Time-sensitive network-oriented side-end collaborative task scheduling strategy method

Through detailed evaluation and modeling of time-sensitive network resources, combined with uniformly distributed resource partition scheduling algorithm and experience pool mechanism, the problems of high latency and low efficiency in time-sensitive network are solved, and efficient resource utilization and system stability are achieved. , the effect of adapting to complex environments.

CN119012282AInactive Publication Date: 2024-11-22GUANGDONG VOCATIONAL COLLEGE OF POST & TELECOM

Patent Information

Application Number
CN202411077802.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-07
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The prior art has high latency, low efficiency, untrustworthiness and poor adaptability in end-end intelligent collaborative task scheduling, which is difficult to meet the real-time requirements of time-sensitive networks.

Method used

By conducting detailed evaluation and modeling of time-sensitive network resources, a resource partition scheduling algorithm and experience pool mechanism based on uniform distribution is designed, the computing node load is monitored in real time, the experience pool sampling ratio is adjusted, and the scheduling strategy is optimized to reduce delay and improve resource utilization. efficiency.

Benefits of technology

It realizes efficient allocation and utilization of resources in time-sensitive networks, reduces time delay, improves task response speed and system stability, enhances the credibility and adaptability of the model, and adapts to complex environments and business changes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119012282A_ABST
    Figure CN119012282A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of communication, in particular to a time-sensitive network-oriented edge-end collaborative task scheduling strategy method, which comprises the following steps of: evaluating various resources in a time-sensitive network in detail, establishing a resource model, reflecting the dynamic change and the use condition of the resources, and establishing a time-sensitive network resource model; according to the method, firstly, differentiated slices can be created according to the demand characteristics of time-sensitive industrial services, a uniform distribution-based slice resource partition scheduling algorithm is designed, and the use of wireless resources is optimized while the service resource demand is met by utilizing resource preemption and resource reservation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to an edge collaborative task scheduling strategy method for a time-sensitive network. Background Art

[0002] With the development of artificial intelligence and communication networks, a large number of new services with high real-time requirements have emerged. How to intelligently and collaboratively schedule tasks on the edge and reduce user latency has attracted widespread attention from researchers.

[0003] At present, research results are only aimed at certain local goals or specific scenarios, and have shortcomings such as inefficiency, unreliability, and poor adaptability. This project intends to be based on a new generation of artificial intelligence algorithms with the main goal of reducing user latency. To this end, we propose an edge collaborative task scheduling strategy method for time-sensitive networks to solve the above problems. Summary of the invention

[0004] The purpose of the present invention is to provide an edge collaborative task scheduling strategy method for time-sensitive networks to solve the problems raised in the above background technology.

[0005] The technical solution of the present invention is: a time-sensitive network-oriented edge collaborative task scheduling strategy method, comprising the following steps:

[0006] Step S1: Conduct a detailed evaluation of various resources in the time-sensitive network and establish a resource model to reflect the dynamic changes and usage of resources;

[0007] Step S2: Conduct in-depth analysis of the tasks to be scheduled to clarify their resource requirements and key attributes of latency requirements;

[0008] Step S3: Based on the resource model and task requirements, the resource requirements are evenly distributed on each mini-time slot within the time slot by establishing a scheduling process for periodic time-sensitive services;

[0009] Step S4: Design two experience pools. The purpose of the experience pools is to store and reuse historical scheduling decisions and their corresponding system status, performance indicators and other information, so that these experiences can be referenced or learned in future scheduling processes, thereby optimizing scheduling strategies and improving the efficiency and performance of task scheduling.

[0010] Step S5: design a data structure containing multiple fields to store each experience record, the fields including task identification, scheduling time, system resource status, scheduling decision, execution result and performance index;

[0011] Step S6: In terms of storage mode, memory storage, database storage or file storage mode is selected according to the actual situation;

[0012] Step S7: monitor the load of each computing node in real time;

[0013] Step S8: During the model training process, data is sampled from the two experience pools at the same time to update the strategy or model parameters. The credibility and generalization ability of the model are improved by adjusting the sampling ratio of the two experience pools.

