Calculation network cooperative scheduling method and device

Abstract

The invention provides a computing network cooperative scheduling method and device, and relates to the technical field of data processing, and the method comprises the steps: obtaining computing power resources, network resources and business information containing business priorities in a rail transit environment; based on the computing power resource, the network resource and the service priority, calculating to obtain a comprehensive scheduling score, and determining a target scheduling mode according to the comprehensive scheduling score; and according to the target scheduling mode, cooperatively allocating the computing power resources and the network resources to process the service information.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a computing network collaborative scheduling method and device. Background Art

[0002] With the intelligent upgrade of rail transit, services such as onboard video and hazard identification have surged, with the daily data volume of a single train exceeding 2TB, and the services exhibiting a significant tidal effect. At the same time, business-to-business (B2B) control services and consumer-to-consumer (B2C) services are being carried in a mixed manner on the same 5G network, placing differentiated demands on low latency, high reliability, and large bandwidth. The existing centralized processing model results in a significant amount of idle onboard computing power, excessively high end-to-end latency, and static QoS policies that cannot be dynamically adjusted based on temporal and spatial changes. This results in low resource utilization and makes critical services susceptible to interference. Therefore, how to more effectively coordinate computing and network scheduling has become an urgent issue for the industry. Summary of the Invention

[0003] The present invention provides a computing network collaborative scheduling method and device, which are used to solve the problem of how to more effectively perform computing network collaborative scheduling in the prior art.

[0004] The present invention provides a computing network collaborative scheduling method, comprising: Obtain computing resources, network resources, and business information including business priorities in the rail transit environment; Calculating a comprehensive scheduling score based on the computing resources, the network resources, and the service priority, and determining a target scheduling mode according to the comprehensive scheduling score; According to the target scheduling mode, the computing resources and the network resources are collaboratively allocated to process the business information.

[0005] According to a computing network collaborative scheduling method provided by the present invention, determining a target scheduling mode according to the comprehensive scheduling score includes: When the comprehensive scheduling score is greater than a first preset threshold, determining that the target scheduling mode is a high-fidelity mode; wherein the high-fidelity mode is used to allocate dedicated network slice resources for the service corresponding to the service information; When the comprehensive scheduling score is less than a second preset threshold, the target scheduling mode is determined to be a cache mode, wherein the cache mode is used to temporarily store data corresponding to the service information in the vehicle terminal.

[0006] According to a computing network collaborative scheduling method provided by the present invention, determining a target scheduling mode according to the comprehensive scheduling score includes: When the comprehensive scheduling score is between the first preset threshold and the second preset threshold, determining that the target scheduling mode is a balanced mode; Among them, the balancing mode is used to compress the data corresponding to the business information through the vehicle terminal, and perform quality enhancement processing on the compressed data at the edge computing node.

[0007] According to a computing network collaborative scheduling method provided by the present invention, the method further includes: Based on the spatiotemporal characteristic data of the rail transit environment, identifying the operating mode of the rail transit environment, wherein the operating mode includes a peak mode, an off-peak mode, or an off-peak mode; The computing resources and the network resources are collaboratively allocated according to the target scheduling mode and the operation mode.

[0008] According to the present invention, a computing-network collaborative scheduling method is provided, which collaboratively allocates the computing power resources and the network resources according to the target scheduling mode and the operation mode, including: When the operating mode is peak mode and the target scheduling mode is high-fidelity mode, dedicated network slice resources are allocated to the service corresponding to the service information in the available network resources outside the hard-isolated network channel reserved for the preset highest priority service.

[0009] According to the present invention, a computing-network collaborative scheduling method is provided, which collaboratively allocates the computing power resources and the network resources according to the target scheduling mode and the operation mode, including: When the operation mode is the idle mode, a large-bandwidth network channel is allocated for executing the target scheduling mode.

[0010] The present invention also provides a computing network collaborative scheduling device, comprising the following modules: The acquisition module is used to obtain computing resources, network resources, and business information including business priorities in the rail transit environment; a determination module, configured to calculate a comprehensive scheduling score based on the computing resources, the network resources, and the service priority, and determine a target scheduling mode according to the comprehensive scheduling score; An allocation module is used to collaboratively allocate the computing resources and the network resources to process the business information according to the target scheduling mode.

[0011] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: When the comprehensive scheduling score is greater than a first preset threshold, determining that the target scheduling mode is a high-fidelity mode; wherein the high-fidelity mode is used to allocate dedicated network slice resources for the service corresponding to the service information; When the comprehensive scheduling score is less than a second preset threshold, the target scheduling mode is determined to be a cache mode, wherein the cache mode is used to temporarily store data corresponding to the service information in the vehicle terminal.

[0012] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: When the comprehensive scheduling score is between the first preset threshold and the second preset threshold, determining that the target scheduling mode is a balanced mode; Among them, the balancing mode is used to compress the data corresponding to the business information through the vehicle terminal, and perform quality enhancement processing on the compressed data at the edge computing node.

[0013] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: Based on the spatiotemporal characteristic data of the rail transit environment, identifying the operating mode of the rail transit environment, wherein the operating mode includes a peak mode, an off-peak mode, or an off-peak mode; The computing resources and the network resources are collaboratively allocated according to the target scheduling mode and the operation mode.

