characterized in that: include: Obtaining the pending business flow information of each pending business flow in the current business cycle; According to the service flow characteristics in each of the service flow information to be processed, mapping each of the service flows to be processed into a corresponding real-time service sub-queue and a non-real-time service sub-queue; Determine the real-time computing power resource reservation amount according to the amount of data to be processed of each of the real-time service sub-queues and the computing power reservation error correction coefficient, and determine the real-time computing power scheduling plan according to the real-time computing power resource reservation amount; Determine a non-real-time computing power scheduling plan according to the amount of data to be processed of each of the non-real-time service sub-queues, the current remaining computing power resources, and the reserved amount of the real-time computing power resources; Among them, the computing power reservation error correction coefficient is determined according to the previous actual computing power resource occupancy and the previous real-time computing power resource reservation in the previous business cycle corresponding to the current business cycle.

2. The edge computing computing power arrangement method according to claim 1 is characterized in that: The step of determining the real-time computing power resource reservation amount according to the amount of data to be processed of each of the real-time service sub-queues and the computing power reservation error correction coefficient, and determining the real-time computing power scheduling plan according to the real-time computing power resource reservation amount includes: Determining the total real-time required data volume according to the to-be-processed data volume of each of the real-time service sub-queues; The total real-time computing power requirement is determined by multiplying the total real-time computing power requirement by a preset computing power conversion factor; Determine the difference between the total real-time computing power demand and the computing power reservation error correction coefficient as the real-time computing power resource reservation amount; The real-time computing power resource reservation amount is allocated to the real-time business container corresponding to each real-time business sub-queue according to the proportion of the amount of data to be processed of each real-time business sub-queue in the total real-time demand data amount, and determined as a real-time computing power scheduling plan.

3. The edge computing computing power arrangement method according to claim 1 is characterized in that: The determining of the non-real-time computing power scheduling scheme according to the amount of data to be processed of each of the non-real-time service sub-queues, the current remaining computing power resources and the reserved amount of the real-time computing power resources includes: Determine the difference between the current remaining computing power resources and the reserved amount of the real-time computing power resources as the non-real-time available computing power resources; Determine the total non-real-time computing power requirement according to the amount of data to be processed of each of the non-real-time service sub-queues and a preset computing power conversion factor; If the total non-real-time computing power demand is less than or equal to the non-real-time available computing power resources, the non-real-time available computing power resources are allocated according to the ratio between the amounts of data to be processed of each of the non-real-time service sub-queues to determine a non-real-time computing power scheduling plan; If the total non-real-time computing power demand is greater than the non-real-time available computing power resources, the non-real-time service sub-queues are aggregated to generate an aggregated service block, and the non-real-time computing power scheduling plan is determined based on the aggregated data volume of the aggregated service block, the non-real-time available computing power resources and the preset non-real-time delay threshold.

4. The edge computing computing power arrangement method according to claim 3 is characterized in that: The determining of the non-real-time computing power scheduling scheme according to the aggregated data volume of the aggregated business block, the non-real-time available computing power resources and the preset non-real-time delay threshold includes: Determine the ratio of the aggregated data volume to the non-real-time available computing resources as the processing delay of the aggregated business block; If the processing delay is less than or equal to the preset non-real-time delay threshold, the aggregated business blocks are processed as a batch and determined as a non-real-time computing power scheduling solution; If the processing delay is greater than the preset non-real-time delay threshold, the aggregated business block is split, and the split aggregated business block is used as a new aggregated business block, and the process returns to execute the step of determining the processing delay of the aggregated business block as the ratio of the aggregated data volume to the non-real-time available computing resources.