[0014] Preferably, the exploratory experience to be stored is the experience generated when trying new strategies or actions, and the exploitative experience to be stored is the experience generated based on the current strategy or action.

[0015] Preferably, the resource partition scheduling algorithm based on uniform distribution follows three principles:

[0016] (1) Allocate resources to time-sensitive services after they arrive;

[0017] (2) The transmission of time-sensitive services must be within the maximum delay n deadline Internal transmission completed;

[0018] (3) Periodic delay-sensitive services complete data transmission based on resource usage.

[0019] Preferably, the monitoring of node load includes monitoring of CPU usage, memory occupancy and network bandwidth.

[0020] Preferably, the elimination mechanism needs to consider network congestion control, transmission protocol selection, traffic optimization, data packet scheduling strategy, clock synchronization, transmission path determinism and reliability factors.

[0021] Preferably, the various types of resources of the time-sensitive network include routers, switches, gateways, and repeaters.

[0022] Preferably, the sampling ratio samples data from two experience pools respectively, and adjusts the proportion of the two data sources.

[0023] Preferably, the resource modeling comprises the following steps:

[0024] Step S1.1: Model the tasks and resources in the network and clarify their attributes and constraints;

[0025] Step S1.2: Generate an initial scheduling strategy based on task requirements and resource conditions;

[0026] Step S1.3: Run the scheduling strategy in a simulation environment or in an actual network and collect performance data;

[0027] Step S1.4: Analyze and evaluate the collected performance data to determine whether the scheduling strategy meets the requirements;

[0028] Step S1.5: If the scheduling strategy does not meet the requirements, adjust and optimize it according to the evaluation results, repeat steps S1.2 to S1.4, and find the optimal scheduling strategy through an iterative optimization algorithm.

[0029] Preferably, the iterative optimization algorithm comprises the following steps:

[0030] Step S1.5.1: Initialization: Generate a set of initial solutions based on heuristic rules;

[0031] Step S1.5.2: Evaluation: Calculate the fitness value of each solution and evaluate each solution based on the objective function and constraints;

[0032] Step S1.5.3: Selection: Select a portion of solutions as candidate solutions for the next generation according to the fitness value. The selection process can be based on a roulette wheel selection strategy.

[0033] Step S1.5.4: Crossover: Perform a crossover operation on the selected solution to generate a new solution;

[0034] Step S1.5.5: Iteration: When the stopping condition is not met, repeat steps S1.5.2 to S1.5.4 until the maximum number of iterations is reached and the fitness value no longer changes significantly.

[0035] In step S6, for experience records that need to be accessed frequently, memory storage is used to improve access speed. For historical data that needs to be preserved for a long time, database or file storage is used. In terms of update strategy, new scheduling experience needs to be added to the experience pool regularly according to specific conditions, and at the same time, a cleanup or elimination mechanism for old experience is considered to maintain the timeliness and effectiveness of the experience pool.

[0036] The present invention provides an edge collaborative task scheduling strategy method for time-sensitive networks by improving the method, which has the following improvements and advantages compared with the prior art:

[0037] First, the present invention first creates differentiated slices based on the demand characteristics of time-sensitive industrial services, and designs a slice resource partition scheduling algorithm based on uniform distribution, using resource preemption and resource reservation to meet business resource requirements while optimizing wireless resource usage.

