Adaptive intelligent task scheduling method and its application for power data processing

Through task type division and optimization of uninstallation sequence, the problem of high task response time in edge environments is solved, and the task response time is reduced and system performance is improved.

CN115080213BActive Publication Date: 2024-11-26INFORMATION & COMM BRANCH OF STATE GRID JIANGSU ELECTRIC POWER +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210798458.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-06
Publication Date
2024-11-26
Estimated Expiration
2042-07-06

AI Technical Summary

Technical Problem

The existing task offloading and resource allocation methods have high latency in the edge environment, ignoring the node heterogeneity and task differential attributes, and lacking the dynamic change of task delay sensitivity.

Method used

A task scheduling method for power data processing is proposed. By obtaining the task delay upper limit, the task type is divided, and the delay tolerance, delay high sensitivity and delay low sensitivity are offloaded to the cloud computing center or edge respectively according to the preset unloading order. node, optimize resource allocation to reduce response time.

Benefits of technology

It effectively reduces the response time of system tasks, improves system performance, and adapts to the dynamic changes of different types of tasks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115080213B_ABST
    Figure CN115080213B_ABST
Patent Text Reader

Abstract

The present invention discloses a task scheduling method and application for power data processing, which is applied to a cloud-edge collaborative system, wherein the cloud-edge collaborative system includes a cloud computing center and multiple edge nodes, and the method includes: obtaining the upper limit of delays of multiple tasks to be scheduled to determine the task types of multiple tasks to be scheduled, wherein the task types include delay-tolerant tasks, delay-sensitive tasks, and delay-insensitive tasks; and unloading multiple tasks to be scheduled to a cloud computing center or an edge node according to a preset unloading order, wherein the preset unloading order is to unload delay-tolerant tasks, delay-sensitive tasks, and delay-insensitive tasks in sequence. The method divides the task types according to the different delay sensitivities of the tasks, and optimizes the unloading and resource allocation mechanisms for different types of tasks, thereby reducing the task response time and improving the system performance.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of electric power technology, and in particular to an adaptive intelligent task scheduling method and application for electric power data processing. Background Art

[0002] With the advent of the information age, the rapid development of the Internet of Things technology, artificial intelligence, image processing and other fields has led to a spurt in the number of smart device terminals. Some smart applications are emerging and have attracted great attention, such as face recognition, natural language processing, interactive games and augmented reality. The tasks issued by these applications are usually computationally intensive and delay-sensitive. Although the CPU speed of mobile devices has been greatly improved and the speed of personal computers has also been greatly improved, they are still powerless when faced with applications that require a large amount of computing resources in a short period of time. How to improve the data processing rate and task completion quality has become a current research focus.

[0003] Cloud computing provides dynamically scalable computing and storage resources to terminals through the network, allowing tasks generated by terminals to be offloaded to the cloud for centralized processing. As a solution with high computing power, cloud computing effectively reduces the processing time of tasks and improves the endurance of mobile terminals. However, cloud computing also has certain limitations. Some delay-sensitive tasks issued by IoT applications require very short response times and may generate a large amount of data, which puts a heavy burden on network bandwidth, and cloud computing is not efficient enough to support these applications. In order to solve the high transmission latency problem caused by cloud computing, edge computing models for computing large amounts of data generated by network terminals were born.

[0004] Although edge computing reduces the data transmission distance and makes up for the shortcomings of cloud computing, its limited resources still face huge resistance when processing computing-intensive applications. Cloud computing and edge computing are complementary, and the two must work closely together to adapt to various application scenarios. The main function of edge computing is to process short-term real-time data; cloud computing is responsible for computing tasks that are difficult for edge nodes to handle. The two complement each other and are applied to big data processing in cloud computing centers and edge nodes, resulting in a "cloud-edge collaboration" system architecture, which better solves the problems existing in big data processing in the era of the Internet of Everything. This new computing model makes the processing of intelligent tasks more flexible and can better meet the needs of users. However, the new environment of cloud-edge collaboration also faces some problems. First, the resources owned by edge servers are usually very limited and cannot meet the computing needs of a large number of tasks. The impact of offloading tasks to different edge nodes on system benefits is different. On this basis, the tasks issued by terminal devices at different times are heterogeneous, and the demand for resources and the bandwidth required for transmission are different. Therefore, the task offloading strategy needs to change dynamically.

[0005] In recent years, the problem of task offloading and resource allocation has attracted the attention of many scholars and has achieved a lot of research results, but most of the research is conducted in the mobile edge computing environment. Regarding the problem of task offloading, most of the current research aims to minimize the response time, abstracting the offloading problem into a convex programming and mixed integer linear programming model to achieve task offloading of heterogeneous edge nodes. Due to the differences in geographical location and bandwidth between the cloud layer and the edge layer, offloading tasks to different locations will result in different transmission delay times. This offloading strategy effectively reduces the response time of the task. Some researchers have also proposed an adaptive particle family optimization algorithm with a hybrid genetic operator. By considering the data distribution characteristics of the edge layer and the cloud layer and the influencing factors such as the system bandwidth, computing resources and storage capacity of the edge node, the data transmission delay during task offloading is effectively reduced.

