A method and system for dynamically changing online scalable configuration of time-sensitive networks

Abstract

This invention discloses an online scalable configuration method and system for dynamically changing time-sensitive networks (TSNs), relating to the Internet of Things (IoT) field. The invention proposes an efficient inter-flow conflict detection mechanism based on multi-level flow grouping. By using a correlation analysis of flow period and offset, it avoids traditional conflict detection methods based on link maximal cliques and link hyperperiods, enabling rapid response to large-scale data flow conflict detection needs. It designs an online incremental method based on decoupling routing and scheduling, as well as offline pre-routing, accelerating the computation speed of routing and scheduling schemes while ensuring scheduling space. This invention reduces the computational complexity of data flow set hyperperiod sensitivity and the maximum link slot occupancy in TSNs, accelerating online scheduling and providing efficient conflict detection support for online scalable configuration.

Description

Technical Field

[0001] This invention relates to the field of the Internet of Things (IoT), and more particularly to an online scalable configuration method and system for time-sensitive networks that dynamically change. Background Technology

[0002] With the rapid development of the Industrial Internet of Things (IIoT), more and more IIoT applications are placing higher demands on the quality of service (QoS) of network communication, such as latency, packet loss, and jitter. Traditional Ethernet technology, due to its "best-effort" transmission mode, is prone to packet loss and congestion, making it unsuitable for IIoT applications with high communication requirements. Therefore, establishing a "timely and accurate" deterministic network has become a research hotspot. In industry, several well-known industrial automation organizations have designed several real-time industrial Ethernet technologies with deterministic and time-sensitive characteristics. For example, PROFINET and EtherNet / IP network technologies, respectively led by Siemens and Rockwell, are widely used in factories worldwide. However, due to commercial competition and technological isolation, these network technologies are closed to each other, resulting in poor interoperability. To overcome these limitations, Time-Sensitive Networking (TSN), as an open industrial communication network, has attracted widespread attention in industry. It endows traditional Ethernet with real-time and deterministic characteristics by defining a series of clock synchronization, scheduling, and configuration mechanisms.

[0003] In recent years, research on TSN technology has mainly focused on the design of scheduling mechanisms, with two commonly used mechanisms being the Time-aware Shaper (TAS) and the Cyclic Queuing and Forwarding (CQF). The former emphasizes fine-grained scheduling with stricter latency and jitter requirements, while the latter offers a simpler scheduling configuration with latency and jitter within a controllable range. These two scheduling mechanisms only define the gating behavior of the queues; the specific design of the gating lists still requires further research.

[0004] For the gating list design problem, existing scheduling work mainly relies on two modeling approaches: satisfiability modulo theories (SMT) and integer linear programming (ILP) to construct deterministic real-time constraint sets, thus transforming the problem into constraint satisfaction. This type of problem is equivalent to the 0-1 bin packing problem (NP-hard, exponential complexity), with high computational complexity and time-consuming brute-force solutions, making it unsuitable for medium- to large-scale scheduling scenarios. To improve scalability, some works have proposed iterative grouping scheduling methods that compress global conflicts into group conflicts, balancing schedulability and scalability by designing different grouping principles, which reduces computational complexity to some extent. Other works have proposed incremental scheduling methods, essentially the same as iterative grouping scheduling, except that the number of data streams in each group is set to 1; both are greedy scheduling methods. However, these scheduling methods only focus on reducing the scheduling scale, neglecting the complexity bottleneck of conflict detection between scheduling data streams, resulting in still time-consuming synthesis of scheduling schemes, some of which take several hours in complex scenarios. It is clear that these offline scheduling methods have high scheduling time costs and cannot meet the needs of dynamic scenarios with high response requirements, such as second-level / hundred-millisecond-level time response.

[0005] Furthermore, routing planning is another means of network configuration management; it is the process of determining the transmission paths of various data flows in the network. This includes determining the optimal path for data flows from source to destination, taking into account factors such as network topology, link status, and traffic load. The result of routing planning is the prior input of the scheduling method, and the two work together to achieve network efficiency, reliability, and adaptability. Existing TSN routing and scheduling work is mainly divided into two categories: one is to design routing and scheduling separately, where the routing phase focuses on achieving the shortest data flow hop count or the lowest network link load, without considering the impact of the routing result on the scheduling phase, which greatly limits the overall scheduling space; the other is to jointly design routing planning and transmission scheduling, which involves many optimization variables and has very high complexity, making it unsuitable for medium-to-large-scale network scenarios and dynamically changing scenarios. Therefore, how to design an online scalable configuration (routing + scheduling) method that can quickly respond to time-sensitive network dynamic changes while ensuring high scalability is a key problem that urgently needs to be solved.

