Urban rail-oriented edge computing data collaborative processing and transmission scheduling method and system

Abstract

The application relates to the technical field of edge computing, and discloses an edge computing data cooperative processing and transmission scheduling method and system for urban rail, which comprises the following steps: collecting urban rail multi-source heterogeneous data, outputting a standardized data set and a topological adjacency matrix set after data preprocessing; performing business hierarchical label mapping, constructing a computing-transmission joint baseline library, extracting joint characteristic quantities, outputting a joint baseline parameter set and a business characteristic data set; performing task granularity adaptive slicing and calculating node cooperative adaptation degree, screening a candidate cooperative node set and a task slicing set; solving an optimal time slot allocation and a transmission path scheme through a joint scheduling utility model and outputting a result; calculating an efficiency score and performing parameter incremental evolutionary updating, and feeding back an optimized front-end baseline and model parameters. The application can adapt to a high-dynamic operation scene of urban rail, realizes joint optimal scheduling of computing power and transmission resources, and significantly improves resource utilization rate and time delay guarantee capacity; the whole scheduling process is traceable, and the model is iterative.

Description

Technical Field

[0001] This invention relates to the field of urban rail communication and edge computing technology, and in particular to a method and system for collaborative processing and transmission scheduling of edge computing data for urban rail transit. Background Technology

[0002] Currently, with the rapid development of intelligent urban rail transit and vehicle-to-ground collaborative services, the massive amounts of business data generated by onboard intelligent terminals, trackside sensing equipment, and station control systems place stringent demands on low-latency processing and highly reliable transmission. Edge computing architecture, by deploying computing nodes at the network edge to achieve local data processing and forwarding, has become the core technological foundation supporting various services such as urban rail train control, passenger services, and equipment monitoring.

[0003] Current edge computing and transmission scheduling solutions for urban rail transit still have several technical shortcomings. First, traditional solutions often employ independent scheduling of computing power and transmission, failing to achieve joint optimization between the two. This easily leads to resource mismatches, such as sufficient computing power but insufficient transmission bandwidth, or idle links but saturated computing power, resulting in low overall resource utilization and an inability to adapt to the highly dynamic fluctuations in urban rail services. Second, most solutions use fixed-granularity task offloading strategies, failing to dynamically adjust task fragmentation granularity based on real-time node load and service level. When high-speed train movement causes frequent topology switching, offloading tasks are prone to timeouts or failures, resulting in insufficient robustness. Third, existing scheduling often focuses on single optimization targets such as latency or bandwidth, failing to construct a joint utility model that integrates multiple dimensions such as service priority, resource utilization, and transmission reliability. The scheduling results struggle to balance the high latency guarantee for security services with efficient resource utilization for non-critical services. Fourth, time slot allocation often uses fixed partitioning mechanisms, unable to dynamically adjust time slot proportions based on service priority and real-time needs. This results in insufficient latency guarantee for critical services during peak hours and significant resource idleness during off-peak hours. Fifth, the scheduling decision-making lacks a full-link traceability mechanism and closed-loop performance evaluation. Model parameter updates rely on manual offline tuning and cannot evolve incrementally online based on actual scheduling results, leading to a continuous decline in scheduling performance over long-term operation. Therefore, there is an urgent need for an edge computing data collaborative processing and transmission scheduling method for urban rail transit that features joint optimization of computing power and transmission, adaptive task sharding, dynamic time slot allocation, and closed-loop evolution. Summary of the Invention

[0004] To address the aforementioned technical shortcomings, the purpose of this invention is to propose an edge computing data collaborative processing and transmission scheduling method for urban rail transit, aiming to solve the technical problems in the prior art, such as mismatch between computing power and transmission resources, poor robustness of fixed task fragmentation granularity, single optimization objective, rigid time slot allocation, and inability of the model to evolve online.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides an edge computing data collaborative processing and transmission scheduling method for urban rail transit.

[0006] The edge computing data collaborative processing and transmission scheduling method for urban rail transit includes: Step S10: Collect multi-source heterogeneous data from the computing power domain, transmission domain, and business domain of the three-level edge nodes (vehicle-mounted, trackside, and station) in the urban rail scenario, perform spatiotemporal benchmark alignment and standardization mapping processing, and output the spatiotemporal standardized dataset and the edge node topology adjacency matrix set. Step S20: Based on the spatiotemporal standardized dataset and topological adjacency matrix set obtained in step S10, perform hierarchical label mapping according to business latency sensitivity, construct a computing power-transmission joint baseline library and extract multi-dimensional joint features, and output the joint baseline parameter set and business feature dataset. Step S30: Based on the joint baseline parameter set and business feature dataset output in step S20, perform task-granular adaptive sharding and calculate the collaborative adaptation degree of each edge node, filter and generate a candidate collaborative node set and a corresponding adaptation degree score set, and output the task sharding set and the candidate collaborative node set. Step S40: Based on the task sharding set and candidate collaborative node set output in step S30, construct a computing power-transmission joint scheduling utility model, solve for the optimal time slot allocation and transmission path planning scheme, and output the joint scheduling parameter set and path planning result set; Step S50: Based on the joint scheduling parameter set and path planning result set output in step S40, construct the full-link scheduling performance traceability chain, calculate the scheduling performance score and perform incremental parameter evolution updates, output the final scheduling results and model evolution parameters, and feed back and update the computing power-transmission joint scheduling performance model.