[0006] Regarding the resource allocation problem, existing work has analyzed the relationship between the energy consumption and resource utilization of edge nodes, inferred the energy consumption and optimal resource utilization of task execution, combined search optimization methods and greedy strategies to solve the resource allocation problem, improved the global optimization performance of the algorithm, and reduced the resource consumption of processing tasks.

[0007] Most of the above studies are aimed at mobile edge computing, but in some scenarios, such as the industrial Internet, most devices are wired for power supply, and the mobile edge computing model is no longer applicable. At the same time, these works do not jointly optimize task offloading and resource allocation strategy issues, nor do they study the dynamic changes of task attributes, and lack relevant research on classifying the delay sensitivity of tasks. Summary of the invention

[0008] The purpose of the present invention is to provide a task scheduling method and application for power data processing, so as to solve the problems that most of the delays in existing methods for task offloading or resource allocation are carried out in edge environments, and some of the studies ignore the heterogeneity of nodes and the different attributes between tasks, and lack research on the dynamic changes of task attributes.

[0009] To achieve the above objectives, an embodiment of the present invention provides a task scheduling method for power data processing, which is applied to a cloud-edge collaborative system, wherein the cloud-edge collaborative system includes a cloud computing center and multiple edge nodes.

[0010] In one or more embodiments of the present invention, the method includes: obtaining delay upper limits of multiple tasks to be scheduled to determine the task types of the multiple tasks to be scheduled, wherein the task types include delay-tolerant tasks, delay-sensitive tasks, and delay-insensitive tasks; unloading the multiple tasks to be scheduled to a cloud computing center or an edge node according to a preset unloading order, wherein the preset unloading order is to unload the delay-tolerant tasks, delay-sensitive tasks, and delay-insensitive tasks in sequence.

[0011] In one or more embodiments of the present invention, the multiple tasks to be scheduled are unloaded to a cloud computing center or an edge node according to a preset unloading order, specifically including: within a preset bandwidth range, unloading delay-tolerant tasks among the multiple tasks to be scheduled to a cloud computing center.

[0012] In one or more embodiments of the present invention, the multiple tasks to be scheduled are unloaded to a cloud computing center or an edge node according to a preset unloading order, specifically including: determining a first target edge node for a delay-sensitive task among the multiple tasks to be scheduled based on preset rules; calculating a minimum bandwidth required for unloading the delay-sensitive task to the first target edge node; and determining to unload the delay-sensitive task to the first target edge node or a cloud computing center based on the calculated minimum bandwidth and the current remaining bandwidth.

[0013] In one or more embodiments of the present invention, determining the first target edge node of the delay-sensitive task among the multiple tasks to be scheduled based on a preset rule specifically includes: calculating the distance between the delay-sensitive task among the multiple tasks to be scheduled and the multiple edge nodes; querying the edge nodes for remaining computing capacity in ascending order of the distances; if so, determining the edge node as the first target edge node;

[0014] And / or, based on the calculated minimum bandwidth and the current remaining bandwidth, determining to offload the delay-sensitive task to the first target edge node or the cloud computing center, specifically including: when the calculated minimum bandwidth is less than the current remaining bandwidth, unloading the delay-sensitive task to the first target edge node; and when the calculated minimum bandwidth is greater than the current remaining bandwidth, within the preset bandwidth range, unloading the delay-sensitive task to the cloud computing center.

[0015] In one or more embodiments of the present invention, after offloading the delay-sensitive task to the first target edge node, the method further includes: updating the current remaining bandwidth and / or the remaining computing capacity of the first target edge node.

[0016] In one or more embodiments of the present invention, the multiple tasks to be scheduled are unloaded to a cloud computing center or an edge node according to a preset unloading order, specifically including: determining a second target edge node for a delay-insensitive task among the multiple tasks to be scheduled based on preset rules; calculating an ideal bandwidth required for unloading the delay-insensitive task to the second target edge node; determining an actual allocated bandwidth for the delay-insensitive task based on a ratio of the calculated ideal bandwidth to the current remaining bandwidth; calculating a first response time for unloading the delay-insensitive task to the second target edge node based on the actual allocated bandwidth; calculating a second response time for unloading the delay-insensitive task to the cloud computing center within a preset bandwidth range; and determining to unload the delay-insensitive task to the second target edge node or cloud computing center based on the first response time and the second response time.

[0017] In one or more embodiments of the present invention, determining the second target edge node of the delay-insensitive task among the multiple tasks to be scheduled based on a preset rule specifically includes: calculating the distance between the delay-insensitive task among the multiple tasks to be scheduled and the multiple edge nodes; querying the edge nodes for remaining computing capacity in ascending order of the distances; if so, determining the edge node as the second target edge node;

[0018] And / or, based on the first response time and the second response time, determine to offload the delay-sensitive task to the second target edge node or cloud computing center, specifically including: when the first response time is less than the second response time, unloading the delay-sensitive task to the second target edge node; and when the first response time is greater than the second response time, within the preset bandwidth range, unloading the delay-sensitive task to the cloud computing center.