[0014] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: When the operating mode is peak mode and the target scheduling mode is high-fidelity mode, dedicated network slice resources are allocated to the service corresponding to the service information in the available network resources outside the hard-isolated network channel reserved for the preset highest priority service.

[0015] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: When the operation mode is the idle mode, a large-bandwidth network channel is allocated for executing the target scheduling mode.

[0016] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the computing network collaborative scheduling method as described above is implemented.

[0017] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described computing network collaborative scheduling methods.

[0018] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-described computing network collaborative scheduling methods.

[0019] The computing network collaborative scheduling method and device provided by the present invention continuously collects computing power resources, network resources and business information carrying business priorities in the rail transit environment within the same control plane, and establishes a comprehensive scheduling scoring model with multi-dimensional real-time data as input; based on the dynamic changes of the score, the system instantly selects the adaptive target scheduling mode without manual intervention, and collaboratively allocates and arranges computing power and network resources accordingly. This process transforms the originally static and fragmented resource configuration into a closed-loop adaptive mechanism, allowing computing power to flow on demand between on-board terminals, trackside edges and central clouds, and network bandwidth to converge or release in real time with the urgency of the business. Therefore, under the premise of keeping the total amount of physical resources unchanged, it simultaneously solves existing problems such as high idle rate of on-board computing power, excessive end-to-end latency, and mutual crowding out of critical and non-critical businesses, providing rail transit with intelligent scheduling capabilities with high reliability, low latency and significantly improved resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 This is a flow chart of the computing network collaborative scheduling method provided by the present invention; Figure 2 A schematic diagram of the overall structure provided by the present invention; Figure 3 This is a schematic diagram of the structure of the computing network collaborative scheduling device provided by the present invention; Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0022] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0023] Figure 1 This is a flow chart of the computing network collaborative scheduling method provided by the present invention. Figure 1 As shown, the method includes the following: Step 110: Obtain computing resources, network resources, and service information including service priorities in the rail transit environment; In the present invention, the rail transit environment is centered on high-speed train operation along fixed tracks, covering continuous movement scenarios including stations, tunnels, vehicle depots and control centers. In this scenario, heterogeneous services such as train control, video surveillance and passenger services run simultaneously, generating both millisecond-level low-latency, ultra-high-reliability key instruction streams and high-definition video streams. Passenger flow fluctuates several times between peak and off-peak hours in the morning and evening, resulting in the multi-level computing power of on-board terminals, trackside edge nodes and central clouds, as well as 5G private networks, public networks and wired transmission network resources being highly dynamic, with large load disparities and significant priority differences in the time and space dimensions.

[0024] Computing resources in rail transit environments refer to the real-time available capabilities of three-tier computing facilities distributed across onboard terminals, trackside edge nodes, and the central cloud during train operation. Onboard terminals periodically collect information about the instantaneous utilization of their CPUs, GPUs, and NPUs, remaining memory, battery charge, and remaining capacity and read / write IOPS of their solid-state drives through operating system interfaces, and then encapsulate this data into lightweight messages for reporting. Trackside edge nodes also record their own server node's CPU / GPU usage, GPU temperature, number of allocatable container instances, and cache capacity in a containerized manner; the central cloud aggregates the real-time load of virtual machines, bare metal, and accelerator cards to form a global computing power pool view.

[0025] Network resources cover the real-time performance indicators of 5G base stations, transmission networks, and core networks along the train line: the base station side continuously outputs reference signal received power, signal-to-noise-interference ratio, PDCP layer packet loss rate, and available uplink and downlink throughput based on 5G NR measurement reports; the transmission network reads link latency, FlexE hard slice remaining bandwidth, queue length, and congestion mark through the SRv6 controller; the core network side records end-to-end latency, jitter, packet loss, and available bandwidth, and all indicators enter the computing network perception plane as streaming data.

[0026] Business information containing business priorities refers to the business demand description issued by the management platform or the in-vehicle application, and its main body is a three-level priority classification: emergency services require end-to-end delays of no more than two hundred milliseconds and are assigned the highest priority, critical services allow delays of less than two seconds and are given medium priority, and ordinary services can be completed within five minutes or even longer and correspond to the lowest priority; this information also carries the business type identifier, the required data fidelity level, the single data volume, the return cycle and the maximum tolerable interruption duration. After being encapsulated by the RESTful API or a dedicated protocol, it is parsed by the computing network business perception engine and bound to the corresponding business flow, thereby forming the complete input required for step 110 together with the computing power and network status.

[0027] Step 120: Calculate a comprehensive scheduling score based on the computing resources, the network resources, and the service priority, and determine a target scheduling mode according to the comprehensive scheduling score; In the present invention, computing resources, network resources, and business information carrying business priorities are input into a uniformly constructed dynamic weighted scoring model.

[0028] The dynamic weighted scoring model uses network quality, computing load, and business priority as factors, mapping each of them into sub-scores Nscore, Cscore, and Pscore in the range of 0–100, and linearly superimposing them according to the preset weights of 0.6, 0.3, and 0.1 to generate a comprehensive scheduling score S of 0–100.