[0007] Preferably, the step S10, which outputs the spatiotemporally normalized dataset and the edge node topological adjacency matrix set, specifically includes: Step S101: Synchronously collect CPU computing power, memory capacity, and storage margin computing power domain data of each edge node; synchronously collect link bandwidth, end-to-end latency, packet loss rate, and handover interval transmission domain data; synchronously collect train control, passenger service, and equipment monitoring business domain data; and synchronously collect node ledger and topology connection relationship data. Step S102: Perform millisecond-level spatiotemporal alignment with a unified clock reference, perform dimensional normalization and outlier filtering on heterogeneous data, and remove invalid data segments with continuous out-of-synchronization duration exceeding a preset threshold. Step S103: Generate a spatiotemporally standardized dataset with unified spatiotemporal reference and standardized format, and synchronously output an edge node topology adjacency matrix set containing node adjacency relationships and link weight parameters.

[0008] Preferably, step S20, which involves completing hierarchical label mapping based on service latency sensitivity, constructing a computing power-transmission joint baseline library and extracting multi-dimensional joint features, and outputting the joint baseline parameter set and service feature dataset, specifically includes: Step S201: Divide services into three levels—security, service, and monitoring—based on latency sensitivity, and map corresponding latency thresholds and resource guarantee priority labels to each type of service; Step S202: Classify and statistically analyze the labeled historical normal operation data according to service level, fit the joint distribution characteristics of computing power consumption and transmission delay for each service level, calculate the mean, covariance and boundary threshold, and generate a joint baseline library of computing power and transmission. Step S203: Extract four types of joint feature quantities from real-time data: computing power demand intensity, transmission bandwidth demand, latency margin, and packet loss sensitivity. Calculate feature statistics for each scheduling frame and integrate and output the joint baseline parameter set and the service feature quantity dataset.

[0009] Preferably, step S30, which involves performing task-granular adaptive sharding and calculating the collaborative fit of each edge node, filtering and generating a candidate collaborative node set and a corresponding fit score set, and outputting the task shard set and the candidate collaborative node set, specifically includes: Step S301: Based on the business level label and the proportion of remaining computing power of the local node, combined with the total data volume of the task and the baseline data volume of the unit shard, round up to calculate the optimal sharding granularity of the task, and split the original task into several sub-shards that can be processed in parallel. Step S302: For each task fragment, traverse the adjacent edge nodes, and calculate the collaborative adaptation score by combining the three indicators of node computing power remaining rate, link available bandwidth ratio and transmission latency cost according to the preset weights. Step S303: Filter nodes with a collaboration fit greater than the preset fit threshold, generate a candidate collaboration node set, and synchronously output the split task fragment set and the corresponding fit score set.

[0010] Preferably, step S40, which involves constructing a joint scheduling utility model for computing power and transmission, solving for the optimal time slot allocation and transmission path planning scheme, and outputting a joint scheduling parameter set and a path planning result set, specifically includes: Step S401: For each task slice and its candidate node set, retrieve the full-dimensional joint feature quantity of the current scheduling frame as the model input, and construct the computing power-transmission joint scheduling utility function; Step S402: With the goal of maximizing the overall scheduling utility, solve for the computing power allocation, transmission time slot ratio, and optimal transmission path for each task fragment; the calculation of the overall scheduling utility for a single scheduling frame satisfies:

[0011] in, For the total scheduling utility of a single scheduling frame, The total number of task fragments, For the first The priority weight of the business to which each shard belongs. For the first Each segment belongs to a service with latency requirements thresholds. For the first The total actual processing and transmission latency of each segment For the first The amount of resources occupied by each fragment. For the first The total resource amount of each shard corresponding to the node and link. For index value, , The utility weight coefficient is used to proportionally allocate the transmission slot length within a single scheduling frame based on the utility optimal solution, according to the priority weight of each fragment and the proportion of unit utility value. Step S403: Summarize the computing power allocation, time slot configuration and path planning results of all shards to generate a joint scheduling parameter set and a path planning result set.