[0006] A review of existing literature revealed that the most similar implementation is Chinese patent application number 202211204257.X, entitled "Time-Sensitive Traffic Online Scheduling Method and Apparatus Based on Deep Reinforcement Learning." Its specific approach involves: using deep reinforcement learning, inputting network resource configuration, network topology, and pre-scheduled traffic information to establish a system model. The system model is then updated using online traffic information, and hyperparameters of the traffic scheduling model are selected. Next, a traffic scheduling action model and a reward model are constructed, and the optimal traffic scheduling action is selected. Then, the queue resource information in the system model and the network parameters of the traffic scheduling model are updated. Finally, the traffic scheduling planning results are distributed to the gating lists of each switch. However, this method, utilizing deep reinforcement learning for scheduling, requires a large number of samples for training, resulting in a long training time and an inability to guarantee rapid response to dynamic changes during actual operation. The patent application, number 202010741539.8, entitled "A Traffic Shaping and Routing Planning and Scheduling Method for Time-Sensitive Network Gating Mechanism," describes a method that divides the output port queues of each switch into two categories: Time-Sensitive (TT) queues and non-TT queues. First, it uses the modulo satisfying theory to perform routing planning and transmission scheduling for TT data flows, generating a TT queue gating list. Based on this, it designs the gating list for the non-TT queues, ultimately synthesizing the output queues (GCLs) of each switch port. Then, it uses relevant configuration software to generate configuration files for the switch ports and terminal devices, configuring them on each switch and terminal device. However, this method employs a global scheduling approach, resulting in high scheduling time costs and making it difficult to apply in medium-to-large-scale and dynamically changing scenarios. Patent application number 202310590065.5, entitled "Time-Sensitive Network Traffic Scheduling Method and Apparatus," describes a method that involves: inputting the network state into a state information encoding module to encode the network state; inputting the state encoding into a multi-task policy model to obtain a target scheduling policy; wherein the target scheduling policy includes a ranking policy and a routing policy; and the multi-task policy model is trained based on a deep neural network using traffic scheduling samples and an enhancement policy reinforcement learning algorithm. However, this method employs a model-free reinforcement learning approach, with each training session taking hours, and its scheduling performance is highly sensitive to parameter settings, making it unsuitable for scenarios with dynamic network changes and high response requirements.

[0007] Therefore, those skilled in the art are dedicated to developing an online scalable configuration method and system for time-sensitive networks that dynamically changes, quickly responds to the need for large-scale data flow conflict detection, and accelerates the calculation speed of routing and scheduling schemes while ensuring scheduling space. Summary of the Invention

[0008] In view of the above-mentioned deficiencies of the prior art, the technical problem to be solved by the present invention is to configure routing and scheduling online in a scalable manner, and to respond quickly to time-sensitive network dynamic changes while ensuring high schedulability.

[0009] To achieve the above objectives, this invention provides an online scalable configuration system for time-sensitive network dynamic changes, including a network input module, an offline pre-routing module, and an online configuration module. The network input module is responsible for generating network topology, storing old routes and scheduling schemes, and inputting network parameters. It is also responsible for updating network dynamic changes, including adding and deleting terminal devices, changing data flow terminals, and thus updating the changed data flow set. The offline pre-routing module generates a set of candidate routing paths between network nodes based on the network topology from the network input module, serving as the prior routing input for the online configuration module. The online configuration module includes an online routing unit, an online scheduling unit, a schedulability analysis unit, an inter-flow conflict detection unit, and a deployment unit. The online routing unit generates a set of routing paths for new data flow sets based on the candidate path set between network nodes and inputs it into the online scheduling unit. The online scheduling unit is responsible for generating a scheduling scheme for the new data flow set. The schedulability analysis unit and the inter-flow conflict detection unit are the basic support for the online routing unit and the online scheduling unit. The deployment unit is responsible for converting the routing and scheduling schemes into configuration instructions that can be recognized by terminal devices and network devices.