[0019] In another aspect of the present invention, a task scheduling device for power data processing is provided, the device comprising an acquisition module and an unloading module.

[0020] An acquisition module, used for acquiring delay upper limits of a plurality of tasks to be scheduled, so as to determine task types of the plurality of tasks to be scheduled, wherein the task types include delay-tolerant tasks, delay-high-sensitive tasks, and delay-low-sensitive tasks;

[0021] The unloading module is used to unload the multiple tasks to be scheduled to the cloud computing center or edge node according to a preset unloading order, wherein the preset unloading order is to unload the delay-tolerant tasks, the delay-sensitive tasks and the delay-insensitive tasks in sequence.

[0022] In one or more embodiments of the present invention, the offloading module is further configured to: offload the delay-tolerant tasks among the multiple tasks to be scheduled to a cloud computing center within a preset bandwidth range.

[0023] In one or more embodiments of the present invention, the unloading module is also used to: determine the first target edge node of the highly delay-sensitive task among the multiple tasks to be scheduled based on preset rules; calculate the minimum bandwidth required to unload the highly delay-sensitive task to the first target edge node; and determine to unload the highly delay-sensitive task to the first target edge node or cloud computing center based on the calculated minimum bandwidth and the current remaining bandwidth.

[0024] In one or more embodiments of the present invention, the unloading module is further used to: calculate the distance between the delay-sensitive task in the multiple tasks to be scheduled and the multiple edge nodes; query the edge node whether there is remaining computing capacity in the order of the distance from small to large; if so, determine the edge node as the first target edge node;

[0025] And / or, based on the calculated minimum bandwidth and the current remaining bandwidth, determining to offload the delay-sensitive task to the first target edge node or the cloud computing center, specifically including: when the calculated minimum bandwidth is less than the current remaining bandwidth, unloading the delay-sensitive task to the first target edge node; and when the calculated minimum bandwidth is greater than the current remaining bandwidth, within the preset bandwidth range, unloading the delay-sensitive task to the cloud computing center.

[0026] In one or more embodiments of the present invention, the unloading module is further configured to: update the current remaining bandwidth and / or the remaining computing capacity of the first target edge node.

[0027] In one or more embodiments of the present invention, the unloading module is also used to: determine the second target edge node of the delay-sensitive task among the multiple tasks to be scheduled based on preset rules; calculate the ideal bandwidth required for unloading the delay-sensitive task to the second target edge node; determine the actual allocated bandwidth of the delay-sensitive task based on the ratio of the calculated ideal bandwidth to the current remaining bandwidth; calculate the first response time for unloading the delay-sensitive task to the second target edge node based on the actual allocated bandwidth; calculate the second response time for unloading the delay-sensitive task to the cloud computing center within a preset bandwidth range; and determine to unload the delay-sensitive task to the second target edge node or the cloud computing center based on the first response time and the second response time.

[0028] In one or more embodiments of the present invention, the unloading module is further used to: calculate the distance between the delay-insensitive task in the multiple tasks to be scheduled and the multiple edge nodes; query the edge node whether there is remaining computing capacity in the order of the distance from small to large; if so, determine the edge node as the second target edge node;

[0029] And / or, based on the first response time and the second response time, determine to offload the delay-sensitive task to the second target edge node or cloud computing center, specifically including: when the first response time is less than the second response time, unloading the delay-sensitive task to the second target edge node; and when the first response time is greater than the second response time, within the preset bandwidth range, unloading the delay-sensitive task to the cloud computing center.

[0030] In another aspect of the present invention, an electronic device is provided, comprising: at least one processor; and a memory, wherein the memory stores instructions, and when the instructions are executed by the at least one processor, the at least one processor executes the task scheduling method for power data processing as described above.

[0031] In another aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the task scheduling method for power data processing as described above are implemented.

[0032] Compared with the prior art, the task scheduling method and application for power data processing according to the implementation mode of the present invention divides task types according to different task delay sensitivities, and optimizes the unloading and resource allocation mechanisms for different types of tasks, thereby reducing the system task response time and improving the system performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart of a task scheduling method for power data processing according to an embodiment of the present invention;

[0034] Figure 2 is a network topology diagram of a cloud-edge collaboration system according to an embodiment of the present invention;

[0035] Figure 3 is a schematic diagram of a task offloading and resource allocation mechanism of a cloud-edge collaboration system according to an embodiment of the present invention;

[0036] Figure 4 is a schematic diagram of a delay-tolerant task offloading and resource allocation mechanism according to an embodiment of the present invention;

[0037] Figure 5 is a schematic diagram of a delay-sensitive task offloading and resource allocation mechanism according to an embodiment of the present invention;

[0038] Figure 6 is a schematic diagram of a delay-insensitive task offloading and resource allocation mechanism according to an embodiment of the present invention;

[0039] Figure 7 is a structural diagram of a device for task scheduling for power data processing according to an embodiment of the present invention;

[0040] Figure 8 4 is a hardware structure diagram of an electronic device for task scheduling for power data processing according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific implementation modes.