[0012] Preferably, step S50, which involves constructing a full-link scheduling performance traceability chain, calculating the scheduling performance score, performing incremental parameter evolution updates, outputting the final scheduling results and model evolution parameters, and feeding back and updating the computing power-transmission joint scheduling performance model, specifically includes: Step S501: Associate each scheduling result with the original data identifier, feature parameters, model version, and execution log to construct a full-link performance traceability chain from original data collection to scheduling execution; Step S502: Statistically calculate the latency compliance rate, task completion rate, and average resource utilization rate for a single scheduling cycle. Calculate the comprehensive scheduling performance score using preset weights. Based on the deviation between the performance score and the baseline threshold, perform incremental evolutionary updates to the baseline parameters and the scheduling model. The parameter update calculation satisfies the following:

[0013] in, For model parameter update amount, For adaptive learning step size coefficients, Score the current scheduling efficiency. As the baseline performance threshold, These are the current model parameter values; Step S503: Output the final scheduling result with traceability chain, synchronously feed back the evolution parameters to the previous steps, and adaptively update the joint baseline parameter set and the computing power-transmission joint scheduling utility model.

[0014] Preferably, step S50 further includes automatically including periodic data with scheduling performance scores greater than a preset qualified threshold into the labeled sample pool, and marking periodic data with scores lower than the preset qualified threshold as a state to be analyzed, and not directly participating in parameter evolution.

[0015] This invention also provides an edge computing data collaborative processing and transmission scheduling system for urban rail transit, including: Multi-domain data spatiotemporal mapping module: used to collect multi-source heterogeneous data from computing power domain, transmission domain, and business domain of three-level edge nodes in urban rail transit scenarios (vehicle-mounted, trackside, and station-mounted), perform spatiotemporal benchmark alignment and standardization mapping processing, and output spatiotemporal standardized dataset and edge node topology adjacency matrix set; Business Classification and Joint Baseline Module: Based on the spatiotemporal standardized dataset and topological adjacency matrix set obtained by the multi-domain data spatiotemporal mapping module, it completes the classification label mapping according to the business latency sensitivity, constructs the computing power-transmission joint baseline library and extracts multi-dimensional joint features, and outputs the joint baseline parameter set and business feature dataset. Task sharding and node filtering module: Based on the joint baseline parameter set and business feature dataset output by the business classification and joint baseline module, it performs task-granular adaptive sharding and calculates the collaborative adaptation of each edge node, filters and generates a candidate collaborative node set and a corresponding adaptation score set, and outputs the task sharding set and the candidate collaborative node set. Joint scheduling slot planning module: Based on the task sharding set and candidate collaborative node set output by the task sharding and node screening module, it constructs a computing power-transmission joint scheduling utility model, solves the optimal slot allocation and transmission path planning scheme, and outputs a joint scheduling parameter set and path planning result set. The performance traceability and parameter evolution module is used to construct a full-link scheduling performance traceability chain based on the joint scheduling parameter set and path planning result set output by the joint scheduling time slot planning module, calculate the scheduling performance score and perform incremental parameter evolution updates, output the final scheduling results and model evolution parameters, and provide feedback and update the computing power-transmission joint scheduling performance model.

[0016] This invention also provides an edge computing data collaborative processing and transmission scheduling device for urban rail transit, the edge computing data collaborative processing and transmission scheduling device for urban rail transit comprising: The system includes a memory, a processor, and an edge computing data collaborative processing and transmission scheduling program for urban rail transit stored in the memory and executable on the processor. When the edge computing data collaborative processing and transmission scheduling program for urban rail transit is executed by the processor, the above method is implemented.

[0017] The present invention also provides a computer program product, the computer program product including an edge computing data collaborative processing and transmission scheduling program for urban rail transit, the edge computing data collaborative processing and transmission scheduling program for urban rail transit implementing the above method when executed by a processor.

[0018] The beneficial effects of this invention are as follows: This invention constructs a unified joint scheduling data foundation by standardizing the spatiotemporal mapping of multi-source data from the computing power domain, transmission domain, and service domain. This breaks through the information barriers of independent scheduling of computing power and transmission, providing global data support for joint optimization. By employing service-level labeling and computing power-transmission joint baseline modeling, it accurately depicts the resource demand distribution characteristics of different service levels, significantly improving the matching degree of service requirements compared to a single-dimensional baseline.

[0019] Furthermore, this invention employs a task-adaptive sharding and collaborative adaptability screening mechanism, which can dynamically adjust the sharding granularity based on the real-time load of nodes and match the optimal collaborative nodes, significantly improving scheduling robustness and offloading success rate in scenarios with frequent topology switching. A multi-objective joint scheduling utility model is constructed to achieve dynamic time slot allocation and optimal path planning, maximizing resource utilization while ensuring low latency for safe services. The overall scheduling performance is significantly better than single-objective solutions. Through a full-link performance traceability and incremental parameter evolution mechanism, scheduling decisions are traceable throughout the entire process. The model can be iteratively optimized online based on actual operational results, exhibiting strong long-term performance retention capabilities and meeting the high reliability and compliance requirements of urban rail transit operations. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the first embodiment of the edge computing data collaborative processing and transmission scheduling method for urban rail transit according to the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0022] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the edge computing data collaborative processing and transmission scheduling method for urban rail transit according to the present invention. The first embodiment of the edge computing data collaborative processing and transmission scheduling method for urban rail transit according to the present invention is presented.

