Cloud video platform system supporting access of multi-source heterogeneous video equipment

Abstract

The invention belongs to the technical field of nuclear power, and particularly relates to a cloud video platform system supporting access of multi-source heterogeneous video equipment. According to the system provided by the invention, by standardizing the configuration file, the access difficulty and development cost of the new equipment / manufacturer platform are greatly reduced, and the access of the configuration is realized. Resources (CPU / MEM / GPU) are allocated as required based on accurate pre-estimation of configuration files and Kubernetes dynamic scheduling, so that waste is avoided; the HPA realizes automatic elastic expansion and contraction to cope with load fluctuation. A high availability mechanism built in Kubernetes and a strategy based on configuration are utilized to ensure high reliability and business continuity of video processing service. Platform resources and service life cycles are uniformly managed through Kubernetes, the automation degree is high, and the operation and maintenance complexity is reduced. And flexible addition of new equipment types, processing services and resource types is supported. The resource utilization rate is improved, and excessive configuration is reduced; and the labor cost is reduced through automatic operation and maintenance.

Description

Technical Field

[0001] This invention belongs to the field of nuclear power technology, specifically relating to a cloud video platform system that supports access to multi-source heterogeneous video devices. Background Technology

[0002] Existing cloud video platforms typically encounter the following problems when connecting to video devices or platforms from different vendors, using different protocols (RTSP, RTMP, ONVIF, GB / T28181, proprietary protocols, etc.), and with varying capabilities (resolution, encoding formats, AI analysis requirements): Customized development: Writing specific access adapters for each device type or vendor API results in high development and maintenance costs. Static resource configuration: Pre-allocating fixed computing resources (CPU, memory, GPU) to different devices or services cannot be dynamically adjusted based on real-time load, leading to low resource utilization or performance bottlenecks. High availability challenges: Achieving failover and load balancing for services such as video stream processing, transcoding, and analysis usually requires complex middleware or customized architectures, making deployment and maintenance difficult. Insufficient elasticity: Resource expansion / contraction is not agile enough to handle sudden traffic spikes or device additions / removals. Summary of the Invention

[0003] To overcome the problems existing in related technologies, a cloud video platform system supporting the access of multi-source heterogeneous video devices is provided. The system includes: a device configuration file management module;

[0004] The device configuration file management module is used to store and manage configuration files that describe device / platform access attributes, media stream parameters, required processing services, and resource requirements.

[0005] The unified access gateway is used to receive device / platform requests and load the corresponding configuration file based on the request identifier.

[0006] The dynamic resource scheduling engine based on Kubernetes includes a configuration file parser and a custom controller. The parser parses the device configuration file into Kubernetes resource requirements. The custom controller dynamically generates or updates Kubernetes resource definitions based on the parsed Kubernetes resource requirements, including defining the workload of the processing service, the service exposure method, and the automatic scaling strategy.

[0007] The video processing service instance running on the Kubernetes cluster is created and managed by the dynamic resource scheduling engine and is used to execute the video processing tasks specified in the device configuration file.

[0008] In one possible implementation, the device configuration file is defined using a structured data format and includes at least the access protocol type, video encoding format, resolution, required video processing service type, estimated resource requirements, and minimum number of replicas.

[0009] In one possible implementation, the custom controller in the dynamic resource scheduling engine sets the initial number of replicas for the workload based on the minimum number of replicas in the configuration file, and configures Pod anti-affinity rules to achieve cross-node distribution.

[0010] In one possible implementation, the Kubernetes resource definition dynamically generated or updated by the custom controller includes a Horizontal Pod Autoscaler resource, whose scaling metrics are based on resource estimation parameters in the device configuration file or real-time performance metrics of the video processing service.

[0011] In one possible implementation, the unified access gateway supports multiple video streaming protocols and vendor platform API protocols.

[0012] In one possible implementation, the video processing service instance includes at least one of a video transcoding service, a video recording service, or an AI video analysis service.

[0013] In one possible implementation, the Kubernetes cluster comprises a pool of nodes with different types of computing resources, and the dynamic resource scheduling engine schedules workloads to the corresponding type of node pool based on the resource requirement type specified in the device configuration file.