[0029] The scoring results are refreshed in real time, and the target scheduling mode is determined based on the comprehensive scheduling score: when the comprehensive scheduling score S≥80, the high-fidelity mode is triggered to allocate a dedicated network slice for the service; when the comprehensive scheduling score is 60≤S<80, the balanced mode is enabled to achieve data compression and super-resolution reconstruction; when the comprehensive scheduling score S<60, it switches to cache mode, temporarily stores the data on the on-board SSD and transmits it back when the network is idle.

[0030] The dynamic weighted scoring model uses three indicators, network quality, computing power margin, and service priority, collected in real time from rail transit as input. It converts RSRP, SINR, and packet loss rate into network scores, maps CPU, GPU, and memory usage into computing power scores, and directly assigns scores to services according to T1, T2, and T3. These scores are then linearly superimposed with fixed weights of 0.6, 0.3, and 0.1 to form a comprehensive score S between 0 and 100. The score is refreshed once a second based on a sliding window average, and the threshold is adaptively adjusted when the network load is below or above a specific threshold for a long time. This allows the system to always use a unique, quantifiable S value to instantly decide on the subsequent scheduling strategies of real-time fidelity, compression enhancement, or cache delay under conditions of high-speed movement and tidal passenger flow, thereby achieving coordinated optimization of computing power and network resources.

[0031] Step 130: Coordinately allocate the computing resources and the network resources to process the business information according to the target scheduling mode.

[0032] In the present invention, if the mode is high-fidelity mode, the scheduling center first calls the network management engine to reserve a FlexE hard slice for the service along the SRv6 path. The bandwidth and time slot are locked at the same time at the base station, transmission node and core network along the way to ensure that the end-to-end delay is locked within 10ms; at the same time, the computing power offloading engine maps the inference or transcoding task directly to the edge GPU instance closest to the user, and the container image is sent down through KubeEdge in seconds. After the task completes all calculations at the edge, only the results are returned, eliminating the round-trip delay in the cloud.

[0033] If the mode is balanced, the system first starts a lightweight encoding container on the onboard terminal NPU, compresses the original video at 50% of the target bit rate, and sends the compressed code stream to the trackside MEC via the existing 5G bearer; the MEC runs a super-resolution reconstruction model on the GPU in the same cabinet, restores the 720p stream to 1080p visual quality, and then sends it to the central platform via a normal backhaul link. During the entire process, the network occupancy is halved while the image quality is intact.

[0034] If the mode is cache mode, the computing power offloading engine will immediately write the data to be transmitted into the on-board SSD array, and at the same time send a "suspend scheduling" instruction to the network management engine to release the air interface bandwidth; when the train enters the terminal and the network utilization rate is lower than 40% for five consecutive minutes, the dispatching center will trigger the batch backhaul task again, and use the 200% peak bandwidth channel during off-peak hours to upload the cached data in full, ensuring data integrity and optimal filling of network resources.

[0035] In the present invention, by continuously collecting computing resources, network resources and business information carrying business priorities in the rail transit environment within the same control plane, a comprehensive scheduling scoring model is established with multi-dimensional real-time data as input; based on the dynamic changes of the score, the system instantly selects the appropriate target scheduling mode without manual intervention, and coordinates and arranges computing power and network resources accordingly. This process transforms the originally static and fragmented resource configuration into a closed-loop adaptive mechanism, allowing computing power to flow on demand between on-board terminals, trackside edges and central clouds, and network bandwidth to converge or release in real time with the urgency of the business. Under the premise of keeping the total amount of physical resources unchanged, it simultaneously solves existing problems such as high idle rate of on-board computing power, excessive end-to-end latency, and mutual crowding out of critical and non-critical businesses, providing rail transit with intelligent scheduling capabilities with high reliability, low latency and significantly improved resource utilization.

[0036] Optionally, determining a target scheduling mode according to the comprehensive scheduling score includes: When the comprehensive scheduling score is greater than a first preset threshold, determining that the target scheduling mode is a high-fidelity mode; wherein the high-fidelity mode is used to allocate dedicated network slice resources for the service corresponding to the service information; When the comprehensive scheduling score is less than a second preset threshold, the target scheduling mode is determined to be a cache mode, wherein the cache mode is used to temporarily store data corresponding to the service information in the vehicle terminal.

[0037] In the present invention, in the rail transit operation scenario, when the comprehensive scheduling score is higher than the first preset threshold, the system determines that the wireless coverage, trackside edge nodes and on-board terminals in the section where the current train is located are all in a state of abundant available resources, and then starts the high-fidelity mode.

[0038] In high-fidelity mode, the network side collaborates with the bearer network through train-specific 5G base stations to establish an end-to-end hard-isolated network slice for train control or driver behavior monitoring services. This slice uses exclusive FlexE time slots on the base station side and hop-by-hop locking through SRv6 Segment Routing on the transmission network side. This ensures that the train-to-ground communication latency is always less than 10 milliseconds and the jitter does not exceed 2 milliseconds during the entire train's movement at speeds of 80 to 120 kilometers per hour. At the same time, the on-board computing unit offloads the corresponding video analysis tasks to the trackside MEC server at the nearest station, using its GPU resources to complete real-time inference and only transmits the recognition results back to the control center, avoiding the delay caused by the long-distance round-trip to the cloud and meeting the time limit requirements for train emergency braking or abnormal driver behavior alerts.