[0024] In the first embodiment, the edge computing data collaborative processing and transmission scheduling method for urban rail transit includes: Step S10: Collect multi-source heterogeneous data from the computing power domain, transmission domain, and business domain of the three-level edge nodes (vehicle-mounted, trackside, and station) in the urban rail scenario, perform spatiotemporal benchmark alignment and standardization mapping processing, and output the spatiotemporal standardized dataset and the edge node topology adjacency matrix set.

[0025] It should be noted that this step covers the full-dimensional data collection of computing power, transmission, and services across three levels of nodes: onboard, trackside, and station. It also outputs a topological adjacency matrix, providing a global topological basis for subsequent joint scheduling. Spatiotemporal benchmark alignment unifies heterogeneous data from different sampling frequencies and clock sources into the same spatiotemporal coordinate system, eliminating spatiotemporal offsets caused by train movement. Dimensional normalization maps computing power, bandwidth, and latency indicators of different dimensions to a unified numerical range, facilitating subsequent joint calculations. Invalid segment removal removes continuously out-of-synchronization fault data, avoiding interference with baseline modeling and scheduling decisions. The output standardized dataset and adjacency matrix set are input into step S20, providing a data foundation for joint baseline construction.

[0026] Understandably, through multi-domain acquisition and spatiotemporal standardization mapping in this step, scattered heterogeneous data are integrated into a global dataset that is spatiotemporally unified, formatted, and topologically clear. This solves the problems of spatiotemporal misalignment and inconsistent dimensions of multi-node data, laying a reliable data foundation for subsequent joint scheduling.

[0027] It should be understood that traditional scheduling schemes often only collect data from a single domain, lacking a unified global spatiotemporal perspective and thus failing to support joint optimization of computing power and transmission. The multi-domain data collection and spatiotemporal mapping design in this step ensures the feasibility and accuracy of joint scheduling from the data source level.

[0028] For example, a certain urban rail line is equipped with 24 train-mounted edge units, 60 trackside edge nodes, and 15 station edge centers. The system synchronously collects data on the CPU computing power, memory, and storage computing power of each node, as well as data on the transmission domains of vehicle-to-ground link bandwidth, latency, and packet loss rate, and three types of business data: train control, passenger information, and equipment monitoring. After unified clock reference alignment, dimension normalization, and outlier filtering, four consecutive out-of-synchronization invalid data segments are removed, generating a spatiotemporally standardized dataset and a corresponding node topology adjacency matrix set.

[0029] Step S20: Based on the spatiotemporal standardized dataset and topological adjacency matrix set obtained in step S10, perform hierarchical label mapping according to business latency sensitivity, construct a computing power-transmission joint baseline library and extract multi-dimensional joint features, and output the joint baseline parameter set and business feature dataset.

[0030] It should be noted that this step categorizes services into security, service, and monitoring levels based on latency sensitivity, each corresponding to different latency requirements and resource priorities, thus achieving differentiated service guarantees. The computing power-transmission joint baseline fits the joint distribution of computing power usage and transmission latency for each service level, rather than independently calculating two-dimensional baselines, enabling a more accurate characterization of the resource coupling needs of the services. Four types of joint features characterize the resource requirements of the services from four dimensions: computing power requirement, bandwidth requirement, latency margin, and packet loss sensitivity, covering the core input dimensions for scheduling decisions. Real-time feature values ​​are calculated frame-by-frame for each scheduling frame, providing computational units for subsequent sharding and scheduling. The output joint baseline parameters and feature dataset are input into step S30 for task sharding and node selection.

[0031] Understandably, this step achieves a refined characterization of business requirements through business grading and joint baselines, solving the problem that independent baselines cannot represent resource coupling relationships. This not only improves the ability to provide differentiated support for different levels of business, but also ensures the matching degree between the baseline and actual business requirements.

[0032] It should be understood that traditional single-dimensional baseline schemes neglect the coupling relationship between computing power and transmission, which can easily lead to resource mismatch. The joint baseline design in this step achieves coordinated matching of computing power and transmission at the baseline level.