5. The edge computing computing power arrangement method according to claim 1 is characterized in that: Mapping each of the to-be-processed service flows into a corresponding real-time service sub-queue and non-real-time service sub-queue according to the service flow characteristics in the to-be-processed service flow information includes: Input each of the business flow information to be processed into a pre-trained business type determination model, so as to extract the business flow features of each of the business logistics information to be processed through the business type determination model, and perform business classification operations based on each of the business flow features to determine the business type corresponding to each of the business flows to be processed; Mapping a to-be-processed service flow of a real-time service type to a corresponding real-time service sub-queue; Mapping a to-be-processed service flow of a non-real-time service type to a corresponding non-real-time service sub-queue; Among them, the business type determination model is a lightweight long short-term memory network model.

6. The edge computing computing power arrangement method according to any one of claims 1 to 5, characterized in that: The method for determining the computing power reservation error correction coefficient includes: Substitute the last actual computing power resource occupancy corresponding to each moment in the last business cycle corresponding to the current business cycle, and the last real-time computing power resource reservation corresponding to each moment into the preset correction coefficient determination formula to determine the computing power reservation error correction coefficient corresponding to the current business cycle; Wherein, the preset correction coefficient determination formula is: Among them, the f C is the computing power reservation error correction coefficient corresponding to the current business cycle; T is the length of the previous business cycle; T i is the previous business cycle; i+1 is the current business cycle; real (t) is the actual computing resource usage at the last moment t; resv (t) is the previous real-time computing resource reservation amount at time t.

7. The edge computing computing power arrangement method according to claim 6 is characterized in that: After determining the computing power reservation error correction coefficient corresponding to the current business cycle, the method further includes: Determine the last processing delay of the last business cycle according to the last unfinished data volume corresponding to the last business cycle, the last real-time demand data volume corresponding to the last real-time computing power resource reservation volume, and the last real-time computing power resource reservation volume; If the previous processing delay is less than or equal to the preset real-time delay threshold, the computing power reservation error correction coefficient corresponding to the current business cycle is kept unchanged; If the previous processing delay is greater than the preset real-time delay threshold, the computing power reservation error correction coefficient corresponding to the current business cycle is updated by a preset additional step size.

8. A computing power arrangement device for edge computing, characterized in that: include: An information acquisition module is used to acquire the pending business flow information of each pending business flow in the current business cycle; A queue mapping module, used for mapping each of the to-be-processed service flows into a corresponding real-time service sub-queue and non-real-time service sub-queue according to the service flow characteristics in the information of each of the to-be-processed service flows; A first scheme determination module is used to determine the real-time computing power resource reservation amount according to the amount of data to be processed of each of the real-time service sub-queues and the computing power reservation error correction coefficient, and determine the real-time computing power scheduling scheme according to the real-time computing power resource reservation amount; A second scheme determination module is used to determine a non-real-time computing power scheduling scheme according to the amount of data to be processed of each of the non-real-time service sub-queues, the current remaining computing power resources and the reserved amount of the real-time computing power resources; Among them, the computing power reservation error correction coefficient is determined according to the previous actual computing power resource occupancy and the previous real-time computing power resource reservation in the previous business cycle corresponding to the current business cycle.

9. The computing power arrangement device for edge computing according to claim 8, characterized in that: The first solution determination module is specifically used to: Determining the total real-time required data volume according to the to-be-processed data volume of each of the real-time service sub-queues; The total real-time computing power requirement is determined by multiplying the total real-time computing power requirement by a preset computing power conversion factor; Determine the difference between the total real-time computing power demand and the computing power reservation error correction coefficient as the real-time computing power resource reservation amount; The real-time computing power resource reservation amount is allocated to the real-time business container corresponding to each real-time business sub-queue according to the proportion of the amount of data to be processed of each real-time business sub-queue in the total real-time demand data amount, and determined as a real-time computing power scheduling plan.