[0039] If the comprehensive dispatch score is lower than the second preset threshold, the system recognizes that the current train is entering a tunnel or a base station overlapping coverage weak area, and the network instantaneous capacity is insufficient, so it switches to cache mode.

[0040] In cache mode, the onboard gateway immediately writes the high-definition video data collected by the onboard camera in the form of file blocks to the NVMe solid-state storage array in the train cabinet, and writes the train number, car number and UTC timestamp in the file index to ensure that the data corresponds one-to-one with the vehicle's running trajectory; at the same time, the network side releases the air interface resources occupied by this service, giving priority to the CBTC signal system and emergency call services.

[0041] When the train reaches the depot or terminal and the network utilization rate is less than 40% for five consecutive minutes, the system uses the dedicated high-bandwidth channel of the depot's nighttime maintenance window to transfer the cached data back to the central cloud in batches. After completing the data integrity check, the solid-state storage array automatically reclaims the space and prepares for the next round of operation cycle.

[0042] In the present invention, during train operation, high-priority services always enjoy dedicated network slices, low-priority services are automatically cached and transmitted back in batches during idle time, the utilization rate of on-board computing power is improved, the network peak bandwidth demand is reduced, and the reliability of security services is ensured.

[0043] Optionally, determining a target scheduling mode according to the comprehensive scheduling score includes: When the comprehensive scheduling score is between the first preset threshold and the second preset threshold, determining that the target scheduling mode is a balanced mode; Among them, the balancing mode is used to compress the data corresponding to the business information through the vehicle terminal, and perform quality enhancement processing on the compressed data at the edge computing node.

[0044] When the comprehensive scheduling score falls between the first and second preset thresholds, the system determines that the current network and computing power are in an available but tight state, and then enables the balancing mode.

[0045] In balanced mode, the onboard terminal immediately invokes the NPU to perform lightweight encoding on the original video, compressing the bitrate by approximately 50% and encapsulating it into a low-latency frame stream. This low-latency frame stream is transmitted via the 5G train-to-ground link to the trackside MEC at the nearest station. The GPU within the MEC completes super-resolution reconstruction within 30ms, restoring the 720p image to 1080p visual quality. The frame is then delivered to the central platform via the existing backhaul channel. This process ensures that key frames arrive within 30ms while halving air interface bandwidth usage, alleviating wireless congestion during peak hours while fully utilizing idle computing power onboard and at the edge, achieving balanced optimization of latency, bandwidth, and image quality.

[0046] Optionally, the method further includes: Based on the spatiotemporal characteristic data of the rail transit environment, identifying the operating mode of the rail transit environment, wherein the operating mode includes a peak mode, an off-peak mode, or an off-peak mode; The computing resources and the network resources are collaboratively allocated according to the target scheduling mode and the operation mode.

[0047] In the present invention, during the operation of the train, the system continuously collects the spatiotemporal characteristic data of the rail transit environment, namely the line section, time, station passenger flow density and historical passenger flow curve of the current train, and divides the overall operation status into peak mode, off-peak mode or off-peak mode accordingly.

[0048] The dispatch center matches the determined target dispatch mode with the operating mode in two dimensions. When a train is at an interchange hub during morning and evening peak hours, the system first reserves a FlexE hard-isolated channel for the highest-priority services, such as CBTC signaling and driver monitoring, and then executes the current target mode within the remaining bandwidth. If a train enters the depot during nighttime idle time, a dedicated backhaul channel with 200% of the peak bandwidth is opened, and software upgrades or batch inspection data are sent back in advance. During off-peak hours, with the goal of maximizing the QoS / energy consumption ratio, key frames are placed in the TSN deterministic channel, and non-key frames share elastic bandwidth, thereby achieving coupled optimization of the operating mode and scheduling strategy, so that computing power, storage, and network resources can be accurately allocated in both time and space dimensions, avoiding the coexistence of peak resource gaps and off-peak redundancy.

[0049] Optionally, collaboratively allocating the computing resources and the network resources according to the target scheduling mode and the operating mode includes: When the operating mode is peak mode and the target scheduling mode is high-fidelity mode, dedicated network slice resources are allocated to the service corresponding to the service information in the available network resources outside the hard-isolated network channel reserved for the preset highest priority service.

[0050] In this invention, during peak mode, when trains run at transfer hubs with heavy passenger flow, the system first locks a FlexE hard-isolated channel for the highest-priority services, such as CBTC signaling and driver monitoring. This channel is configured with exclusive fixed time slots at base stations, transmission nodes, and the core network along the route, ensuring that latency and jitter are always within safety thresholds.

[0051] Subsequently, under the triggering condition of high-fidelity mode, the scheduling hub immediately establishes a dedicated network slice in the remaining network resources outside the above hard-isolated channel for the second-highest priority service corresponding to the current service information; This slice completes path orchestration at the millisecond level through SRv6 segmented routing, and the bandwidth size is dynamically adjusted based on the real-time score. It neither competes for resources with train control instructions nor provides predictable low-latency transmission for the service during peak congestion periods, thereby achieving differentiated, reliable fidelity for multi-level services in peak scenarios.

