A video traffic adaptive scheduling system

Abstract

The application discloses a video flow adaptive scheduling system, and belongs to the technical field of rail transit video monitoring. The video flow adaptive scheduling system comprises a center-level device and a station-level device which are connected through a backbone transmission network. The center-level device comprises a network management device, a storage device, a center server, a core switch device, an equal protection device, an AI intelligent scheduling server and a dispatching station. The AI intelligent scheduling server is used for deployment of an AI large model and generation of an intelligent scheduling strategy. The disclosed basic AI large model technology summarizes and analyzes tidal flow in a daily subway operation process, formulates an intelligent video scheduling strategy according to important scenes such as concentrated large passenger flow and emergency communication command that may be faced in daily subway operation, realizes dynamic adjustment of video resources, and thus provides stable, clear and reliable live video images.

Description

Technical Field

[0001] This application belongs to the field of rail transit video surveillance technology, and in particular relates to a video traffic adaptive scheduling system. Background Technology

[0002] Video surveillance systems are an indispensable communication subsystem in the daily operation of urban rail transit. Like the "visual nerves" of the rail transit system, they use high-definition cameras distributed throughout stations and trains to achieve 24 / 7, comprehensive monitoring and recording of the rail transit operating environment. Furthermore, in the event of abnormal situations or malfunctions in subway operations, the operating unit can quickly grasp the specific situation on-site through the video surveillance system and rapidly deploy relevant work.

[0003] Currently, all major cities in China's urban rail transit video surveillance systems adopt a fully digital high-definition system solution. Fully digital high-definition network cameras are installed in station concourses, platforms, public areas, equipment areas, and entrance / exit passages for station monitoring. Network storage devices are installed in stations or control centers for centralized storage of video images. Video surveillance terminals are typically installed in station control rooms and control center dispatch halls to enable operators to monitor stations in real time and retrieve video recordings. Due to the increasing demands for video images in the public safety field, the number of cameras in subway stations is rapidly increasing. Furthermore, with the widespread application of video analysis technologies such as facial recognition and intrusion detection, the performance requirements of the video surveillance system itself are constantly increasing to ensure clear and usable video stream quality. This is especially true in emergency command scenarios, which often involve the retrieval and switching of numerous real-time video streams from different stations. To prevent stuttering or even black screens during high-number video stream retrieval, the performance requirements of the video surveillance system itself are continuously rising.

[0004] Currently, conventional video surveillance systems face the following problems in video traffic transmission and viewing: (1) Rigid resource allocation: Traditional video transmission bandwidth is usually statically allocated. However, passenger flow has obvious "tidal" characteristics (more people during morning and evening peak hours, fewer people during off-peak hours). During morning and evening peak hours, due to passenger congestion or slow movement speed, higher definition or more real-time images are needed to monitor safety, but the existing system cannot automatically adjust the transmission strategy according to historical patterns, resulting in bandwidth congestion during peak hours and bandwidth waste during off-peak hours. (2) Weak ability to handle sudden concurrent events: When a sudden security event occurs (such as fire or large passenger flow congestion), the control center or personnel from multiple departments will simultaneously view the video of the same key area, instantly generating high concurrent traffic. The existing system often lacks the concept of "priority", resulting in key video lag or even service crash, which cannot guarantee the needs of emergency command. (3) Poor environmental adaptability: Video front-end equipment (cameras) or transmission networks inevitably experience equipment aging, overheating, or network jitter (packet loss, delay). When existing systems detect poor network conditions, they often disconnect directly or reconnect frequently, lacking a "degradation and survival" mechanism, which causes critical locations to completely lose their images when the network fluctuates.

[0005] In response to the ever-increasing performance requirements of video surveillance systems, the current widespread approach is to continuously enhance the hardware capabilities of the video surveillance system itself to improve its performance. This includes using higher-performance video servers to improve video stream processing capabilities and using higher-capacity storage devices to store higher-resolution video streams. As a result, the construction cost of video surveillance systems is becoming increasingly higher.

[0006] Therefore, it is necessary to provide a new video traffic adaptive scheduling system to solve the above-mentioned technical problems. Summary of the Invention

[0007] The purpose of this disclosure is to provide a video traffic adaptive scheduling system to solve the above-mentioned problems.

[0008] This disclosure achieves the above objectives through the following technical solutions: A video traffic adaptive scheduling system includes central-level equipment and station-level equipment connected via a backbone transmission network; The central-level equipment includes network management equipment, storage equipment, a central server, core switching equipment, security protection equipment, an AI intelligent scheduling server, and a scheduling console. The network management equipment is used for online management and centralized monitoring of the video surveillance system. The storage equipment is used for saving and recording video images. The central server hosts the video surveillance platform. The core switching equipment forms the video network. The security protection equipment provides security protection for the video surveillance system. The scheduling console is used for real-time viewing of video images and playback of recorded videos. The AI ​​intelligent scheduling server is used for deploying large-scale AI models and generating intelligent scheduling strategies. The station-level equipment includes multiple camera terminals.