[0038] Second: The present invention, through detailed evaluation and modeling of resources, can make the system more accurately understand the dynamic changes and usage of resources, thereby optimizing the allocation and utilization of resources. Through evenly distributed resource demand scheduling, it can avoid the waste of reserved resources, reduce the occurrence of resource preemption, and improve the overall resource utilization efficiency. In-depth analysis of the resource demand and delay requirements of tasks can ensure that tasks are completed within a reasonable delay range. Through reasonable matching and scheduling of tasks, it can minimize unnecessary waiting time and improve the response speed of tasks. The credibility evaluation mechanism can ensure the stability and reliability of the model in complex environments, reduce system performance degradation or failures caused by environmental changes, and achieve By monitoring the load of each computing node in real time, potential performance bottlenecks or fault points can be discovered in time, and corresponding measures can be taken to repair or adjust them. The design of two experience pools can simultaneously consider the learning and exploration capabilities of the model, improve the model's generalization ability and ability to adapt to new environments, and update strategies or model parameters by sampling data from the two experience pools. This can more effectively utilize historical data and improve the learning efficiency and accuracy of the model. This process is not only suitable for the current network environment and task requirements, but can also be flexibly adjusted and expanded according to future changes. By updating resource models and scheduling strategies, it can adapt to new business needs and changes in the network environment, and maintain the system's continued competitiveness and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The present invention will be further explained below in conjunction with the accompanying drawings and embodiments:

[0040] Figure 1 It is a schematic diagram of the scheduling process of the periodic time-sensitive service of the present invention;

[0041] Figure 2 It is a schematic diagram of simulation parameters of the present invention;

[0042] Figure 3 Schematic diagram of the value of the bias N_bi as of the present invention changing with NTSN (rat io=30);

[0043] Figure 4 It is a schematic diagram of the value of the bias amount rat i o_bias of the present invention changing with rat io (NTSN=30). DETAILED DESCRIPTION

[0044] The present invention is described in detail below, and the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] The present invention provides an edge collaborative task scheduling strategy method for a time-sensitive network by improving the method. The technical solution of the present invention is:

[0046] like Figure 1 - Figure 4 As shown in FIG. 1 , the edge collaborative task scheduling strategy method for time-sensitive networks includes the following steps:

[0047] Step S1: Conduct a detailed evaluation of various resources in the time-sensitive network and establish a resource model to reflect the dynamic changes and usage of resources;

[0048] Step S2: Conduct in-depth analysis of the tasks to be scheduled to clarify their resource requirements and key attributes of latency requirements;

[0049] Step S3: Based on the resource model and task requirements, a scheduling process for periodic time-sensitive services is established to evenly distribute resource requirements on each mini-time slot within the time slot, thereby avoiding waste of reserved resources and minimizing the occurrence of resource preemption. arrival is the time when a single service arrives at the resource scheduler, n scheduled The time when the scheduler allocates physical resources RB and starts transmission, n end The time when the service transmission ends, n arrival To achieve the maximum acceptable latency for the service, the resource partition scheduling algorithm based on uniform distribution follows the following three principles:

[0050] Step S4: Design two experience pools. The purpose of the experience pools is to store and reuse historical scheduling decisions and their corresponding system status, performance indicators and other information, so that these experiences can be referenced or learned in future scheduling processes, thereby optimizing scheduling strategies and improving the efficiency and performance of task scheduling.

[0051] Step S5: design a data structure containing multiple fields to store each experience record, the fields including task identification, scheduling time, system resource status, scheduling decision, execution result and performance index;

[0052] Step S6: In terms of storage mode, memory storage, database storage or file storage mode is selected according to the actual situation. For experience records that need to be accessed frequently, memory storage is used to improve access speed. For historical data that needs to be preserved for a long time, database or file storage is used. In terms of update strategy, new scheduling experience needs to be added to the experience pool regularly according to specific conditions, and at the same time, a cleanup or elimination mechanism for old experience is considered to maintain the timeliness and effectiveness of the experience pool;

[0053] Step S7: monitor the load of each computing node in real time;

[0054] Step S8: During the model training process, data is sampled from the two experience pools at the same time to update the strategy or model parameters. The credibility and generalization ability of the model are improved by adjusting the sampling ratio of the two experience pools.