[0010] This invention also provides an online scalable configuration method for dynamically changing time-sensitive networks, comprising the following steps:

[0011] Step 1: Initialize the offline pre-routing module, set the number of candidate paths, and input the information into the online configuration module;

[0012] Step 2: Based on dynamic network changes, update the terminal addition / deletion status and data stream set changes, and input the new data stream set, deleted data stream set, original data stream set, original routing and scheduling scheme, network bandwidth and time slot size network parameters into the online configuration module;

[0013] Step 3: Call the schedulability analysis unit to determine the latency condition of the new data stream set. If the condition is met, activate the online routing unit to select the optimal path for the new data stream set; otherwise, adjust the time slot size to meet the latency condition.

[0014] Step 4: Based on the optimal path set of data flow calculated by the online routing unit, update the data flow set traversed by each link in the network, call the schedulability analysis unit to perform conflict judgment on the data flow set of each link, and if the conflict is satisfied, return "yes"; otherwise, return "no".

[0015] Step 5: If yes, call the online scheduling unit to generate a scheduling scheme; otherwise, the data stream set is not schedulable.

[0016] Step 6: Invoke the deployment unit to convert the routing and scheduling scheme into configuration instructions, and send them to network devices and terminal devices to update and take effect;

[0017] Step 7: Update the network link status and the new routing and scheduling scheme to the network input module.

[0018] Furthermore, the efficient inter-flow conflict detection mechanism based on multi-level flow grouping avoids conflict detection methods based on link maximal cliques and link overperiods by using a correlation analysis of flow period and offset.

[0019] Furthermore, the link data stream is grouped into multiple levels based on the data stream period and transmission offset. For any two data streams belonging to different first-level groups, there is no conflict as they do not occupy the same time slot. Similarly, for any two data streams belonging to different third-level groups, there is no conflict as they do not occupy the same time slot. For any two data streams belonging to different second-level groups, there is at least one conflict as they occupy the same time slot.

[0020] Furthermore, the multi-level grouping principle is as follows: the first-level flow grouping label of the data stream is equal to the remainder of the link offset divided by the greatest common divisor of the periods of all data streams in the link; the second-level grouping of the link is implemented based on the data stream period analysis. First, the processing period set of the corresponding link data stream is obtained, then the maximum processing period of the link data stream is calculated, and prime numbers in the range of 1 to the maximum processing period are found. Each prime number represents a second-level group. If the prime number corresponding to the second-level group is the prime number that is the largest divisor of the data stream processing period, the data stream is assigned to the second-level group. The data stream with a processing period of 1 is assigned to a separate second-level group; the third-level flow grouping label of the data stream is equal to the remainder obtained by dividing the link offset by the quotient of the greatest common divisor of the periods of all data streams in the link and then dividing by the processing period of the second-level group in which it belongs.

[0021] Furthermore, based on the decoupling of routing and scheduling and the online incremental method of offline pre-routing, the computation speed of routing and scheduling schemes is accelerated while ensuring scheduling space.

[0022] Furthermore, the routing and scheduling are decoupled by constructing a link load balancing metric that reflects flow scheduling to guide the optimal online route. The link load balancing metric is based on period analysis and secondary grouping, which aggregates data flows with multiples of each other in period on the same link and distributes data flows with coprime periods on different links.

[0023] Furthermore, the offline pre-routing provides the online routing with a set of shortest candidate paths between network nodes. For new data flows, it is not necessary to re-invoke the loop-free routing algorithm to traverse the network topology nodes. Instead, it is only necessary to check the directly linked network nodes of the terminal devices corresponding to the new data flows and quickly filter out the set of selectable paths based on real-time constraints.

[0024] Furthermore, schedulability analysis conditions are constructed from the perspectives of transmission delay and flow conflict, including delay judgment conditions and conflict judgment conditions. The analysis is conducted to determine whether the data stream set can be scheduled under the best conditions to satisfy real-time performance and conflict-free operation.

[0025] Furthermore, the latency judgment condition is whether the minimum network forwarding latency of the new data stream is less than or equal to the corresponding deadline, wherein the minimum network forwarding latency is equal to the product of the minimum routing hop count and the time slot size. If the data stream set does not meet the latency judgment condition, the time slot size can be adjusted downward. The conflict judgment condition is whether the maximum network link utilization of all data streams is less than or equal to the network bandwidth, wherein the network link utilization is equal to the sum of the ratios of the frame size to the period of all data streams on the link.