[0042] Unless explicitly stated otherwise, throughout the specification and claims, the term “comprise” or variations such as “include” or “comprising”, etc., will be understood to include the stated elements or components but not to exclude other elements or components.

[0043] The technical solutions provided by various embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0044] Ginseng Figure 2 A typical application scenario of the task scheduling method for power data processing of the present invention can be a cloud-edge collaborative system.

[0045] The cloud-edge collaborative system is a heterogeneous system, including a cloud computing center and multiple edge nodes. The tasks issued by the terminal devices are sent to the cloud computing center or edge server nodes through wireless access points (APs). Different edge server nodes may have different amounts of resources and computing power, and the distance from each terminal device to each edge server node may also be different.

[0046] For example, let the edge server node set be N = {1, 2, ..., N}, and the terminal device set be M = {1, 2, ..., M}. The attribute description of the edge server node n at time t is a two-tuple: F n represents the computing power of edge server node n, Represents the computing capacity of edge server node n at time t.

[0047] The task request sent by the terminal device is transmitted to the edge server node n or the cloud center computing through the AP, and the bandwidth resources owned by the AP are represented by W. Generally, the bandwidth resources of the AP are limited. Therefore, the present invention hopes to study a feasible task scheduling method based on the cloud-edge collaborative system.

[0048] like Figure 1 As shown, an implementation of the task scheduling method for power data processing of the present invention is introduced. In this implementation, the method includes the following steps.

[0049] In step S101, delay upper limits of a plurality of tasks to be scheduled are obtained to determine task types of the plurality of tasks to be scheduled.

[0050] The tasks issued by each terminal device at the same time in the cloud-edge collaborative system are recorded as m. The attributes of task m are described as a triplet P m ={d m ,r m ,h m}. Among them, d m represents the amount of data that task m needs to transmit, r m represents the amount of resources required by task m, h m is the upper limit of the delay of task m.

[0051] In this embodiment, two system task delay thresholds can be set, namely H1 and H2, by comparing the delay upper limit h of task m m and the size of the two system task delay thresholds H1 and H2, task m is divided into delay-tolerant, delay-sensitive, and delay-insensitive types. Specifically: if h m ≥H2, then task m is determined as a delay-tolerant task; if h m≤ H1, then task m is determined as a high latency - sensitive task; if H1 < h m < H2, then task m is determined as a low latency - sensitive task.

[0052] In step S102, the multiple tasks to be scheduled are unloaded to the cloud computing center or the edge node according to the preset unloading order.

[0053] See Figure 3 , in this embodiment, the preset unloading order is to unload the latency - tolerant tasks, high latency - sensitive tasks, and low latency - sensitive tasks in sequence.

[0054] To describe the unloading situation of task m, first define the unloading matrix Y. For the unloading variable (m ∈ M, n ∈ N), if represents that task m is unloaded to edge server node n at time slot t; if represents that task m is not unloaded at edge server node n at time slot t.

[0055] The response time generated when unloading each task to be scheduled includes both the transmission delay generated by transmitting task data from the terminal device to the edge server node or the cloud computing center, and the computing delay generated by the edge server node or the cloud computing center calculating the received task data. Furthermore, by summing the transmission delay and the computing delay, the response time of each task to be scheduled can be obtained.

[0056] Exemplarily, set the response time of task m as T m , then there is:

[0057] T m = TT m + CT m (1)

[0058] Wherein, TT m represents the transmission delay generated by task m transmitting from the terminal device to edge server node n, and its calculation formula is:

[0059]

[0060] Wherein, is the value of the unloading variable between task m and edge server node n, d m is the amount of data that task m needs to transmit; is the transmission rate of task m transmitting to edge server node n at time t, which can be calculated according to the distance between the terminal device and edge server node n and the bandwidth allocated by the wireless access point during the task transmission process, and its calculation formula is:

[0061]

[0062] Among them, L mn represents the distance between the terminal device that issues task m and the edge server node n; It represents the bandwidth allocated by AP to task m at time t.

[0063] CT m It represents the computational delay generated by edge server node n for computing the received task m, and its calculation formula is:

[0064]

[0065] in, is the value of the offloading variable between task m and edge server node n, r m is the bandwidth required by task m; F n is the computing capacity of edge server node n.

[0066] If a task is decided to be offloaded to an edge server node, the computation time CT of this task needs to be calculated. m and transmission time TT m If this task is decided to be offloaded to the cloud computing center, since the resources of the cloud computing center are sufficient, the computing time of this task is CT m Can be ignored.