[0055] Furthermore, the exploratory experience used to store is the experience generated when trying new strategies or actions, and the exploitative experience used to store is the experience generated based on the current strategy or action. In reinforcement learning or similar problem solving processes, balancing exploration and exploitation is a key challenge. Exploration means trying new strategies or actions to find better solutions, while exploitation is based on the known best strategy or action to maximize benefits. By storing these two types of experience separately, the system can manage the two strategies more effectively and ensure that they can be balanced when needed. When the system needs to quickly adapt to new environments or tasks, the exploitative experience can quickly provide effective feedback based on the current strategy. At the same time, exploratory experience can provide inspiration and reference when the system needs improvement or innovation. By storing these two types of experience separately, the system can adjust its learning strategy according to needs more quickly. In a complex decision-making environment, the system may need to choose different strategies according to different situations. By storing exploratory experience and exploitative experience separately, the system can more easily evaluate the effectiveness of each strategy and make more informed decisions based on these experiences. By storing exploratory experience, the system can accumulate knowledge about different strategies or actions, thereby improving its generalization ability, which means that when the system encounters new and unseen situations, it can use these experiences to quickly adapt and find effective solutions. By storing exploratory experience and exploitative experience separately, the system can more easily evaluate the performance of different strategies and compare them, which helps the system select the best strategy and continuously improve its decision-making process.

[0056] Furthermore, the resource partition scheduling algorithm based on uniform distribution follows three principles:

[0057] (1) Allocate resources to time-sensitive services after they arrive;

[0058] (2) The transmission of time-sensitive services must be within the maximum delay n deadline Internal transmission completed;

[0059] (3) Periodic delay-sensitive services complete data transmission based on resource usage. When real-time or time-sensitive services (such as urgent communication requests, financial transactions, etc.) arrive, the algorithm can quickly allocate resources for them. Immediacy ensures that key services can be executed without delay, thus meeting the service's requirements for time sensitivity. The second principle ensures that the transmission of time-sensitive services is completed within the maximum delay. This is crucial for services that require strict time guarantees, such as video streaming and online games. By controlling the maximum delay, the algorithm can maintain or improve user experience and prevent service interruptions. Periodic delay-sensitive services (such as periodic reports, periodic data transmission, etc.) complete data transmission based on resource usage. Flexibility enables the algorithm to use network resources more efficiently. When the network load is low, these services can be completed faster, while when the network is busy, they may need to wait more time. This dynamic adjustment helps avoid resource waste and ensures fair allocation of network resources. The evenly distributed resource partition scheduling algorithm helps reduce the risk of system congestion and overload. By allocating resources reasonably, the algorithm can ensure that critical tasks are processed in a timely manner while reducing the occupation of system resources by non-critical tasks. This balance improves the overall stability and reliability of the system. The resource partition scheduling algorithm based on even distribution usually has a relatively simple implementation logic, making it easy to deploy and manage at both the hardware and software levels. In addition, because its rules are clear and easy to understand, the algorithm is also easy to debug and optimize. As the network environment and business needs change, the resource partition scheduling algorithm based on even distribution can adapt to these changes by adjusting parameters or introducing new strategies. This scalability and adaptability enable the algorithm to maintain its effectiveness and efficiency in different scenarios.

[0060] Furthermore, monitoring the node load includes monitoring the CPU usage, memory usage, and network bandwidth. By monitoring these key performance indicators in real time, system administrators or operation and maintenance personnel can quickly discover performance bottlenecks or potential problems. For example, if the CPU usage of a node is continuously too high, it may mean that the node needs to be expanded or optimized. Understanding the node load helps to allocate resources more reasonably. If the memory usage of a node is low and the CPU usage is high, it may be possible to consider allocating more computing tasks to the node while maintaining the load balance of other nodes. By long-term monitoring and analyzing historical data, potential failures or performance degradation can be predicted and prevented in advance. This helps reduce the impact of system failures on the business and improve the stability and availability of the system. By optimizing resource allocation and timely discovering performance bottlenecks, it can ensure that the system always runs in the best state, thereby improving the overall efficiency of the system. Accurate load data can provide decision support for management. Expansion plans and resource procurement plans based on load data can be more scientific and reasonable. By optimizing system performance, it can ensure that users have a better experience when using the system and reduce delays and freezes caused by system performance problems.