[0033] For example, select normal operation data from the past 30 days, and statistically analyze it according to three types of services: security level, service level, and monitoring level. Fit the joint distribution of computing power usage and transmission latency, calculate the mean, covariance, and 95% confidence boundary, and generate a joint baseline library of computing power and transmission for the three types of services. At the same time, using a 100-millisecond scheduling frame sliding window, calculate in real time the four joint feature values ​​of computing power demand intensity, bandwidth demand, latency margin, and packet loss sensitivity for each service, forming a service feature dataset.

[0034] Step S30: Based on the joint baseline parameter set and business feature dataset output in step S20, perform task-granular adaptive sharding and calculate the collaborative adaptation degree of each edge node, filter and generate a candidate collaborative node set and a corresponding adaptation degree score set, and output the task sharding set and the candidate collaborative node set.

[0035] It should be noted that this step dynamically calculates the optimal number of shards based on the amount of business data and the remaining computing power of the local node. When local computing power is sufficient, the number of shards is reduced to lower scheduling overhead; when local computing power is insufficient, the number of shards is increased to fully utilize the resources of neighboring nodes. The collaborative adaptability comprehensively measures the remaining computing power, link bandwidth, and transmission latency of candidate nodes, evaluating the adaptability of nodes to handle task shards from three dimensions. The higher the score, the better the adaptability. Qualified candidate nodes are selected through an adaptability threshold, narrowing the scope of subsequent optimization solutions and improving scheduling computation efficiency. The output task shard set and candidate node set are input into step S40, providing input for joint scheduling optimization.

[0036] Understandably, this step achieves dynamic adjustment of task granularity through adaptive sharding and rapid focusing of candidate nodes through collaborative adaptability screening, balancing scheduling robustness and computational efficiency. It is a key transformation step from global tasks to schedulable units.

[0037] It should be understood that fixed sharding granularity schemes cannot adapt to dynamically changing node loads and topology states, easily leading to unloading failures or resource waste. The adaptive sharding and adaptability filtering mechanism in this step can dynamically match node states, significantly improving the unloading success rate.

[0038] For example, for the six service tasks in the current frame, the optimal number of fragments is calculated for each task. Two security-level tasks are split into two fragments, three service-level tasks are split into three fragments, and one monitoring-level task is split into five fragments, generating a total of 18 task fragments. For each fragment, the 3-5 adjacent edge nodes are traversed to calculate the collaborative adaptation degree. The adaptation threshold is set to 0.6, and a total of 42 candidate nodes are selected to generate a set of candidate collaborative nodes and a corresponding adaptation degree score set.

[0039] Step S40: Based on the task sharding set and candidate collaborative node set output in step S30, construct a computing power-transmission joint scheduling utility model, solve for the optimal time slot allocation and transmission path planning scheme, and output the joint scheduling parameter set and path planning result set.

[0040] It should be noted that this step aims to maximize the overall scheduling utility. The utility function incorporates both latency protection and resource utilization benefits, and differentiated protection is achieved through service priority weights, ensuring that security-level services receive resources first. Based on the optimal utility solution, transmission time slots for each segment are dynamically allocated. High-priority, high-utility segments receive more time slot resources, while low-priority segments have their time slot usage reduced accordingly, achieving on-demand dynamic resource allocation. The optimal transmission path is simultaneously solved, and the link with the minimum latency and sufficient bandwidth is selected based on the topology adjacency matrix. The output scheduling parameters can be directly distributed to each edge node for execution.

[0041] Understandably, this step achieves global collaborative optimization of computing power and transmission through a multi-objective joint utility model, and realizes on-demand resource adaptation through dynamic time slot allocation. Compared with the single-objective fixed time slot scheme, the overall scheduling performance is significantly improved.

[0042] It should be understood that traditional single-objective scheduling cannot simultaneously guarantee latency and resource utilization, and fixed time slot allocation lacks flexibility. The joint utility model and dynamic time slot allocation in this step improve scheduling performance from two dimensions: optimizing objectives and resource allocation.

[0043] For example, for 18 task shards and 42 candidate nodes, a joint scheduling utility function is constructed to solve for the optimal allocation scheme with the goal of maximizing the total utility. The solution obtains the computing power allocation node, transmission time slot length and optimal link corresponding to each shard, in which the time slot ratio of security-level service shards is increased by 30% and the time slot ratio of monitoring-level service shards is correspondingly compressed. The joint scheduling parameter set and path planning result set are then generated.

[0044] Step S50: Based on the joint scheduling parameter set and path planning result set output in step S40, construct the full-link scheduling performance traceability chain, calculate the scheduling performance score and perform incremental parameter evolution updates, output the final scheduling results and model evolution parameters, and feed back and update the computing power-transmission joint scheduling performance model.