[0067] This paper studies the real-time task offloading and resource allocation strategy in the cloud-edge collaborative environment, with the goal of minimizing the response time of tasks in the system. Its objective function is:

[0068]

[0069] The constraints include:

[0070]

[0071]

[0072]

[0073] Among them, the constraint condition shown in formula (6) is that the number of tasks offloaded to the edge server node n does not exceed the computing capacity of the edge server node n; the constraint condition shown in formula (7) indicates that the sum of the bandwidth allocated to each task does not exceed the total bandwidth currently owned by the cloud-edge collaborative system; the constraint condition shown in formula (8) indicates that the offload variable can only take 0 or 1.

[0074] Due to the task attribute P in the cloud-edge collaborative environment mIt will change with time, and the status of the cloud-edge collaborative system, including the remaining bandwidth of the AP and the computing capacity of the edge server nodes, is different at each moment. Therefore, it is necessary to dynamically adjust the task offloading and resource allocation strategies of different task types according to the changes in the remaining bandwidth of the AP and the computing capacity of the edge server nodes.

[0075] Ginseng Figure 4 , introduces a method for offloading delay-tolerant tasks from tasks to be scheduled.

[0076] In this implementation, for delay-tolerant tasks issued by the terminal device at time t, all of them are offloaded to the cloud computing center within the preset bandwidth range. Here, considering the bandwidth limitation of the cloud-edge collaborative system, a fixed amount of bandwidth resources, i.e., the preset bandwidth, is allocated to the tasks offloaded to the cloud computing center.

[0077] In this process, the response time of the delay-tolerant task to be scheduled can also be calculated. Usually, the computing resources of the cloud computing center are sufficient and the computing power is strong. Therefore, the computing time CT of task m in the cloud computing center can be calculated. m Approximately 0, set to CT m = 0. However, the distance between the cloud computing center and the terminal device is usually far, resulting in a long transmission time for the task.

[0078] Specifically, the preset bandwidth and the distance L between the terminal device that issues the task m and the cloud computing center can be combined. mn , according to formula (3), calculate the transmission speed between this terminal device and the cloud computing center Then, combined with the transmission speed The transmission time TT of task m is calculated based on formula (2) m Finally, the final response time of task m is calculated according to formula (1).

[0079] Since all delay-tolerant tasks are offloaded to the cloud computing center and do not occupy the capacity of the edge server nodes, it is necessary to update the offloading variables of task m on the edge server nodes. Make

[0080] Ginseng Figure 5 , introduces a method for offloading delay-sensitive tasks from tasks to be scheduled.

[0081] For delay-sensitive tasks issued by terminal devices at time t, if the system bandwidth and edge server node resources are sufficient, the first choice is to offload them to the edge server nodes. If there are not enough edge server node resources or system bandwidth, the task will be offloaded to the cloud computing center.

[0082] In this implementation, the first target edge node of the delay-sensitive task among the multiple tasks to be scheduled can be determined based on a preset rule. Exemplarily, the preset rule can be to use a greedy algorithm to find the edge node n that is closest to the terminal device that issues the delay-sensitive task and has remaining computing capacity as the first target edge node for offloading the delay-sensitive task.

[0083] Specifically, the distances between the delay-sensitive tasks and the edge nodes among the multiple tasks to be scheduled can be calculated first, and the edge nodes can be queried in ascending order of distances to see whether there is any remaining computing capacity; if so, the edge node is determined as the first target edge node, and the computing time CT at the first target edge node is calculated. m .

[0084] Then, the minimum bandwidth required for offloading the delay-sensitive task to the first target edge node is calculated. m And the calculated calculation time CT m , calculate the upper limit of the task transmission time according to formula (1); then use formula (2) and formula (3) to obtain the minimum AP bandwidth w that needs to be allocated while meeting the transmission time requirement of task m.

[0085] Finally, based on the calculated minimum bandwidth and the current remaining bandwidth, it is determined to offload the delay-sensitive task to the first target edge node or cloud computing center.

[0086] Specifically, the required bandwidth resource w can be compared with the current remaining bandwidth W of the edge server node. If W ≥ w, the bandwidth of w is allocated to task m and the offload variable is updated. At the same time, update the value of W and the remaining computing capacity of edge server node n, and calculate the response delay of task m. If W < w, it means that there is not enough system bandwidth to allocate to this task m, and task m is offloaded to the cloud computing center. After determining the offloading location of task m, according to the attribute P of task m m The response time of task m is calculated based on factors such as the offloading location and the like, and the offloading of this part of the task is completed.

[0087] Here, the specific method of offloading the delay-sensitive task to the cloud computing center can refer to the above-mentioned implementation method of offloading the delay-tolerant task to the cloud computing center, which will not be repeated here. In addition, when offloading the delay-sensitive task to the cloud computing center, the occupied bandwidth can be the preset bandwidth allocated to the delay-tolerant task, that is, the delay-sensitive task is offloaded to the cloud computing center within the preset bandwidth range.