10. The computing power arrangement device for edge computing according to claim 8, characterized in that: The second solution determination module is specifically used to: Determine the difference between the current remaining computing power resources and the reserved amount of the real-time computing power resources as the non-real-time available computing power resources; Determine the total non-real-time computing power requirement according to the amount of data to be processed of each of the non-real-time service sub-queues and a preset computing power conversion factor; If the total non-real-time computing power demand is less than or equal to the non-real-time available computing power resources, the non-real-time available computing power resources are allocated according to the ratio between the amounts of data to be processed of each of the non-real-time service sub-queues to determine a non-real-time computing power scheduling plan; If the total non-real-time computing power demand is greater than the non-real-time available computing power resources, the non-real-time service sub-queues are aggregated to generate an aggregated service block, and the non-real-time computing power scheduling plan is determined based on the aggregated data volume of the aggregated service block, the non-real-time available computing power resources and the preset non-real-time delay threshold.

11. The computing power arrangement device for edge computing according to claim 10, characterized in that: The determining of the non-real-time computing power scheduling scheme according to the aggregated data volume of the aggregated business block, the non-real-time available computing power resources and the preset non-real-time delay threshold includes: Determine the ratio of the aggregated data volume to the non-real-time available computing resources as the processing delay of the aggregated business block; If the processing delay is less than or equal to the preset non-real-time delay threshold, the aggregated business blocks are processed as a batch and determined as a non-real-time computing power scheduling solution; If the processing delay is greater than the preset non-real-time delay threshold, the aggregated business block is split, and the split aggregated business block is used as a new aggregated business block, and the process returns to execute the step of determining the processing delay of the aggregated business block as the ratio of the aggregated data volume to the non-real-time available computing resources.

12. The computing power arrangement device for edge computing according to claim 8, characterized in that: The queue mapping module is specifically used for: Input each of the business flow information to be processed into a pre-trained business type determination model, so as to extract the business flow features of each of the business logistics information to be processed through the business type determination model, and perform business classification operations based on each of the business flow features to determine the business type corresponding to each of the business flows to be processed; Mapping a to-be-processed service flow of a real-time service type to a corresponding real-time service sub-queue; Mapping a to-be-processed service flow of a non-real-time service type to a corresponding non-real-time service sub-queue; Among them, the business type determination model is a lightweight long short-term memory network model.

13. The computing power arrangement device for edge computing according to any one of claims 8 to 12, characterized in that: The method for determining the computing power reservation error correction coefficient includes: Substitute the last actual computing power resource occupancy corresponding to each moment in the last business cycle corresponding to the current business cycle, and the last real-time computing power resource reservation corresponding to each moment into the preset correction coefficient determination formula to determine the computing power reservation error correction coefficient corresponding to the current business cycle; Wherein, the preset correction coefficient determination formula is: Among them, the f C is the computing power reservation error correction coefficient corresponding to the current business cycle; T is the length of the previous business cycle; T i is the previous business cycle; i+1 is the current business cycle; real (t) is the actual computing resource usage at the last moment t; resv (t) is the previous real-time computing resource reservation amount at time t.

14. The computing power arrangement device for edge computing according to claim 13, characterized in that: Also includes: Coefficient update module; The coefficient updating module is used to determine the previous processing delay of the previous business cycle according to the previous unfinished data volume corresponding to the previous business cycle, the previous real-time demand data volume corresponding to the previous real-time computing power resource reservation amount, and the previous real-time computing power resource reservation amount after determining the computing power reservation error correction coefficient corresponding to the current business cycle; If the previous processing delay is less than or equal to the preset real-time delay threshold, the computing power reservation error correction coefficient corresponding to the current business cycle is kept unchanged; If the previous processing delay is greater than the preset real-time delay threshold, the computing power reservation error correction coefficient corresponding to the current business cycle is updated by a preset additional step size.

15. A computing power arrangement device for edge computing, characterized in that: include: at least one processor; and a memory communicatively coupled to the at least one processor; In which, the memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the edge computing computing power orchestration method described in any one of claims 1-7.

16. A storage medium containing computer executable instructions, characterized in that: When the computer executable instructions are executed by a computer processor, they are used to execute the edge computing computing power orchestration method as described in any one of claims 1-7.

17. A computer program product, comprising a computer program, which, when executed by a processor, implements the edge computing computing power orchestration method as described in any one of claims 1-7.

Images