[0045] It should be noted that the end-to-end traceability chain links original data, feature parameters, model version, and execution logs. Any scheduling decision can be traced back to the entire chain, meeting the requirements for fault diagnosis and operational compliance. The scheduling performance score comprehensively evaluates scheduling effectiveness across three dimensions: latency compliance rate, task completion rate, and resource utilization rate. Parameters are incrementally updated based on performance deviations. When performance is better than the baseline, corresponding parameters are positively reinforced; when performance is lower than the baseline, parameters are negatively corrected, achieving online adaptive evolution of the model without requiring full retraining. Updated parameters are fed back to previous steps, forming a closed-loop evolution between the baseline and the model. The final output includes the scheduling scheme, execution results, and a complete traceability chain, which can directly generate an operational scheduling report.

[0046] Understandably, this step not only achieves full-link traceability of the scheduling process, but also realizes online self-optimization of the model through performance-driven incremental evolution, enabling the scheduling performance of the system to continuously improve iteratively during long-term operation, forming a complete closed-loop self-evolutionary system.

[0047] It should be understood that traditional static scheduling systems experience continuous performance degradation after deployment, and decision-making processes lack verifiable data. The performance tracking and incremental evolution mechanism in this step enhances the system's practical value from both compliance and long-term performance perspectives.

[0048] For example, each scheduling result is associated with the original data ID, feature parameters, model V2.0 version number, and execution log to form a complete performance traceability chain; the latency compliance rate of the current scheduling cycle is 98.2%, the task completion rate is 99.5%, and the resource utilization rate is 76.3%, and the overall performance score is calculated to be better than the baseline threshold; the system performs positive incremental updates, adjusts the baseline and model parameters, and feeds them back to step S20; low-performance boundary data are marked as pending analysis and do not participate in parameter evolution for the time being.

[0049] Example 2: Furthermore, the edge computing data collaborative processing and transmission scheduling system for urban rail transit provided by this invention, employing the edge computing data collaborative processing and transmission scheduling method for urban rail transit described in the above embodiments, can solve the technical problems in the prior art such as mismatch between independent scheduling of computing power and transmission resources, poor robustness of fixed task fragmentation granularity, single optimization objective, rigid time slot allocation, and inability of the model to evolve online. The beneficial effects of the edge computing data collaborative processing and transmission scheduling system for urban rail transit provided by this invention are the same as those of the edge computing data collaborative processing and transmission scheduling method for urban rail transit provided in the above embodiments, and other technical features of the edge computing data collaborative processing and transmission scheduling system for urban rail transit are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0050] Example 3: This invention provides an edge computing data collaborative processing and transmission scheduling device for urban rail transit. The device includes: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor, which are then executed to enable the at least one processor to perform the edge computing data collaborative processing and transmission scheduling method for urban rail transit described in Example 1. The edge computing data collaborative processing and transmission scheduling device for urban rail transit in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This edge computing data collaborative processing and transmission scheduling device for urban rail transit is merely an example and should not limit the functionality or scope of the embodiments of this invention. An edge computing data collaborative processing and transmission scheduling device for urban rail transit may include processing units (such as a central processing unit, graphics processing unit, etc.) that can perform various appropriate actions and processes based on programs stored in read-only memory or programs loaded from storage devices into random access memory. The random access memory also stores various programs and data required for the operation of the edge computing data collaborative processing and transmission scheduling device for urban rail transit. The processing units, read-only memory, and random access memory are interconnected via a bus. I / O interfaces are also connected to the bus. Typically, the following systems can be connected to the I / O interface: input devices including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices including, for example, magnetic tapes, hard disks, etc.; and communication devices. The communication device allows the edge computing data collaborative processing and transmission scheduling device for urban rail transit to communicate wirelessly or wiredly with other devices to exchange data. Although edge computing data collaborative processing and transmission scheduling devices for urban rail transit with various systems have been described, it should be understood that implementation or possession of all the described systems is not required. It can be implemented alternatively or with more or fewer systems.

[0051] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the edge computing data collaborative processing and transmission scheduling method for urban rail transit as described above. The computer program product provided by this invention can solve the technical problems in the prior art, such as mismatch between independent scheduling of computing power and transmission resources, poor robustness of fixed task fragmentation granularity, single optimization objective, rigid time slot allocation, and inability of the model to evolve online. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the edge computing data collaborative processing and transmission scheduling method for urban rail transit provided in the above embodiments, and will not be repeated here.