[0052] Optionally, collaboratively allocating the computing resources and the network resources according to the target scheduling mode and the operating mode includes: When the operation mode is the idle mode, a large-bandwidth network channel is allocated for executing the target scheduling mode.

[0053] During off-peak hours when trains are returning to their destination or the entire line is shut down at night, the dispatching hub aggregates the idle FlexE timeslots and optical layer channels on the entire transmission link into a single high-bandwidth network channel, with a capacity of up to 200% of the daytime peak.

[0054] This channel uses 100Gbps port bundling at the physical layer and is marked as an "off-peak" slice in the SRv6 controller. It is only open to low-real-time services such as cache backhaul, software upgrades, and equipment inspection data.

[0055] The onboard terminal, trackside MEC and central cloud upload previously cached high-definition videos, log files or AI model incremental packages in batches in a multi-path parallel manner through this channel, with the average throughput increased to 1.2Gbps. 200GB of data can be transmitted back within the maintenance window of 02:00-04:00. The channel is then automatically released to ensure that resources are returned to zero before the peak of the next day, thereby achieving full utilization of network resources during idle time and precise scheduling of computing network collaboration.

[0056] In an alternative embodiment, Figure 2 The overall structure diagram provided by the present invention is as follows: Figure 2 As shown, this solution analyzes rail transit user service types and real-time requirements. On the one hand, it fully leverages the advantages of computing power and network synergy, offloading some computing tasks to the end-side. Combined with the computing power of the end-to-end computing gateway, it performs on-demand video compression and encoding during peak hours, reducing pressure on network bandwidth and improving user experience. On the other hand, it builds a scheduling system that adopts dynamic and adaptive transmission and processing strategies in different time periods and at different sites based on traffic forecasts and business indicator requirements. While ensuring real-time processing, it can meet the relevant industry requirements for video fidelity and achieve a balance between latency and quality. Specifically, it includes a computing network resource access component, a computing network dynamic perception component, and a computing network intelligent scheduling component.

[0057] The computing network resource access component primarily collects raw data from computing network resources through basic resource access (cloud, edge, end, network, etc.). It adopts a hybrid access mode, configures Kafka streaming collection channels for time-series data such as signal systems and video surveillance, and uses the Fluentd log collection framework for unstructured data such as device logs. It supports management of the cycle and method of data collection, as well as preprocessing, adaptation, and conversion of various access data. At the same time, through docking end computing gateways and cloud platform interfaces, tasks such as video processing, AI reasoning, and data processing can be flexibly distributed to relevant computing resources, effectively improving the utilization of on-board terminal computing resources. The integrated SRv6 programmable forwarding technology and FlexE hard slicing technology provide deterministic latency fidelity for key rail transit services, improve network resource utilization, and reduce the conflict rate of services of different priorities. Specifically, it includes a basic resource docking engine, a computing power offload engine, and a network management engine.

[0058] Furthermore, the basic resource docking engine docks with the system's underlying cloud, edge, end, and network resources, managing these underlying basic resources through interface acquisition, active reporting, and intermediate table acquisition. Interface acquisition supports passive acquisition of computing power information through an interface; active reporting supports actively opening a reporting interface and actively reporting computing power information through the interface; and intermediate table acquisition supports providing a data table format and periodically reporting data in a fixed format.

[0059] Furthermore, the computing power offloading engine, through the docking end computing gateway, cloud platform interface, etc., can flexibly distribute tasks such as video processing, AI reasoning, and data processing to relevant computing resources, effectively improving the utilization rate of computing power resources of the vehicle terminal, while reducing cloud load and network transmission pressure.

[0060] The end computing gateway is deployed on train-mounted terminals and station edge devices, supports a lightweight container runtime environment, realizes virtualization and encapsulation of local computing resources, and has a built-in priority queue management algorithm. It monitors the CPU / GPU / NPU computing power reserve in real time and reports resource status through the MQTT protocol.

[0061] The cloud platform interface builds a two-way communication channel and supports access to heterogeneous resource pools across cloud service providers. The interface layer uses adaptive protocol conversion technology to achieve seamless integration between the Kubernetes cluster API and the rail transit-specific protocol.

[0062] For example, for time-sensitive AI inference tasks, model segmentation technology is used to deploy the feature extraction layer on the terminal NPU to reduce network transmission pressure.

[0063] Furthermore, the network management engine integrates SRv6 programmable forwarding technology by connecting to the network scheduling interface, network resource allocation management module, etc., to achieve dynamic orchestration of cross-base station service chains, and integrates FlexE hard slicing technology to provide deterministic latency fidelity for key rail transit services, thereby realizing intelligent allocation of network resources under the collaboration of toB (enterprise-level services, including train control, equipment monitoring, etc.) and toC (consumer services, including passenger services, commercial applications, etc.), improving network resource utilization, and reducing the conflict rate of services with different priorities.

[0064] The computing network dynamic perception component senses underlying computing, network, and storage resources, building a two-tiered computing power assessment system at the edge and terminal levels. This system monitors available computing resources (CPU / GPU / NPU utilization, memory remaining, and battery status) on mobile terminals (vehicle-mounted devices and handheld devices) in real time. Furthermore, it assesses the current service quality by sensing user-side latency, data transmission volume, upload traffic, task priority, and data fidelity requirements (4K / 1080P / 720P). This includes the computing network resource perception engine and the computing network service perception engine.