[0088] Since the cloud-edge collaborative system has allocated a portion of the bandwidth for the transmission of tasks that are highly sensitive to delay, the remaining bandwidth can no longer meet the transmission requirements of the remaining tasks. Therefore, when unloading tasks that are less sensitive to delay, it is necessary to redetermine resource allocation, that is, it is necessary to update the system's current remaining bandwidth and / or the remaining computing capacity of the first target edge node.

[0089] Ginseng Figure 6 , introduces a method for offloading delay-insensitive tasks from tasks to be scheduled.

[0090] In this implementation, the second target edge node of the delay-insensitive task among the multiple tasks to be scheduled may also be determined based on a preset rule. That is, based on a greedy algorithm, an edge server node n that is closest to the terminal device that issues the delay-insensitive task and has remaining computing capacity is found as the second target edge node for unloading the delay-insensitive task.

[0091] Specifically, the distances between the delay-insensitive tasks and the edge nodes among the multiple tasks to be scheduled can be calculated first, and the edge nodes can be queried in ascending order of distances to see whether there is any remaining computing capacity; if so, the edge node is determined as the second target edge node, and the computing time CT of the edge server node n is calculated. m , and then obtain the transmission time requirement of task m.

[0092] Then, the ideal bandwidth required for offloading the delay-insensitive task to the second target edge node is calculated according to formula (2) and formula (3): Then calculate the ratio of the remaining amount of system bandwidth resources at time t to the total ideal bandwidth required by this part of task m, and use the formula Calculate the actual bandwidth allocation for task m.

[0093] Finally, the first response time T of the delay-insensitive task offloaded to the edge server node is calculated according to formulas (1) to (4). e and the second response time T of the low-delay sensitive task offloaded to the cloud computing center c Compare the response time T of the task at the edge server node e and the task response time T in the cloud computing center c The size of T c<T e , then task m is offloaded to the cloud computing center; otherwise, task m is offloaded to edge server node n and bandwidth resources are allocated After the unloading location of task m is finally determined, according to the attribute P of task m m The response time of task m is calculated based on factors such as the offloading location and the like, and the offloading of this part of the task is completed.

[0094] Similarly, the specific method of offloading the delay-insensitive task to the cloud computing center can refer to the above-mentioned implementation method of offloading the delay-tolerant task to the cloud computing center, which will not be repeated here. In addition, when offloading the delay-insensitive task to the cloud computing center, the occupied bandwidth can also be the preset bandwidth allocated to the delay-tolerant task, that is, the delay-insensitive task is offloaded to the cloud computing center within the preset bandwidth range.

[0095] like Figure 7 As shown, a task scheduling device for power data processing according to a preferred embodiment of the present invention.

[0096] In an embodiment of the present invention, the task scheduling device for power data processing includes an acquisition module 201 and an unloading module 202 .

[0097] An acquisition module 201 is used to acquire delay upper limits of a plurality of tasks to be scheduled, so as to determine task types of the plurality of tasks to be scheduled, wherein the task types include delay-tolerant tasks, delay-high-sensitive tasks, and delay-low-sensitive tasks;

[0098] The unloading module 202 is used to unload the multiple tasks to be scheduled to the cloud computing center or edge node according to a preset unloading order, wherein the preset unloading order is to unload the delay-tolerant tasks, the delay-sensitive tasks and the delay-insensitive tasks in sequence.

[0099] The unloading module 202 is further configured to: unload the delay-tolerant tasks among the plurality of tasks to be scheduled to the cloud computing center within a preset bandwidth range.

[0100] The unloading module 202 is also used to: determine the first target edge node of the delay-sensitive task among the multiple tasks to be scheduled based on preset rules; calculate the minimum bandwidth required to unload the delay-sensitive task to the first target edge node; and determine to unload the delay-sensitive task to the first target edge node or cloud computing center based on the calculated minimum bandwidth and the current remaining bandwidth.

[0101] The unloading module 202 is further used to: calculate the distance between the delay-sensitive task in the multiple tasks to be scheduled and the multiple edge nodes; query whether the edge node has remaining computing capacity according to the order of the distance from small to large; if so, determine the edge node as the first target edge node;

[0102] And / or, based on the calculated minimum bandwidth and the current remaining bandwidth, determining to offload the delay-sensitive task to the first target edge node or the cloud computing center, specifically including: when the calculated minimum bandwidth is less than the current remaining bandwidth, unloading the delay-sensitive task to the first target edge node; and when the calculated minimum bandwidth is greater than the current remaining bandwidth, within the preset bandwidth range, unloading the delay-sensitive task to the cloud computing center.

[0103] The unloading module 202 is further configured to update the current remaining bandwidth and / or the remaining computing capacity of the first target edge node.