[0052] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a read-only memory. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0053] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0054] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the present invention and its equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A method for collaborative processing and transmission scheduling of edge computing data for urban rail transit, characterized in that, The methods include: Step S10: Collect multi-source heterogeneous data from the computing power domain, transmission domain, and business domain of the three-level edge nodes (vehicle-mounted, trackside, and station) in the urban rail scenario, perform spatiotemporal benchmark alignment and standardization mapping processing, and output the spatiotemporal standardized dataset and the edge node topology adjacency matrix set. Step S20: Based on the spatiotemporal standardized dataset and topological adjacency matrix set obtained in step S10, perform hierarchical label mapping according to business latency sensitivity, construct a computing power-transmission joint baseline library and extract multi-dimensional joint features, and output the joint baseline parameter set and business feature dataset. Step S30: Based on the joint baseline parameter set and business feature dataset output in step S20, perform task-granular adaptive sharding and calculate the collaborative adaptation degree of each edge node, filter and generate a candidate collaborative node set and a corresponding adaptation degree score set, and output the task sharding set and the candidate collaborative node set. Step S40: Based on the task sharding set and candidate collaborative node set output in step S30, construct a computing power-transmission joint scheduling utility model, solve for the optimal time slot allocation and transmission path planning scheme, and output the joint scheduling parameter set and path planning result set; Step S50: Based on the joint scheduling parameter set and path planning result set output in step S40, construct the full-link scheduling performance traceability chain, calculate the scheduling performance score and perform incremental parameter evolution updates, output the final scheduling results and model evolution parameters, and feed back and update the computing power-transmission joint scheduling performance model.

2. The edge computing data collaborative processing and transmission scheduling method for urban rail transit according to claim 1, characterized in that, Step S10, which outputs the spatiotemporally normalized dataset and the edge node topological adjacency matrix set, specifically includes: Step S101: Synchronously collect CPU computing power, memory capacity, and storage margin computing power domain data of each edge node; synchronously collect link bandwidth, end-to-end latency, packet loss rate, and handover interval transmission domain data; synchronously collect train control, passenger service, and equipment monitoring business domain data; and synchronously collect node ledger and topology connection relationship data. Step S102: Perform millisecond-level spatiotemporal alignment with a unified clock reference, perform dimensional normalization and outlier filtering on heterogeneous data, and remove invalid data segments with continuous out-of-synchronization duration exceeding a preset threshold. Step S103: Generate a spatiotemporally standardized dataset with unified spatiotemporal reference and standardized format, and synchronously output an edge node topology adjacency matrix set containing node adjacency relationships and link weight parameters.

3. The edge computing data collaborative processing and transmission scheduling method for urban rail transit according to claim 1, characterized in that, Step S20, which involves completing hierarchical label mapping based on service latency sensitivity, constructing a computing power-transmission joint baseline library and extracting multi-dimensional joint features, and outputting the joint baseline parameter set and service feature dataset, specifically includes: Step S201: Divide services into three levels—security, service, and monitoring—based on latency sensitivity, and map corresponding latency thresholds and resource guarantee priority labels to each type of service; Step S202: Classify and statistically analyze the labeled historical normal operation data according to service level, fit the joint distribution characteristics of computing power consumption and transmission delay for each service level, calculate the mean, covariance and boundary threshold, and generate a joint baseline library of computing power and transmission. Step S203: Extract four types of joint feature quantities from real-time data: computing power demand intensity, transmission bandwidth demand, latency margin, and packet loss sensitivity. Calculate feature statistics for each scheduling frame and integrate and output the joint baseline parameter set and the service feature quantity dataset.

4. The edge computing data collaborative processing and transmission scheduling method for urban rail transit according to claim 1, characterized in that, The steps in step S30, which involve performing task-granular adaptive sharding, calculating the collaborative fit of each edge node, filtering and generating a candidate collaborative node set and a corresponding fit score set, and outputting the task shard set and the candidate collaborative node set, specifically include: Step S301: Based on the business level label and the proportion of remaining computing power of the local node, combined with the total data volume of the task and the baseline data volume of the unit shard, round up to calculate the optimal sharding granularity of the task, and split the original task into several sub-shards that can be processed in parallel. Step S302: For each task fragment, traverse the adjacent edge nodes, and calculate the collaborative adaptation score by combining the three indicators of node computing power remaining rate, link available bandwidth ratio and transmission latency cost according to the preset weights. Step S303: Filter nodes with a collaboration fit greater than the preset fit threshold, generate a candidate collaboration node set, and synchronously output the split task fragment set and the corresponding fit score set.

5. The edge computing data collaborative processing and transmission scheduling method for urban rail transit according to claim 1, characterized in that, Step S40, which involves constructing a joint scheduling utility model for computing power and transmission, solving for the optimal time slot allocation and transmission path planning scheme, and outputting the joint scheduling parameter set and path planning result set, specifically includes: Step S401: For each task slice and its candidate node set, retrieve the full-dimensional joint feature quantity of the current scheduling frame as the model input, and construct the computing power-transmission joint scheduling utility function; Step S402: With the goal of maximizing the overall scheduling utility, solve for the computing power allocation, transmission time slot ratio, and optimal transmission path for each task fragment; the calculation of the overall scheduling utility for a single scheduling frame satisfies: in, For the total scheduling utility of a single scheduling frame, The total number of task fragments, For the first The priority weight of the business to which each shard belongs. For the first Each segment belongs to a service with latency requirements thresholds. For the first The total actual processing and transmission latency of each segment For the first The amount of resources occupied by each fragment. For the first The total resource amount of each shard corresponding to the node and link. For index value, , The utility weight coefficient is used to proportionally allocate the transmission slot length within a single scheduling frame based on the utility optimal solution, according to the priority weight of each fragment and the proportion of unit utility value. Step S403: Summarize the computing power allocation, time slot configuration and path planning results of all shards to generate a joint scheduling parameter set and a path planning result set.