[0009] As a further optimization of this disclosure, the AI ​​intelligent scheduling server includes an intelligent prediction and mode switching module, a global resource preemption scheduling module, and a device status perception and keep-alive module; The intelligent prediction and mode switching module is used to process historical passenger flow and traffic data to construct a tidal model. The global resource preemption and scheduling module is used to monitor concurrent requests in real time and handle bandwidth allocation under sudden events; The device status awareness and keep-alive module is used to monitor the health of devices and networks and execute degradation strategies.

[0010] As a further optimization of this disclosure, the intelligent prediction and mode switching module periodically retrieves historical passenger flow and traffic data, automatically defines the time thresholds for peak and off-peak periods, and sets two different strategy modes, peak period mode and off-peak period mode, to achieve smooth switching of the video process.

[0011] As a further optimization of this disclosure, the intelligent prediction and mode switching module prioritizes image quality details in peak mode by setting high bitrate, high frame rate, and low compression ratio, and reserves high bandwidth through QoS policies; in off-peak mode, it releases redundant bandwidth for other services or storage by setting standard bitrate, standard frame rate, and high compression ratio.

[0012] As a further optimization of this disclosure, the intelligent prediction and mode switching module does not use a forced interruption and restart method at the mode switching critical point, but instead achieves a seamless and smooth transition of the video stream by dynamically adjusting the frame interval and encoding parameters, thus avoiding black screens or abrupt changes in the picture.

[0013] As a further optimization of this disclosure, when the global resource preemption scheduling module detects a sudden high-concurrency viewing request, it automatically triggers a preemption mechanism, including marking the video stream of the key involved area as a critical resource with a priority of Level S; and marking other ordinary monitoring areas as non-critical resources with a priority of Level C.

[0014] As a further optimization of this disclosure, when the number of concurrent viewing channels of a certain video or a certain area exceeds a preset threshold, or when an external alarm signal is received, the global resource preemption scheduling module automatically reduces the bitrate or frame rate of Level C video, and even temporarily suspends the preview stream of non-priority areas and forcibly reclaims bandwidth when there is extreme bandwidth shortage; all the reclaimed bandwidth resources are injected into Level S video, and the acceleration channel of the multicast or distribution server is opened to ensure that even if multiple people watch at the same time, the video stream is still clear, smooth and has low latency.

[0015] As a further optimization of this disclosure, the device and network health status includes the device's CPU utilization, memory usage, and the network link's packet loss rate, latency, and jitter.

[0016] As a further optimization of this disclosure, the device state awareness and keep-alive module executes a degradation strategy, including: If slight network fluctuations occur, the scalable video encoding frame dropping strategy will be activated, prioritizing the discarding of enhancement layer data and retaining base layer data to ensure continuous and uninterrupted video playback. If moderate fluctuations occur, the system will automatically switch to high-speed mode. By reducing the resolution and frame rate, the system will maintain the image online with a very low data volume, ensuring that the control center can see the dynamic outline of the scene. If a device malfunctions, it will automatically issue a command to shut down unnecessary functions, reduce the device load, and ensure video stream output.

[0017] As a further optimization of this disclosure, minor network fluctuations include packet loss rates less than a preset value; moderate fluctuations include insufficient bandwidth; device malfunctions include overheating of the encoding chip; and non-essential functions include intelligent analysis functions.

[0018] The beneficial effects of this disclosure are as follows: This invention utilizes AI big data modeling technology to analyze the traffic patterns of daily video surveillance systems and formulate different video traffic allocation strategies for special scenarios that may occur during daily subway operations, thereby achieving intelligent scheduling of video resources. Through AI big data modeling technology, the hardware resources of existing video surveillance systems can be allocated in real time, eliminating the need to blindly add hardware to improve the performance of the video platform and enabling stable viewing of video images in any scenario. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the overall system structure in an embodiment of this disclosure. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] like Figure 1 As shown, a video traffic adaptive scheduling system includes central-level equipment and station-level equipment connected through a backbone transmission network; Central-level equipment includes network management equipment, storage equipment, central servers, core switching equipment, information security compliance equipment, AI intelligent dispatch servers, dispatch consoles, and other equipment. The network management equipment mainly enables online management and centralized monitoring of the video surveillance system; Storage devices enable the saving and recording of video images; The central server is responsible for hosting the video surveillance platform; Core switching equipment is responsible for the composition of the video network; The equipment covered by the information security classification system is responsible for the security protection of the video surveillance system; The dispatch console enables real-time viewing and playback of video drawings; The AI ​​intelligent scheduling server is responsible for deploying large AI models and generating intelligent scheduling strategies.