[0014] The beneficial effects of this disclosure are as follows: The system provided by this disclosure greatly reduces the difficulty and development cost of integrating new devices / vendor platforms through standardized configuration files, achieving configuration-based integration. Based on accurate prediction using configuration files and dynamic scheduling by Kubernetes, resources (CPU / MEM / GPU) are allocated on demand, avoiding waste; HPA enables automatic elastic scaling to cope with load fluctuations. Utilizing Kubernetes' built-in high availability mechanism and configuration-based policies, high reliability and business continuity of video processing services are ensured. The platform manages resources and service lifecycles uniformly through Kubernetes, achieving a high degree of automation and reducing operational complexity. It supports the flexible addition of new device types, processing services, and resource types. It improves resource utilization and reduces over-configuration; automated operation and maintenance reduce labor costs. Attached Figure Description

[0015] Figure 1 This is a block diagram of a cloud video platform system that supports access from multiple heterogeneous video devices, as shown in an embodiment of this disclosure. Detailed Implementation

[0016] The present disclosure will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0017] Unless otherwise defined, the technical and scientific terms used in this disclosure have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains; the terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of this disclosure; the terms used in this disclosure include, and any variations thereof, are intended to cover non-exclusive inclusion. Clearly, the embodiments described in this disclosure are only a part of the embodiments of this disclosure, and not all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0018] References to embodiments in this disclosure mean that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this disclosure. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0019] Figure 1 This is a block diagram of a cloud video platform system supporting access from multiple heterogeneous video devices, as shown in an embodiment of this disclosure. Figure 1 As shown, the system includes: a device configuration file management module, a unified access gateway, a Kubernetes-based dynamic resource scheduling engine, and video processing service instances running on a Kubernetes cluster.

[0020] The device configuration file management module is used to store and manage configuration files. These files describe device / platform access attributes, media stream parameters, required processing services, and resource requirements. The device configuration files are defined using a structured data format and include at least the access protocol type, video encoding format, resolution, required video processing service type, estimated resource requirements, and minimum number of copies. For example, the system can predefine standard device configuration file templates (such as YAML or JSON). This template describes the key attributes of the device / platform, which may include: access protocol type (e.g., RTSP, ONVIF, GB28181, SDK API URL), media stream parameters (e.g., video encoding H.264 / H.265, resolution, frame rate, audio encoding, bitrate), required processing services (e.g., basic transcoding, adaptive bitrate transcoding, video recording, AI analysis type - face recognition / behavior analysis / license plate recognition), estimated resource requirements (e.g., minimum / recommended number of CPU cores, memory size, whether a GPU is required and its model), high availability requirements (e.g., minimum number of replicas, maximum fault recovery time), and vendor / device-specific parameters (e.g., ONVIFProfile, GB28181 device ID, private API key). The device configuration file management module provides a configuration management interface for administrators or device manufacturers to upload, register, and manage device configuration files.

[0021] The unified access gateway receives requests (API call requests) from different devices / platforms and matches and loads the corresponding configuration file based on the request identifier (which can be, for example, MAC, SN, or a unique ID). The unified access gateway supports multiple video streaming protocols and vendor platform API protocols, is responsible for device authentication, and can automatically match and load the corresponding configuration file based on device information. Furthermore, the system containerizes various video processing programs (such as transcoding, intelligent analysis, and recording), builds them into service images, and pushes them to a private image repository.

[0022] The Kubernetes cluster contains node pools with different types of computing resources, and the dynamic resource scheduling engine schedules workloads to the corresponding type of node pool based on the resource requirement type specified in the device configuration file.

[0023] As an example of this embodiment, the dynamic resource scheduling engine includes a configuration file parser and a custom controller; the parser parses the device configuration file into resource requirements that Kubernetes can recognize; the custom controller dynamically generates or updates Kubernetes resource definitions based on the parsed Kubernetes resource requirements, including defining the workload of the processing service, the service exposure method, and the automatic scaling strategy; the video processing service instance is created and managed by the dynamic resource scheduling engine and is used to execute the video processing tasks specified in the device configuration file.

[0024] For example, a custom controller can listen for new device access or configuration change events. Based on the required processing services and resource needs estimated in the configuration file, it dynamically generates or updates the corresponding Kubernetes resource definitions, including defining the number of Pod replicas, container images, and resource requests / limits for running video processing services (such as FFmpeg transcoders, AI inference engines, and recording services); providing stable internal access endpoints to achieve load balancing; automatically scaling the number of Pod replicas according to resource estimates or custom metrics (such as CPU utilization and video frame processing latency) in the configuration file to meet elasticity requirements; defining higher-level video processing task abstractions; and ensuring that the generated resources meet the high availability requirements in the configuration file (such as setting minReadySeconds, replicas>=2, and configuring PodAntiAffinity to distribute Pods across different Nodes).

[0025] Custom controllers are also used for resource pool management. Custom controllers store the resource pools (node ​​groups) of the defined ERP cluster, including general CPU nodes, GPU nodes (different types), high IO nodes, etc., and automatically schedule the Pod to the appropriate node according to its resource requirements (CPU, Memory, GPU Type).