[0065] Furthermore, the computing network resource perception engine integrates computing power (end computing power, edge computing power, and industry cloud), networks, and business applications. It uniformly collects and manages performance parameters of network equipment such as 5G base stations, cloud management platforms, and other perception data sources. It also implements network performance and resource load perception for typical computing network business scenarios in the rail transit industry, including computing power resource perception and network resource perception.

[0066] The computing resource perception mainly uses Linux cgroup technology to collect real-time CPU usage and GPU temperature of computing resources, and SSD remaining capacity and IOPS of storage resources.

[0067] The network resource perception mainly obtains RSRP, SINR, and PDCP layer packet loss rate based on 5G NR measurement reports.

[0068] Furthermore, the computing network service perception engine perceives user-side latency, data transmission volume, upload traffic, task priority, data fidelity requirements and other indicators, analyzes and processes the data based on the indicator system, and builds a user service classification model to support real-time decision-making in computing network scheduling, operational fidelity, intelligent operation and maintenance and other demand scenarios. The user service classification model includes: Emergency (T1): Driver dangerous behavior identification (latency requirement ≤ 200ms); Critical category (T2 level): equipment fault diagnosis (allowable delay ≤ 2s); General category (T3): carriage monitoring video backhaul (allowed delay ≥ 5 minutes); The intelligent scheduling component of the computing network mainly forms capabilities such as rail transit business priority scheduling and time-sharing tidal scheduling through system modeling, business rule analysis, optimization strategy solution and system module docking, and builds a computing network collaborative intelligent scheduling system for rail transit applications, including hierarchical scheduling strategies such as real-time fidelity mode, terminal computing power compression transmission + rail transit local enhancement mode, pause transmission + local cache mode for different priority task scenarios, and dynamic adaptive transmission and processing strategies in different time periods and different sites for tidal effect business scenarios, effectively improving the utilization efficiency of computing power resources, reducing network transmission pressure, and improving user experience. Specifically, it includes a rail transit business priority scheduling engine and a rail transit time-sharing tidal scheduling engine.

[0069] Furthermore, the rail transit service priority scheduling engine, targeting scenarios with different priority levels of emergency, critical, and ordinary tasks, such as rail transit video, conducts intelligent analysis and decision-making based on perceived computing network resource indicators and business operation indicators through system modeling, algorithm optimization, and strategy generation, and selects appropriate computing nodes, routing strategies, and bandwidth allocation strategies to provide services. The steps include: Users issue video capture tasks from the management platform and determine service quality indicators such as latency. Combined with resource indicators such as computing power and network, a dynamic weighted scoring model is established. Based on network quality (RSRP / SINR), terminal computing power load, and service priority, the strategy score is calculated to provide a basis for selecting the optimal transmission strategy.

[0070] ; in, .

[0071] Network rating : Calculated based on RSRP / SINR; Hashrate Rating : Calculated based on terminal CPU / GPU / NPU utilization and edge node load; Business Rating : T1=100, T2=60, T3=20.

[0072] The dispatching platform implements hierarchical dispatching strategies based on dynamic weighted scores, fully dispatching the end, edge computing power, and network, and adopts various strategy modes such as real-time fidelity mode, end computing power compression transmission + rail transit local enhancement mode, and pause transmission + local cache mode. The scoring range is shown in Table 1 below: Table 1

[0073] High-fidelity mode allocates dedicated network slices for T1 high-priority services (such as driver behavior monitoring); Balanced mode, namely end-to-end computing power compression transmission + rail transit local enhancement mode, mainly implements layered encoding (bitrate compression of 50%) for T2 medium-priority service video streams, and performs super-resolution reconstruction on the trackside MEC nodes; Caching mode, which is a suspended transmission + local caching mode, is primarily used to temporarily store low-priority T3 data in the vehicle's SSD array during network congestion. For example, in cases of network congestion, real-time surveillance video data transmission can be temporarily suspended, or some pre-processed low-fidelity data can be transmitted by offloading computing power to the end, thereby preserving the fidelity of high-priority data such as driver risk monitoring, thereby achieving coordinated scheduling of computing power and network.

[0074] Based on the global dispatching management strategy, when the network is idle (network utilization <40% and lasts ≥5 minutes) or when the train enters the terminal maintenance area (GPS coordinate matching), the locally stored fidelity data will be uploaded to ensure data integrity.

[0075] Based on changes in perceived network quality, resource load and other indicators, the current service scheduling strategy is dynamically adjusted to form a feedback optimization closed loop.

