CN-122027629-A - Cloud edge task scheduling method and device, computer equipment and readable storage medium

CN122027629ACN 122027629 ACN122027629 ACN 122027629ACN-122027629-A

Abstract

The application relates to a cloud edge task scheduling method, a cloud edge task scheduling device, computer equipment and a readable storage medium. The method comprises the steps of obtaining a hierarchical virtual network formed by networking cloud equipment, edge side equipment and terminal equipment, receiving video tasks to be processed, determining a current hierarchical virtual network corresponding to the video tasks to be processed from the hierarchical virtual network, determining algorithm capability information and load information of each equipment in the current hierarchical virtual network, carrying out affinity verification on the equipment and the video tasks to be processed according to the algorithm capability information and the load information to obtain corresponding matching degrees, and determining task issuing execution equipment of the video tasks to be processed from the current hierarchical virtual network according to the matching degrees. By adopting the method, the task and the equipment can be accurately matched, and the scheduling success rate is improved.

Inventors

CAO MIN
LUO TINGTING
XU YIBIAO

Assignees

浙江大华技术股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260324

Claims (10)

1. The cloud side task scheduling method is characterized by comprising the following steps of: Acquiring a hierarchical virtual network formed by networking cloud equipment, edge side equipment and terminal equipment; Receiving a video task to be processed, and determining a current level virtual network corresponding to the video task to be processed from the level virtual network; determining algorithm capability information and load information of each device in the current-level virtual network; carrying out affinity verification on the equipment and the video task to be processed according to the algorithm capability information and the load information to obtain corresponding matching degree; and determining task issuing execution equipment of the video task to be processed from the current-level virtual network according to the matching degree.
2. The method of claim 1, wherein the obtaining a hierarchical virtual network formed by networking cloud devices, edge devices, and terminal devices comprises: Acquiring equipment network attribute information of the terminal equipment; And determining a registration networking node of the terminal equipment from the edge side equipment and the cloud equipment based on the equipment network attribute information, and adding the terminal equipment to a network taking the registration networking node as a center to obtain a hierarchical virtual network.
3. The method according to claim 2, wherein the determining, based on the device network attribute information, the registered networking node of the terminal device from the edge side device and the cloud device includes: Determining network delay, bandwidth, region to which the equipment belongs and equipment type in the equipment network attribute information; According to the network delay, the bandwidth, the region to which the equipment belongs and the equipment type, respectively determining the index values between the terminal equipment and the edge side equipment and the cloud equipment according to a networking calculation mode; And determining the edge side equipment corresponding to the index value with the largest value and/or the cloud equipment as a registration networking node of the terminal equipment.
4. The method according to claim 1, wherein the performing affinity checking on the device and the video task to be processed according to the algorithm capability information and the load information to obtain a corresponding matching degree includes: If the video task to be processed comes from the first equipment in the hierarchical virtual network, carrying out affinity verification on the first equipment and the video task to be processed according to the algorithm capability information and the load information of the first equipment to obtain corresponding matching degree; and determining task issuing execution equipment of the video task to be processed from the current-level virtual network according to the matching degree, wherein the task issuing execution equipment comprises the following components: And if the matching degree meets the preset condition, determining the first equipment as task issuing execution equipment of the video task to be processed.
5. The method according to claim 4, wherein the method further comprises: If the matching degree does not meet the preset condition, determining algorithm capability information and load information of a second device except the first device of the current-level virtual network where the first device is located; And executing the algorithm capability information and the load information of the root, and performing affinity verification on the equipment and the video task to be processed to obtain a corresponding matching degree, thereby obtaining task issuing execution equipment of the video task to be processed.
6. The method according to any one of claims 1 to 5, further comprising: If the task issuing execution equipment of the video task to be processed does not exist in the current-level virtual network, determining an upper-level virtual network of the current-level virtual network from the hierarchical virtual network; Updating the current level virtual network into an upper level virtual network, executing the step of determining algorithm capability information and load information of each device in the current level virtual network, and determining task issuing execution equipment of the video task to be processed from the upper level virtual network.
7. The method according to any one of claims 1 to 4, wherein performing affinity checking on the device and the video task to be processed according to the algorithm capability information and the load information to obtain a corresponding matching degree includes: determining a target algorithm and task priority required by processing the video task to be processed; And carrying out affinity verification on the equipment and the video task to be processed according to the algorithm capability information, the load information, the target algorithm and the task priority to obtain corresponding matching degree.
8. The method according to claim 7, wherein performing affinity checking on the device and the video task to be processed according to the algorithm capability information, the load information, the target algorithm and the task priority to obtain a corresponding matching degree includes: determining a decoding capability matching value of the equipment for processing the video task to be processed according to a task list, algorithm resolving capability and equipment decoding capability existing on the equipment in the algorithm capability information; Determining an algorithm capability matching value of the equipment for processing the video task to be processed according to the loaded algorithm, the algorithm switching operation capability and the type of the target algorithm of the equipment in the algorithm capability information; Determining a resource load penalty value according to the load of the equipment of the load information and/or the load of the virtual network of the current hierarchy where the equipment is located; determining a priority gain value of the video task to be processed according to the task priority; and obtaining the corresponding matching degree according to the decoding capability matching value, the algorithm capability matching value, the resource load penalty value and the priority gain value.
9. A cloud end task scheduling device, the device comprising: the virtual network determining module is used for obtaining a hierarchical virtual network formed by networking the cloud end equipment, the edge side equipment and the terminal equipment; the task acquisition module is used for receiving a video task to be processed and determining a current-level virtual network corresponding to the video task to be processed from the hierarchical virtual network; The data determining module is used for determining algorithm capability information and load information of each device in the current-level virtual network; The task scheduling module is used for carrying out affinity verification on the equipment and the video task to be processed according to the algorithm capability information and the load information to obtain corresponding matching degree; and determining task issuing execution equipment of the video task to be processed from the current-level virtual network according to the matching degree.
10. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 8 when the computer program is executed.