[0104] The unloading module 202 is also used to: determine the second target edge node of the delay-insensitive task among the multiple tasks to be scheduled based on preset rules; calculate the ideal bandwidth required for unloading the delay-insensitive task to the second target edge node; determine the actual allocated bandwidth of the delay-insensitive task based on the ratio of the calculated ideal bandwidth to the current remaining bandwidth; calculate the first response time for unloading the delay-insensitive task to the second target edge node based on the actual allocated bandwidth; calculate the second response time for unloading the delay-insensitive task to the cloud computing center within a preset bandwidth range; and determine to unload the delay-insensitive task to the second target edge node or the cloud computing center based on the first response time and the second response time.

[0105] The unloading module 202 is further used to: calculate the distance between the delay-insensitive task in the multiple tasks to be scheduled and the multiple edge nodes; query whether there is remaining computing capacity of the edge node according to the order of the distance from small to large; if so, determine the edge node as the second target edge node;

[0106] And / or, based on the first response time and the second response time, determine to offload the delay-sensitive task to the second target edge node or cloud computing center, specifically including: when the first response time is less than the second response time, unloading the delay-sensitive task to the second target edge node; and when the first response time is greater than the second response time, within the preset bandwidth range, unloading the delay-sensitive task to the cloud computing center.

[0107] Figure 8FIG. 2 shows a hardware structure diagram of an electronic device 30 for task scheduling for power data processing according to an embodiment of the present specification. Figure 8 As shown, the electronic device 30 may include at least one processor 301, a memory 302 (e.g., a non-volatile memory), a memory 303, and a communication interface 304, and the at least one processor 301, the memory 302, the memory 303, and the communication interface 304 are connected together via a bus 305. At least one processor 301 executes at least one computer-readable instruction stored or encoded in the memory 302.

[0108] It should be understood that the computer executable instructions stored in the memory 302, when executed, cause at least one processor 301 to perform the above combined operations in various embodiments of this specification. Figure 1-6 Describes the various operations and functions.

[0109] In the embodiments of the present specification, the electronic device 30 may include, but is not limited to, personal computers, server computers, workstations, desktop computers, laptop computers, notebook computers, mobile computing devices, smart phones, tablet computers, cellular phones, personal digital assistants (PDAs), handheld devices, messaging devices, wearable computing devices, consumer electronic devices, and the like.

[0110] According to one embodiment, a program product such as a computer-readable storage medium is provided. The computer-readable storage medium may have instructions (i.e., the above-mentioned elements implemented in the form of software), which, when executed by a computer, causes the computer to perform the above-mentioned combination of various embodiments of this specification. Figure 1-6 Specifically, a system or device equipped with a readable storage medium may be provided, on which a software program code implementing the functions of any of the above-mentioned embodiments is stored, and a computer or processor of the system or device reads and executes the instructions stored in the readable storage medium.

[0111] According to the task scheduling method and application for power data processing according to the implementation mode of the present invention, the tasks are first divided according to the different delay sensitivities of the tasks, and then the unloading and resource allocation mechanisms of tasks of different delay types are studied and optimized respectively, so as to obtain the unloading position and resource allocation amount of each task, and calculate the response time of each task; finally, the response time of each task is minimized, so as to reduce the task response time and improve the system performance.

[0112] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0113] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0114] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0115] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0116] The foregoing description of specific exemplary embodiments of the present invention is for the purpose of illustration and demonstration. These descriptions are not intended to limit the present invention to the precise form disclosed, and it is clear that many changes and variations can be made based on the above teachings. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present invention and its practical application, so that those skilled in the art can realize and utilize various different exemplary embodiments of the present invention and various different selections and changes. The scope of the present invention is intended to be limited by the claims and their equivalents.

Claims

1. A task scheduling method for power data processing, applied in a cloud-edge collaborative system, wherein the cloud-edge collaborative system includes a cloud computing center and multiple edge nodes, characterized in that: The method comprises: Obtaining delay upper limits of a plurality of tasks to be scheduled to determine task types of the plurality of tasks to be scheduled, wherein the task types include delay-tolerant tasks, delay-high-sensitive tasks, and delay-low-sensitive tasks; Unloading the plurality of tasks to be scheduled to a cloud computing center or an edge node according to a preset unloading order, wherein the preset unloading order is to sequentially unload the delay-tolerant tasks, the delay-sensitive tasks, and the delay-insensitive tasks; Among them, unloading the multiple tasks to be scheduled to the cloud computing center or edge node according to the preset unloading order specifically includes: unloading the delay-tolerant tasks among the multiple tasks to be scheduled to the cloud computing center within the preset bandwidth range; determining the first target edge node of the delay-sensitive tasks among the multiple tasks to be scheduled based on the preset rules; calculating the minimum bandwidth required for unloading the delay-sensitive tasks to the first target edge node; determining to unload the delay-sensitive tasks to the first target edge node or the cloud computing center based on the calculated minimum bandwidth and the current remaining bandwidth; determining the second target edge node of the delay-insensitive tasks among the multiple tasks to be scheduled based on the preset rules; calculating the ideal bandwidth required for unloading the delay-insensitive tasks to the second target edge node; determining the actual allocated bandwidth of the delay-insensitive tasks based on the ratio of the calculated ideal bandwidth to the current remaining bandwidth; calculating the first response time of unloading the delay-insensitive tasks to the second target edge node based on the actual allocated bandwidth; calculating the second response time of unloading the delay-insensitive tasks to the cloud computing center within the preset bandwidth range; determining to unload the delay-insensitive tasks to the second target edge node or the cloud computing center based on the first response time and the second response time; Based on the calculated minimum bandwidth and the current remaining bandwidth, determining to offload the delay-sensitive task to the first target edge node or the cloud computing center specifically includes: when the calculated minimum bandwidth is less than the current remaining bandwidth, offloading the delay-sensitive task to the first target edge node; and when the calculated minimum bandwidth is greater than the current remaining bandwidth, within the preset bandwidth range, offloading the delay-sensitive task to the cloud computing center; Based on the first response time and the second response time, determine to offload the delay-sensitive task to the second target edge node or the cloud computing center, specifically including: when the first response time is less than the second response time, unloading the delay-sensitive task to the second target edge node; and when the first response time is greater than the second response time, within the preset bandwidth range, unloading the delay-sensitive task to the cloud computing center.