[0076] Furthermore, the rail transit time-sharing tidal scheduling engine is mainly aimed at business scenarios with tidal effects such as rail transit during commuting hours. It coordinates the supply and scheduling of distributed computing power, storage, and traffic, improves the application of terminal computing power, adopts dynamic and adaptive transmission and processing strategies in different time periods and different sites, and monitors the scheduling execution process end-to-end. It automatically handles anomalies and errors according to relevant strategies. On the basis of ensuring real-time processing, it can meet the requirements of relevant industries for video fidelity, reduce network peak bandwidth pressure, and improve customer experience. It includes the following steps: The temporal and spatial characteristics of rail transit are analyzed for different time periods and different stations, and divided into three modes: peak, off-peak, and off-peak, as shown in Table 2 below: Table 2

[0077] Different modes are matched with differentiated, dynamically adaptive resource allocation and optimized scheduling strategies. The details are as follows: a) In peak mode (S ≥ 80), a hard-isolated channel (FlexE slice ≥ 800 Mbps) is reserved for the train control system. This channel is used only for critical traffic such as train emergency braking and ATO commands, ensuring SRv6 direct transmission of critical services (latency ≤ 10 ms). Dynamic bandwidth preemption is also implemented. When the control command latency exceeds 15 ms, the passenger commercial traffic bandwidth is automatically reduced proportionally (up to 70%). SRv6 segment routing is used to achieve millisecond-level resource reallocation, ensuring the absolute reliability and real-time performance of high-priority services such as train control and emergency communications.

[0078] b) In off-peak mode (40 ≤ S < 80), with the goal of maximizing the overall performance ratio (QoS / energy consumption), key frames are allocated TSN deterministic channels (latency ≤ 30ms), and non-critical frames use elastic shared bandwidth. This maximizes resource utilization while maintaining basic service quality.

[0079] c) In off-peak mode (S<40), a dedicated channel (with bandwidth up to 200% of the daytime peak) is opened from 02:00 to 04:00 daily for batch transmission of equipment inspection data. This channel also utilizes idle network resources during off-peak hours for intelligent pre-distribution of software upgrades, delivering 80% of upgrade packages in advance and completing the remaining 20% ​​of incremental updates before peak hours.

[0080] Establish a strategy knowledge base to store verified effective scheduling plans, and continuously improve scheduling strategies based on changes in resource indicators and business indicators to meet the needs of different rail transit business scenarios.

[0081] The solution of the present invention adopts hierarchical scheduling strategies such as real-time fidelity mode, terminal computing power compression transmission + rail transit local enhancement mode, pause transmission + local cache mode, etc. for rail transit task scenarios of different priorities such as emergency, critical, and ordinary, based on computing network resource indicators and business indicator requirements. In addition, for rail transit tidal effect business scenarios, dynamic adaptive transmission and processing strategies are adopted in different time periods and different stations. On the basis of ensuring real-time processing, it can meet the requirements of related businesses for video fidelity, achieve a balance between latency and quality, and provide effective support for the intelligent evolution of rail transit services.

[0082] Specifically, through the docking end computing gateway, cloud platform interface, etc., tasks such as video processing, AI reasoning, and data processing can be flexibly distributed to relevant computing resources, effectively improving the utilization rate of vehicle-mounted terminal computing resources by more than 30%, while reducing cloud load and network transmission pressure; by adopting dynamic and adaptive transmission and processing strategies in different time periods and different sites, the absolute reliability and real-time performance of high-priority services such as train control and emergency communications can be ensured, and business efficiency can be improved by more than 50%.

[0083] The computing network collaborative scheduling device provided by the present invention is described below. The computing network collaborative scheduling device described below and the computing network collaborative scheduling method described above can be referenced to each other.

[0084] Figure 3 This is a schematic diagram of the structure of the computing network collaborative scheduling device provided by the present invention, as shown in FIG. Figure 3 Shown, including: The acquisition module 310 is used to obtain computing resources, network resources and business information including business priorities in the rail transit environment; The determination module 320 is configured to calculate a comprehensive scheduling score based on the computing resources, the network resources, and the service priority, and determine a target scheduling mode according to the comprehensive scheduling score; The allocation module 330 is used to collaboratively allocate the computing resources and the network resources to process the business information according to the target scheduling mode.

[0085] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: When the comprehensive scheduling score is greater than a first preset threshold, determining that the target scheduling mode is a high-fidelity mode; wherein the high-fidelity mode is used to allocate dedicated network slice resources for the service corresponding to the service information; When the comprehensive scheduling score is less than a second preset threshold, the target scheduling mode is determined to be a cache mode, wherein the cache mode is used to temporarily store data corresponding to the service information in the vehicle terminal.

[0086] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: When the comprehensive scheduling score is between the first preset threshold and the second preset threshold, determining that the target scheduling mode is a balanced mode; Among them, the balancing mode is used to compress the data corresponding to the business information through the vehicle terminal, and perform quality enhancement processing on the compressed data at the edge computing node.

[0087] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: Based on the spatiotemporal characteristic data of the rail transit environment, identifying the operating mode of the rail transit environment, wherein the operating mode includes a peak mode, an off-peak mode, or an off-peak mode; The computing resources and the network resources are collaboratively allocated according to the target scheduling mode and the operation mode.

[0088] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: When the operating mode is peak mode and the target scheduling mode is high-fidelity mode, dedicated network slice resources are allocated to the service corresponding to the service information in the available network resources outside the hard-isolated network channel reserved for the preset highest priority service.

[0089] According to a computing network collaborative scheduling device provided by the present invention, the device is further used for: When the operation mode is the idle mode, a large-bandwidth network channel is allocated for executing the target scheduling mode.