[0023] The AI ​​intelligent scheduling server mainly consists of three modules: intelligent prediction and mode switching module, global resource preemption and scheduling module, and device status perception and keep-alive module.

[0024] Among them, the intelligent prediction and mode switching module is responsible for processing historical passenger flow and traffic data and building a tidal model; The global resource preemption and scheduling module is responsible for real-time monitoring of concurrent requests and handling bandwidth allocation under sudden events; The device status awareness and keep-alive module is responsible for monitoring the health of devices and networks and executing degradation strategies.

[0025] The equipment at the station level is similar in structure to that of the control center, but it also includes various camera terminals to achieve real-time video image acquisition. Through a video resource scheduling system based on an AI-powered big data model, intelligent scheduling of video resources can be achieved in the following aspects, ensuring clear, stable, and reliable video images: (1) Based on the fluctuations in video data traffic caused by historical data tidal passenger flow, the system automatically constructs "morning / evening peak mode" and "off-peak mode" to achieve smooth switching of video traffic; Currently, most urban rail transit video surveillance systems use front-end cameras of the same brand. The camera specifications, video recording bitrate, and video platform storage bitrate are all fixed. At any given time, the video system allocates resources evenly to cameras at different locations. The video system cannot adjust its video resource allocation based on passenger flow.

[0026] This system incorporates time-series analysis algorithms to train on historical passenger flow data and historical video bandwidth consumption data, identifying peak and off-peak time windows for each station. The system periodically retrieves historical passenger flow data, automatically defining peak periods (e.g., 07:30-09:30) and off-peak time thresholds. It sets two different strategy modes: peak and off-peak, to achieve smooth video stream transitions. In peak mode, high bitrate, high frame rate, and low compression ratio are used to prioritize image quality details, while QoS policies reserve high bandwidth. In off-peak mode, standard bitrate, standard frame rate, and high compression ratio are used to free up redundant bandwidth for other services or storage. Furthermore, at mode switching thresholds (e.g., transitioning from off-peak to peak), the system does not use forced disconnection and restart; instead, it dynamically adjusts frame intervals and encoding parameters to achieve a seamless and smooth transition of the video stream, avoiding black screens or abrupt changes. This system can adapt to the tidal patterns of passenger flow, thereby maximizing the utilization of network resources, ensuring monitoring quality during peak periods while saving storage and transmission costs during off-peak periods.

[0027] (2) When faced with a sudden situation or extreme passenger flow, the system automatically implements "global resource preemption scheduling" to ensure the clarity and reliability of key video data; When simultaneously viewing videos from key areas, the high concurrent traffic can cause buffering and stuttering. This system addresses this by establishing a "priority pyramid" and a "dynamic bandwidth pool" for the video streams. When a sudden surge in high-concurrency viewing requests is detected, a preemption mechanism is automatically triggered. For example, video streams from key affected areas are marked as "critical resources (Level S)," while other ordinary monitored areas are marked as "non-critical resources (Level C)." When the number of concurrent viewings of a particular video stream or area exceeds a certain threshold (e.g., N people watching simultaneously), or when an external alarm signal is received (e.g., fire alarm linkage), the system automatically reduces the bitrate or frame rate of Level C videos. In extreme cases of insufficient bandwidth, it may even temporarily suspend preview streams from non-critical areas, forcibly reclaiming bandwidth. All reclaimed bandwidth is injected into Level S videos, and multicast or distribution server acceleration channels are activated to ensure that even with multiple simultaneous viewers, the video stream remains clear, smooth, and with low latency. In the event of an emergency, this system can preserve core resources by sacrificing non-essential resources, solving the problem of traditional systems "overwhelming all resources and causing complete paralysis" and greatly improving emergency command efficiency.

[0028] (3) In the event of equipment failure or network fluctuations, the system automatically implements "downgrade and keep alive" scheduling to ensure the stable online operation of cameras in critical areas.

[0029] This system can monitor the health status of cameras (such as CPU utilization and memory usage) and network link quality (packet loss rate, latency, jitter, etc.) in real time. It monitors device and network status at millisecond-level frequencies via protocols such as SNMP or private heartbeat protocols. In case of minor network fluctuations (e.g., packet loss rate <5%), a scalable video encoding frame dropping strategy is activated, prioritizing the discarding of enhancement layer data while retaining base layer data. Although image quality decreases slightly, continuous, uninterrupted video playback is guaranteed. In case of moderate fluctuations (e.g., severe bandwidth shortage), the system automatically switches to high-speed mode, reducing resolution and frame rate to maintain the video feed with extremely low data volume, ensuring the control center can see the dynamic outlines of the scene. In case of device malfunction (e.g., encoding chip overheating), the system automatically issues commands to disable unnecessary functions (such as intelligent analysis), reducing device load and ensuring optimal video stream output. Even in poor network environments or when devices are in a sub-optimal state, this system can maintain video feed through degradation strategies, avoiding blind spots and improving overall system stability.