[0061] Furthermore, the elimination mechanism needs to consider network congestion control, transmission protocol selection, traffic optimization, packet scheduling strategy, clock synchronization, transmission path determinism and reliability factors. Congestion control can dynamically adjust the data transmission rate according to the degree of network congestion to avoid further aggravation of network congestion, thereby improving the overall performance and throughput of the network, avoiding data loss or transmission errors caused by network congestion, and ensuring reliable data transmission. By adjusting the data transmission rate, the data transmission delay caused by network congestion can be reduced, and the response speed and real-time performance of the network can be improved. By selecting the appropriate transmission protocol, the reliability and real-time performance of data transmission can be ensured according to the needs of the application scenario. By optimizing the information flow, information can be conveyed to the target audience more quickly and accurately. By optimizing the traffic, the information processing cost can be reduced and the information utilization efficiency can be improved. , fairly allocate network bandwidth according to the degree of network congestion, ensure that each terminal host can share network resources fairly, and through intelligent data packet scheduling strategies, avoid some terminal hosts occupying too much bandwidth and causing other terminal hosts to be unable to use the network normally. In network communication, clock synchronization can ensure the consistency of data packet timestamps, thereby avoiding data transmission errors. Many network communication protocols require the use of timestamps for timing and control. Clock synchronization can ensure the normal operation of these protocols. Timestamps are used for identity authentication and data integrity verification in network security. Clock synchronization can enhance the effectiveness of these security mechanisms. Selecting a stable and reliable transmission path can further ensure the integrity and correctness of data transmission. In the presence of noise, interference, network congestion, etc., selecting the optimal transmission path can reduce the risk of data loss and errors.

[0062] Furthermore, the various resources of a time-sensitive network include routers, switches, gateways, and repeaters. By evaluating routers, we can understand the router's forwarding capabilities, the number of network protocols it supports, and the performance of functions such as network address translation (NAT), which helps ensure that the router can efficiently and stably forward data packets in the time-sensitive network. Switch evaluation can reveal its capabilities in network traffic control and load balancing, ensuring that data can be transmitted quickly and accurately in the time-sensitive network. Gateway evaluation can ensure that data conversion between different network protocols is efficient and accurate, which is crucial for data flow in the time-sensitive network. Repeater evaluation can determine whether it can effectively expand the communication distance, reduce signal attenuation, and ensure the coverage of the time-sensitive network.

[0063] Furthermore, the sampling ratio samples data from each of the two experience pools and adjusts the proportion of the two data sources. Data is sampled from the two experience pools to update the strategy or model parameters. The sampling ratio determines the proportion of the data from each experience pool to the total number of samples. By adjusting the ratio, the direction and effect of model training are affected, thereby improving the credibility and generalization ability of the model.

[0064] Further, resource modeling includes the following steps:

[0065] Step S1.1: Model the tasks and resources in the network and clarify their attributes and constraints;

[0066] Step S1.2: Generate an initial scheduling strategy based on task requirements and resource conditions;

[0067] Step S1.3: Run the scheduling strategy in a simulation environment or in an actual network and collect performance data;

[0068] Step S1.4: Analyze and evaluate the collected performance data to determine whether the scheduling strategy meets the requirements;