[0090] The present invention continuously collects computing resources, network resources, and business information carrying business priorities in the rail transit environment within the same control plane, and establishes a comprehensive scheduling scoring model with multi-dimensional real-time data as input; based on the dynamic changes of the score, the system instantly selects the appropriate target scheduling mode without manual intervention, and coordinates and arranges computing power and network resources accordingly. This process transforms the originally static and fragmented resource configuration into a closed-loop adaptive mechanism, allowing computing power to flow on demand between on-board terminals, trackside edges, and central clouds, and network bandwidth to converge or release in real time with the urgency of the business. Under the premise that the total amount of physical resources remains unchanged, it simultaneously solves existing problems such as high idle rate of on-board computing power, excessive end-to-end latency, and mutual crowding out of critical and non-critical businesses, providing rail transit with intelligent scheduling capabilities with high reliability, low latency, and significantly improved resource utilization.

[0091] Figure 4 Schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other via the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the computing network collaborative scheduling method, which includes: obtaining computing resources, network resources and service information including service priorities in a rail transit environment; Calculating a comprehensive scheduling score based on the computing resources, the network resources, and the service priority, and determining a target scheduling mode according to the comprehensive scheduling score; According to the target scheduling mode, the computing resources and the network resources are collaboratively allocated to process the business information.

[0092] Furthermore, the logic instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0093] On the other hand, the present invention further provides a computer program product, the computer program product including a computer program, the computer program being storable on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer is capable of executing the computing-network collaborative scheduling method provided by the above methods, the method including: obtaining computing resources, network resources, and service information including service priorities in a rail transit environment; Calculating a comprehensive scheduling score based on the computing resources, the network resources, and the service priority, and determining a target scheduling mode according to the comprehensive scheduling score; According to the target scheduling mode, the computing resources and the network resources are collaboratively allocated to process the business information.

[0094] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the computer-network collaborative scheduling method provided by the above methods is implemented, the method comprising: obtaining computing resources, network resources, and service information including service priorities in a rail transit environment; Calculating a comprehensive scheduling score based on the computing resources, the network resources, and the service priority, and determining a target scheduling mode according to the comprehensive scheduling score; According to the target scheduling mode, the computing resources and the network resources are collaboratively allocated to process the business information.

[0095] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0096] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A computing network collaborative scheduling method, characterized in that: include: Obtain computing resources, network resources, and business information including business priorities in the rail transit environment; Calculating a comprehensive scheduling score based on the computing resources, the network resources, and the service priority, and determining a target scheduling mode according to the comprehensive scheduling score; According to the target scheduling mode, the computing resources and the network resources are collaboratively allocated to process the business information.

2. The method for collaborative scheduling of computing networks according to claim 1, characterized in that: Determine the target scheduling mode based on the comprehensive scheduling score, including: When the comprehensive scheduling score is greater than a first preset threshold, determining that the target scheduling mode is a high-fidelity mode; wherein the high-fidelity mode is used to allocate dedicated network slice resources for the service corresponding to the service information; When the comprehensive scheduling score is less than a second preset threshold, the target scheduling mode is determined to be a cache mode, wherein the cache mode is used to temporarily store data corresponding to the service information in the vehicle terminal.

3. The method for collaborative scheduling of computing networks according to claim 2, characterized in that: Determine the target scheduling mode based on the comprehensive scheduling score, including: When the comprehensive scheduling score is between the first preset threshold and the second preset threshold, determining that the target scheduling mode is a balanced mode; Among them, the balancing mode is used to compress the data corresponding to the business information through the vehicle terminal, and perform quality enhancement processing on the compressed data at the edge computing node.

4. The method for collaborative scheduling of computing networks according to claim 1, characterized in that: The method further comprises: Based on the spatiotemporal characteristic data of the rail transit environment, identifying the operating mode of the rail transit environment, wherein the operating mode includes a peak mode, an off-peak mode, or an off-peak mode; The computing resources and the network resources are collaboratively allocated according to the target scheduling mode and the operation mode.

5. The method for collaborative scheduling of computing networks according to claim 4, characterized in that: Collaboratively allocating the computing resources and the network resources according to the target scheduling mode and the operating mode includes: When the operating mode is peak mode and the target scheduling mode is high-fidelity mode, dedicated network slice resources are allocated to the service corresponding to the service information in the available network resources outside the hard-isolated network channel reserved for the preset highest priority service.

6. The method for collaborative scheduling of computing networks according to claim 4, characterized in that: Collaboratively allocating the computing resources and the network resources according to the target scheduling mode and the operating mode includes: When the operation mode is the idle mode, a large-bandwidth network channel is allocated for executing the target scheduling mode.

7. A computing network collaborative scheduling device, characterized in that: include: The acquisition module is used to obtain computing resources, network resources, and business information including business priorities in the rail transit environment; a determination module, configured to calculate a comprehensive scheduling score based on the computing resources, the network resources, and the service priority, and determine a target scheduling mode according to the comprehensive scheduling score; An allocation module is used to collaboratively allocate the computing resources and the network resources to process the business information according to the target scheduling mode.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the computing network collaborative scheduling method as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computing network collaborative scheduling method as described in any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the computing network collaborative scheduling method as described in any one of claims 1 to 6 is implemented.

Images