[0030] The embodiments described above are merely examples of several implementations of this disclosure, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent disclosure. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this disclosure, and these modifications and improvements all fall within the protection scope of this disclosure.

Claims

1. A video traffic adaptive scheduling system, characterized in that, This includes central-level equipment and station-level equipment connected via the backbone transmission network; The central-level equipment includes network management equipment, storage equipment, a central server, core switching equipment, security protection equipment, an AI intelligent scheduling server, and a scheduling console. The network management equipment is used for online management and centralized monitoring of the video surveillance system. The storage equipment is used for saving and recording video images. The central server hosts the video surveillance platform. The core switching equipment forms the video network. The security protection equipment provides security protection for the video surveillance system. The scheduling console is used for real-time viewing of video images and playback of recorded videos. The AI ​​intelligent scheduling server is used for deploying large-scale AI models and generating intelligent scheduling strategies. The station-level equipment includes multiple camera terminals.

2. The video traffic adaptive scheduling system according to claim 1, characterized in that, The AI ​​intelligent scheduling server includes an intelligent prediction and mode switching module, a global resource preemption and scheduling module, and a device status perception and keep-alive module. The intelligent prediction and mode switching module is used to process historical passenger flow and traffic data to construct a tidal model. The global resource preemption and scheduling module is used to monitor concurrent requests in real time and handle bandwidth allocation under sudden events; The device status awareness and keep-alive module is used to monitor the health of devices and networks and execute degradation strategies.

3. The video traffic adaptive scheduling system according to claim 2, characterized in that, The intelligent prediction and mode switching module periodically retrieves historical passenger flow and traffic data, automatically defines the time thresholds for peak and off-peak periods, and sets two different strategy modes, peak period mode and off-peak period mode, to achieve smooth switching of the video process.

4. The video traffic adaptive scheduling system according to claim 3, characterized in that, In peak-hour mode, the intelligent prediction and mode switching module prioritizes image quality details by setting high bitrate, high frame rate, and low compression ratio, and reserves high bandwidth through QoS policies. In off-peak-hour mode, it releases redundant bandwidth for other services or storage by setting standard bitrate, standard frame rate, and high compression ratio.

5. A video traffic adaptive scheduling system according to claim 3, characterized in that, The intelligent prediction and mode switching module does not use a forced interruption and restart method at the mode switching critical point. Instead, it dynamically adjusts the frame interval and encoding parameters to achieve a seamless and smooth transition of the video stream, avoiding black screens or abrupt changes.

6. The video traffic adaptive scheduling system according to claim 1, characterized in that, When the global resource preemption scheduling module detects a sudden high-concurrency viewing request, it automatically triggers the preemption mechanism, including marking the video stream of the key involved area as a critical resource with a priority of Level S; Other ordinary monitoring areas are marked as non-critical resources with a priority of Level C.

7. A video traffic adaptive scheduling system according to claim 6, characterized in that, When the number of concurrent viewing channels for a certain video stream or a certain area exceeds a preset threshold, or when an external alarm signal is received, the global resource preemption scheduling module automatically reduces the bitrate or frame rate of Level C video. In extreme cases of insufficient bandwidth, it may even temporarily suspend the preview stream of non-priority areas and forcibly reclaim bandwidth. All reclaimed bandwidth resources are injected into Level S video and the acceleration channel of the multicast or distribution server is enabled to ensure that the video stream remains clear, smooth, and has low latency even when multiple people are watching at the same time.

8. A video traffic adaptive scheduling system according to claim 1, characterized in that, The device and network health status includes the device's CPU utilization, memory usage, and the network link's packet loss rate, latency, and jitter.

9. A video traffic adaptive scheduling system according to claim 1, characterized in that, The device status awareness and keep-alive module executes a degradation strategy, including: If slight network fluctuations occur, the scalable video encoding frame dropping strategy will be activated, prioritizing the discarding of enhancement layer data and retaining base layer data to ensure continuous and uninterrupted video playback. If moderate fluctuations occur, the system will automatically switch to high-speed mode. By reducing the resolution and frame rate, the system will maintain the image online with a very low data volume, ensuring that the control center can see the dynamic outline of the scene. If a device malfunctions, it will automatically issue a command to shut down unnecessary functions, reduce the device load, and ensure video stream output.

10. A video traffic adaptive scheduling system according to claim 1, characterized in that, Minor network fluctuations include packet loss rates below the preset value; moderate fluctuations include insufficient bandwidth; device malfunctions include overheating of the encoding chip; and non-essential functions include intelligent analysis functions.

Images