[0069] Step S1.5: If the scheduling strategy does not meet the requirements, it is adjusted and optimized according to the evaluation results, and steps S1.2 to S1.4 are repeated. The optimal scheduling strategy is found through an iterative optimization algorithm. First, by modeling the tasks and resources in the network, their attributes and constraints can be clearly defined, which provides a solid foundation for the subsequent scheduling strategy formulation. In this way, during the scheduling process, it can be ensured that the tasks can be executed according to the predetermined requirements while avoiding waste and conflicts of resources. Secondly, the initial scheduling strategy generated based on task requirements and resource conditions provides a feasible solution for actual operations. Although this initial solution may not be optimal, it provides a starting point for subsequent optimization and adjustment. Then, running the scheduling strategy in a simulation environment or an actual network and collecting performance data are key steps in verifying and optimizing the scheduling strategy. Step 1. These data can reflect the performance of the scheduling strategy in actual operation, including key indicators such as task completion time and resource utilization. Through these data, the effect of the scheduling strategy can be objectively evaluated. Next, the collected performance data is analyzed and evaluated to determine whether the scheduling strategy meets the requirements. If the scheduling strategy does not meet the requirements, it needs to be adjusted and optimized according to the evaluation results. This process is iterative. Through continuous adjustment and optimization, the optimal scheduling strategy can be gradually approached. Finally, the optimal scheduling strategy is found through the iterative optimization algorithm, which can greatly improve the efficiency and accuracy of network scheduling. This optimal strategy can maximize the use of network resources while meeting the needs of tasks. This can not only improve the overall performance of the network, but also reduce operating costs, bringing greater value to enterprises and users.

[0070] Further, the iterative optimization algorithm comprises the following steps:

[0071] Step S1.5.1: Initialization: Generate a set of initial solutions based on heuristic rules;

[0072] Step S1.5.2: Evaluation: Calculate the fitness value of each solution and evaluate each solution based on the objective function and constraints;

[0073] Step S1.5.3: Selection: Select a portion of solutions as candidate solutions for the next generation according to the fitness value. The selection process can be based on a roulette wheel selection strategy.

[0074] Step S1.5.4: Crossover: Perform a crossover operation on the selected solution to generate a new solution;

[0075] Step S1.5.5: Iteration: When the stopping condition is not met, repeat steps S1.5.2 to S1.5.4 until the maximum number of iterations is reached and the fitness value no longer changes significantly. First, a set of initial solutions based on heuristic rules is generated in the initialization phase, which can make the algorithm have a certain degree of intelligence from the beginning and avoid the disadvantages of blind search. Secondly, the evaluation phase calculates the fitness value of each solution and evaluates it according to the objective function and constraints, which can ensure that the algorithm always moves towards the optimization goal. The selection phase uses strategies based on fitness values, such as roulette selection, to screen out better solutions as candidate solutions for the next generation, which helps to accelerate the convergence speed of the algorithm. The crossover phase can generate new solutions by performing crossover operations on the selected solutions, which increases the search range and diversity of the algorithm and helps to jump out of the local optimal solution and find the global optimal solution. Finally, the iteration phase continuously repeats the evaluation, selection and crossover steps until the stopping condition is met, such as reaching the maximum number of iterations or the fitness value no longer changes significantly, which ensures that the algorithm can fully search the solution space and find the best possible solution.

[0076] The above description enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. The edge collaborative task scheduling strategy method for time-sensitive networks is characterized by: The following steps are involved: Step S1: Conduct a detailed evaluation of various resources in the time-sensitive network and establish a resource model to reflect the dynamic changes and usage of resources; Step S2: Conduct in-depth analysis of the tasks to be scheduled to clarify their resource requirements and key attributes of latency requirements; Step S3: Based on the resource model and task requirements, the resource requirements are evenly distributed on each mini-time slot within the time slot by establishing a scheduling process for periodic time-sensitive services; Step S4: Design two experience pools. The purpose of the experience pools is to store and reuse historical scheduling decisions and their corresponding system status and performance indicator information so that these experiences can be referenced or learned in future scheduling processes, thereby optimizing scheduling strategies and improving the efficiency and performance of task scheduling. Step S5: design a data structure containing multiple fields to store each experience record, the fields including task identification, scheduling time, system resource status, scheduling decision, execution result and performance index; Step S6: In terms of storage mode, memory storage, database storage or file storage mode is selected according to the actual situation; Step S7: monitor the load of each computing node in real time; Step S8: During the model training process, data is sampled from the two experience pools at the same time to update the strategy or model parameters. The credibility and generalization ability of the model are improved by adjusting the sampling ratio of the two experience pools.

2. The edge collaborative task scheduling strategy method for time-sensitive networks according to claim 1 is characterized in that: The exploratory experience stored is the experience generated when trying new strategies or actions, and the exploitative experience stored is the experience generated based on the current strategy or action.