6. The edge computing data collaborative processing and transmission scheduling method for urban rail transit according to claim 1, characterized in that, Step S50, which involves constructing a full-link scheduling performance traceability chain, calculating the scheduling performance score, performing incremental parameter evolution updates, outputting the final scheduling results and model evolution parameters, and feeding back and updating the computing power-transmission joint scheduling performance model, specifically includes: Step S501: Associate each scheduling result with the original data identifier, feature parameters, model version, and execution log to construct a full-link performance traceability chain from original data collection to scheduling execution; Step S502: Statistically calculate the latency compliance rate, task completion rate, and average resource utilization rate for a single scheduling cycle. Calculate the comprehensive scheduling performance score using preset weights. Based on the deviation between the performance score and the baseline threshold, perform incremental evolutionary updates to the baseline parameters and the scheduling model. The parameter update calculation satisfies the following: in, For model parameter update amount, For adaptive learning step size coefficients, Score the current scheduling efficiency. As the baseline performance threshold, These are the current model parameter values; Step S503: Output the final scheduling result with traceability chain, synchronously feed back the evolution parameters to the previous steps, and adaptively update the joint baseline parameter set and the computing power-transmission joint scheduling utility model.

7. The edge computing data collaborative processing and transmission scheduling method for urban rail transit according to claim 1, characterized in that, Step S50 also includes automatically including periodic data with scheduling performance scores greater than a preset qualified threshold into the labeled sample pool, and marking periodic data with scores lower than the preset qualified threshold as pending analysis status, which does not directly participate in parameter evolution.

8. An edge computing data collaborative processing and transmission scheduling system for urban rail transit, applied to the edge computing data collaborative processing and transmission scheduling method for urban rail transit as described in any one of claims 1 to 7, characterized in that, include: Multi-domain data spatiotemporal mapping module: used to collect multi-source heterogeneous data from computing power domain, transmission domain, and business domain of three-level edge nodes in urban rail transit scenarios (vehicle-mounted, trackside, and station-mounted), perform spatiotemporal benchmark alignment and standardization mapping processing, and output spatiotemporal standardized dataset and edge node topology adjacency matrix set; Business Classification and Joint Baseline Module: Based on the spatiotemporal standardized dataset and topological adjacency matrix set obtained by the multi-domain data spatiotemporal mapping module, it completes the classification label mapping according to the business latency sensitivity, constructs the computing power-transmission joint baseline library and extracts multi-dimensional joint features, and outputs the joint baseline parameter set and business feature dataset. Task sharding and node filtering module: Based on the joint baseline parameter set and business feature dataset output by the business classification and joint baseline module, it performs task-granular adaptive sharding and calculates the collaborative adaptation of each edge node, filters and generates a candidate collaborative node set and a corresponding adaptation score set, and outputs the task sharding set and the candidate collaborative node set. Joint scheduling slot planning module: Based on the task sharding set and candidate collaborative node set output by the task sharding and node screening module, it constructs a computing power-transmission joint scheduling utility model, solves the optimal slot allocation and transmission path planning scheme, and outputs a joint scheduling parameter set and path planning result set. The performance traceability and parameter evolution module is used to construct a full-link scheduling performance traceability chain based on the joint scheduling parameter set and path planning result set output by the joint scheduling time slot planning module, calculate the scheduling performance score and perform incremental parameter evolution updates, output the final scheduling results and model evolution parameters, and provide feedback and update the computing power-transmission joint scheduling performance model.

9. An edge computing data collaborative processing and transmission scheduling device for urban rail transit, characterized in that, include: The system includes a memory, a processor, and an edge computing data collaborative processing and transmission scheduling program for urban rail transit stored in the memory and executable on the processor. When the processor executes the edge computing data collaborative processing and transmission scheduling program for urban rail transit, it implements the edge computing data collaborative processing and transmission scheduling method for urban rail transit as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes an edge computing data collaborative processing and transmission scheduling program for urban rail transit. When the edge computing data collaborative processing and transmission scheduling program for urban rail transit is executed by the processor, it implements the edge computing data collaborative processing and transmission scheduling method for urban rail transit as described in any one of claims 1 to 7.

Images