Description

Cloud edge task scheduling method and device, computer equipment and readable storage medium Technical Field The present application relates to the field of task scheduling technologies, and in particular, to a cloud edge task scheduling method, a cloud edge task scheduling device, a computer device, and a readable storage medium. Background With the development of intelligent technology, algorithms show a diversified trend, including a series of algorithms such as specialized algorithms, small micro-model algorithms, algorithms capable of being generated in a customized manner, large multi-model algorithms and the like, however, the environments, devices and the like on which the algorithms depend are different. And then, correspondingly producing a plurality of algorithm application devices with different scales corresponding to different scenes. For example, the devices include large central clusters, small and medium devices, client local devices, and the like. Various devices can bear different task numbers, algorithm types and process data which can be processed respectively, and the devices are primarily divided into three types of devices of cloud, side and end. In the field of algorithm analysis of videos, a scene is generally encountered in which the same video needs to be analyzed by multiple algorithms, or different videos need to be analyzed by the same algorithm. Meanwhile, the method has the advantages that different equipment can load different kinds of algorithms, different algorithms have different analysis capacities on different equipment, the video decoding capacity of the same equipment and the analysis capacity of the algorithms have a gap, part of equipment has algorithm switching capacity, part of equipment does not support algorithm switching and the like. However, the existing task scheduling performs task allocation according to the idle computing capacity, memory size and other resource conditions of the equipment, so that the matching between the task and the equipment is low, and the scheduling success rate is low. Disclosure of Invention In view of the foregoing, it is desirable to provide a cloud end task scheduling method, apparatus, computer device, computer readable storage medium, and computer program product that can achieve accurate matching of tasks and devices and improve scheduling success rate. In a first aspect, the present application provides a cloud edge task scheduling method, including: Acquiring a hierarchical virtual network formed by networking cloud equipment, edge side equipment and terminal equipment; Receiving a video task to be processed, and determining a current level virtual network corresponding to the video task to be processed from the level virtual network; determining algorithm capability information and load information of each device in the current-level virtual network; carrying out affinity verification on the equipment and the video task to be processed according to the algorithm capability information and the load information to obtain corresponding matching degree; And determining target equipment from the current-level virtual network according to the matching degree, and determining the target equipment as task issuing execution equipment of the video task to be processed. In one embodiment, the obtaining a hierarchical virtual network formed by networking a cloud device, an edge device, and a terminal device includes: Acquiring equipment network attribute information of the terminal equipment; And determining a registration networking node of the terminal equipment from the edge side equipment and the cloud equipment based on the equipment network attribute information, and adding the terminal equipment to a network taking the registration networking node as a center to obtain a hierarchical virtual network. In one embodiment, the determining, based on the device network attribute information, the registered networking node of the terminal device from the edge device and the cloud device includes: Determining network delay, bandwidth, region to which the equipment belongs and equipment type in the equipment network attribute information; According to the network delay, the bandwidth, the region to which the equipment belongs and the equipment type, respectively determining the index values between the terminal equipment and the edge side equipment and the cloud equipment according to a networking calculation mode; And determining the edge side equipment corresponding to the index value with the largest value and/or the cloud equipment as a registration networking node of the terminal equipment. In one embodiment, the performing affinity checking on the device and the video task to be processed according to the algorithm capability information and the load information to obtain a corresponding matching degree includes: If the video task to be processed comes from the first equipment in the hierarchical virtual network, carrying out af