3. The edge collaborative task scheduling strategy method for time-sensitive networks according to claim 1 is characterized in that: The resource partition scheduling algorithm based on uniform distribution follows three principles: (1) Allocate resources to time-sensitive services after they arrive; (2) The transmission of time-sensitive services must be within the maximum delay n deadline Internal transmission completed; (3) Periodic delay-sensitive services complete data transmission based on resource usage.

4. The edge collaborative task scheduling strategy method for time-sensitive networks according to claim 1 is characterized in that: The monitoring node load includes monitoring of CPU usage, memory occupancy and network bandwidth.

5. The edge collaborative task scheduling strategy method for time-sensitive networks according to claim 1 is characterized by: The elimination mechanism needs to consider network congestion control, transmission protocol selection, traffic optimization, data packet scheduling strategy, clock synchronization, transmission path determinism and reliability factors.

6. The edge collaborative task scheduling strategy method for time-sensitive networks according to claim 1 is characterized by: The various resources of the time-sensitive network include routers, switches, gateways, and repeaters.

7. The edge collaborative task scheduling strategy method for time-sensitive networks according to claim 1 is characterized by: The sampling ratio samples data from two experience pools respectively, and adjusts the proportions of the two data sources.

8. The edge collaborative task scheduling strategy method for time-sensitive networks according to claim 1 is characterized by: The resource modeling includes the following steps: Step S1.1: Model the tasks and resources in the network and clarify their attributes and constraints; Step S1.2: Generate an initial scheduling strategy based on task requirements and resource conditions; Step S1.3: Run the scheduling strategy in a simulation environment or in an actual network and collect performance data; Step S1.4: Analyze and evaluate the collected performance data to determine whether the scheduling strategy meets the requirements; Step S1.5: If the scheduling strategy does not meet the requirements, adjust and optimize it according to the evaluation results, repeat steps S1.2 to S1.4, and find the optimal scheduling strategy through an iterative optimization algorithm.

9. The edge collaborative task scheduling strategy method for time-sensitive networks according to claim 8 is characterized by: The iterative optimization algorithm comprises the following steps: Step S1.5.1: Initialization: Generate a set of initial solutions based on heuristic rules; Step S1.5.2: Evaluation: Calculate the fitness value of each solution and evaluate each solution based on the objective function and constraints; Step S1.5.3: Selection: Select a portion of solutions as candidate solutions for the next generation according to the fitness value. The selection process can be based on a roulette wheel selection strategy. Step S1.5.4: Crossover: Perform a crossover operation on the selected solution to generate a new solution; Step S1.5.5: Iteration: When the stopping condition is not met, repeat steps S1.5.2 to S1.5.4 until the maximum number of iterations is reached and the fitness value no longer changes significantly.

10. The edge collaborative task scheduling strategy method for time-sensitive networks according to claim 9 is characterized in that: In step S6, for experience records that need to be accessed frequently, memory storage is used to improve access speed. For historical data that needs to be preserved for a long time, database or file storage is used. In terms of update strategy, new scheduling experience needs to be added to the experience pool regularly according to specific conditions, and at the same time, a cleanup or elimination mechanism for old experience is considered to maintain the timeliness and effectiveness of the experience pool.

Citation Information

Patent Citations

  • Cluster task scheduling method and system based on deep reinforcement learning

    CN115904652A

  • Downlink wireless resource scheduling method and device for 5G-TSN fusion deployment scene

    CN116234018A

  • 5G-TSN fusion network slice management method and system based on deep reinforcement learning

    CN116582855A

  • SDN enhanced path allocation device and method employing bin-packing technique

    WO2020181761A1

  • Delay deterministic transmission method based on route scheduling and joint optimization

    WO2023004898A1

Cited By

  • Offshore computing task unloading method based on resource reservation mechanism and medium

    CN119883662A

  • Energy acquisition network-oriented double-layer dynamic information updating method and product

    CN120151946A