[0026] The custom controller in the dynamic resource scheduling engine sets the initial number of replicas for the workload based on the minimum number of replicas in the configuration file, and configures Pod anti-affinity rules to achieve cross-node distribution.

[0027] The video processing service instance is created and managed by the dynamic resource scheduling engine, and includes at least one of video transcoding service, video recording service, or AI video analysis service, used to execute video processing tasks specified in the device configuration file (such as transcoding to a target format, performing AI analysis, and recording to storage). The processed results (video stream, analysis results, and recorded files) are distributed to downstream systems (CDN, storage, and application platform).

[0028] In one possible implementation, the system implements service instance redundancy, automatically manages Pod replicas, and ensures that the number of instances of the required service (such as transcoding or analytics) is greater than or equal to the minimum number of replicas specified in the configuration file. This allows the system to automatically restart or create the Pod on a new node in the event of any Pod failure.

[0029] In one possible implementation, the system implements load balancing, providing VIPs and load balancing for backend Pods. Clients (players, storage systems, and analytics result consumers) access services through the system frontend without needing to be aware of changes to the backend Pods.

[0030] In one possible implementation, the system enables streaming state management: for stateful services (such as recording), a combination of StatefulSet and persistent storage (PV / PVC) ensures no data loss. Services are designed to support recovery from checkpoints or rapid task takeover.

[0031] In one possible implementation, the system is capable of performing health checks: configuring Liveness Probe and Readiness Probe for the video processing service container to ensure that the system can accurately perceive the service status and perform recovery.

[0032] Compared with the prior art, this disclosure has the following beneficial effects:

[0033] Simplified access: Standardized configuration files greatly reduce the difficulty and development cost of accessing new devices / vendor platforms, enabling access as soon as configuration is completed.

[0034] Resource efficiency: Based on accurate prediction of configuration files and dynamic scheduling of Kubernetes, resources (CPU / MEM / GPU) are allocated on demand to avoid waste; HPA enables automatic elastic scaling to cope with load fluctuations.

[0035] High availability assurance: Utilizing Kubernetes' built-in high availability mechanisms (replicas, self-healing, load balancing) and configuration-based policies, we ensure high reliability and business continuity for video processing services.

[0036] Convenient operation and maintenance: The platform manages resources and service lifecycles through Kubernetes, with a high degree of automation, reducing the complexity of operation and maintenance.

[0037] Highly scalable: Supports flexible addition of new device types, processing services, and resource types (such as new GPU models).

[0038] Reduce costs: Improve resource utilization and reduce over-configuration; automated operation and maintenance reduce labor costs.

[0039] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A cloud video platform system for supporting multi-source heterogeneous video device access, characterized in that, The system includes: a device configuration file management module; The device configuration file management module is used to store and manage configuration files, which describe device / platform access attributes, media stream parameters, required processing services, and resource requirements. The unified access gateway is used to receive device / platform requests and load the corresponding configuration file based on the request identifier. The dynamic resource scheduling engine based on Kubernetes includes a configuration file parser and a custom controller. The parser parses the device configuration file into Kubernetes resource requirements. The custom controller dynamically generates or updates Kubernetes resource definitions based on the parsed Kubernetes resource requirements, including defining the workload of the processing service, the service exposure method, and the automatic scaling strategy. The video processing service instance running on the Kubernetes cluster is created and managed by the dynamic resource scheduling engine and is used to execute the video processing tasks specified in the device configuration file.

2. The system of claim 1, wherein, The device configuration file is defined using a structured data format and includes at least the access protocol type, video encoding format, resolution, required video processing service type, estimated resource requirements, and minimum number of replicas.

3. The system of claim 1 or 2, wherein, The custom controller in the dynamic resource scheduling engine sets the initial number of replicas for the workload based on the minimum number of replicas in the configuration file, and configures Pod anti-affinity rules to achieve cross-node distribution.

4. The system of claim 1, wherein, The Kubernetes resource definitions dynamically generated or updated by the custom controller include Horizontal Pod Autoscaler resources, whose scaling metrics are based on resource estimation parameters in the device configuration file or real-time performance metrics of the video processing service.

5. The system of claim 1, wherein, The unified access gateway supports multiple video streaming protocols and vendor platform API protocols.

6. The system of claim 1, wherein, The video processing service examples include at least one of video transcoding service, video recording service, or AI video analysis service.

7. The system of claim 1, wherein, The Kubernetes cluster contains node pools with different types of computing resources, and the dynamic resource scheduling engine schedules workloads to the corresponding type of node pool based on the resource requirement type specified in the device configuration file.

Images