2. The task scheduling method for power data processing according to claim 1, characterized in that: Determining a first target edge node of a delay-sensitive task among the multiple tasks to be scheduled based on a preset rule specifically includes: Calculating the distances between the delay-sensitive tasks among the multiple tasks to be scheduled and the multiple edge nodes; According to the order of the distance from small to large, query whether the edge node has remaining computing capacity; if so, The edge node is determined as a first target edge node.

3. The task scheduling method for power data processing according to claim 2, characterized in that: After offloading the delay-sensitive task to the first target edge node, the method further includes: The current remaining bandwidth and / or the remaining computing capacity of the first target edge node is updated.

4. The task scheduling method for power data processing according to claim 1, characterized in that: Determining the second target edge node of the delay-insensitive task among the multiple tasks to be scheduled based on a preset rule specifically includes: Calculating the distance between the delay-insensitive task among the multiple tasks to be scheduled and the multiple edge nodes; According to the order of the distance from small to large, query whether the edge node has remaining computing capacity; if so, The edge node is determined as a second target edge node.

5. A task scheduling device for power data processing, characterized in that: The device comprises: An acquisition module, used for acquiring delay upper limits of a plurality of tasks to be scheduled, so as to determine task types of the plurality of tasks to be scheduled, wherein the task types include delay-tolerant tasks, delay-high-sensitive tasks, and delay-low-sensitive tasks; An unloading module, used for unloading the plurality of tasks to be scheduled to a cloud computing center or an edge node according to a preset unloading order, wherein the preset unloading order is to unload the delay-tolerant tasks, the delay-sensitive tasks, and the delay-insensitive tasks in sequence; Among them, unloading the multiple tasks to be scheduled to the cloud computing center or edge node according to the preset unloading order specifically includes: unloading the delay-tolerant tasks among the multiple tasks to be scheduled to the cloud computing center within the preset bandwidth range; determining the first target edge node of the delay-sensitive tasks among the multiple tasks to be scheduled based on the preset rules; calculating the minimum bandwidth required for unloading the delay-sensitive tasks to the first target edge node; determining to unload the delay-sensitive tasks to the first target edge node or the cloud computing center based on the calculated minimum bandwidth and the current remaining bandwidth; determining the second target edge node of the delay-insensitive tasks among the multiple tasks to be scheduled based on the preset rules; calculating the ideal bandwidth required for unloading the delay-insensitive tasks to the second target edge node; determining the actual allocated bandwidth of the delay-insensitive tasks based on the ratio of the calculated ideal bandwidth to the current remaining bandwidth; calculating the first response time of unloading the delay-insensitive tasks to the second target edge node based on the actual allocated bandwidth; calculating the second response time of unloading the delay-insensitive tasks to the cloud computing center within the preset bandwidth range; determining to unload the delay-insensitive tasks to the second target edge node or the cloud computing center based on the first response time and the second response time; Based on the calculated minimum bandwidth and the current remaining bandwidth, determining to offload the delay-sensitive task to the first target edge node or the cloud computing center specifically includes: when the calculated minimum bandwidth is less than the current remaining bandwidth, offloading the delay-sensitive task to the first target edge node; and when the calculated minimum bandwidth is greater than the current remaining bandwidth, within the preset bandwidth range, offloading the delay-sensitive task to the cloud computing center; Based on the first response time and the second response time, determine to offload the delay-sensitive task to the second target edge node or the cloud computing center, specifically including: when the first response time is less than the second response time, unloading the delay-sensitive task to the second target edge node; and when the first response time is greater than the second response time, within the preset bandwidth range, unloading the delay-sensitive task to the cloud computing center.

6. An electronic device, characterized in that: The electronic device comprises: at least one processor; and A memory storing instructions, which, when executed by the at least one processor, enables the at least one processor to execute the task scheduling method for power data processing as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the task scheduling method for power data processing according to any one of claims 1 to 4 is implemented.