[0026] In a preferred embodiment of the present invention, considering that most existing TSN scheduling methods reduce computational complexity by decreasing the scale of each scheduling operation, they neglect the complexity bottleneck of inter-flow conflict detection, resulting in long processing times in complex scenarios and an inability to respond to dynamic network changes in real time. This invention proposes an efficient inter-flow conflict detection mechanism based on multi-level flow grouping. Through a correlation analysis of flow period and offset, it avoids traditional conflict detection methods based on link maximal cliques and link over-periods, enabling rapid response to large-scale data flow conflict detection needs. Supported by integer division theory, this invention performs multi-level grouping of link data flows based on data flow period and transmission offset. For any two data flows belonging to different first-level groups, they do not occupy the same time slot, i.e., they do not conflict; for any two data flows belonging to different third-level groups, they also do not occupy the same time slot. For any two data flows belonging to different second-level groups, they occupy at least the same time slot, i.e., they conflict. The multi-level grouping principle is described as follows: the first-level flow group label of a data flow is equal to the remainder of the link offset divided by the greatest common divisor of the periods of all data flows in the link. Link-level grouping is implemented based on data flow cycle analysis. First, the processing cycle set of the corresponding link data flow is obtained (see the embodiment definition: the integer cycle of the link data flow is equal to the quotient of the data flow cycle divided by the time slot size; the corresponding processing cycle is equal to the quotient of the integer cycle divided by the time slot size, which is equal to the greatest common divisor of the integer cycles of all data flows in the link). Then, the maximum processing cycle of the link data flow is calculated, and prime numbers within the range of 1 to the maximum processing cycle are identified. Each prime number represents a secondary group. If the prime number corresponding to the secondary group is the largest divisible prime number of the data flow processing cycle, the data flow is assigned to that secondary group. Furthermore, data flows with a processing cycle of 1 are assigned to a separate secondary group. The tertiary flow grouping label of a data flow is equal to the remainder obtained by dividing the link offset by the quotient of the greatest common divisor of the cycles of all data flows in the link, and then dividing by the processing cycle of the corresponding secondary group.

[0027] Existing TSN routing and scheduling methods fall into two categories: The first is the joint design method, which involves numerous routing and scheduling variables, resulting in high computational complexity and long processing times, making it unsuitable for medium-to-large-scale network scenarios and dynamically changing network conditions. The second is the separate design method, which employs brute-force decoupling, ignoring the inherent relationship between routing planning and transmission scheduling, severely limiting the overall scheduling space and consequently restricting the schedulable performance of the TSN network. This invention designs an online incremental method based on routing and scheduling decoupling and offline pre-routing, accelerating the computation speed of routing and scheduling schemes while ensuring sufficient scheduling space. The routing and scheduling decoupling method of this invention involves constructing a link load balancing metric reflecting flow scheduling to guide online optimal routing. This metric is based on periodic analysis and two-level grouping, aggregating data flows with periodic multiples on the same link and distributing data flows with coprime periods on different links, thereby improving schedulable performance. This is because coprime period data flows are more prone to conflict and have a smaller scheduling space. Offline pre-routing provides online routing with a set of K shortest candidate paths between network nodes. For new data flows, it is not necessary to re-invoke the loop-free routing algorithm to traverse the network topology nodes. Instead, it is only necessary to check the directly linked network nodes of the terminal devices corresponding to the new data flows and quickly filter out the set of selectable paths based on real-time constraints.

[0028] Existing TSN scheduling technologies focus on algorithm design and scheme generation, lacking schedulability analysis of data flow sets under a given network topology, and directly proceed to scheduling implementation. In fact, schedulability analysis plays a preliminary screening role in this process, removing unschedulable data flow sets, effectively avoiding resource and time waste. This invention derives schedulability analysis conditions for new data flow sets, filtering out unschedulable data flow sets under current network parameters, and also providing guidance for adjusting network parameters such as time slot size. This invention constructs schedulability analysis conditions from the perspectives of transmission delay and flow conflict, analyzing whether a data flow set can be scheduled under the best-case scenario, i.e., satisfying real-time performance and conflict-free operation. The delay judgment condition is whether the minimum network forwarding delay of the new data flow is less than or equal to the corresponding deadline, where the minimum network forwarding delay equals the product of the minimum routing hop count and the time slot size. If the data flow set does not meet the delay judgment condition, the time slot size can be adjusted downwards. The conflict judgment condition is whether the maximum network link utilization of all data flows is less than or equal to the network bandwidth, where the network link utilization equals the sum of the ratios of frame size to period of all data flows on that link.

[0029] Compared with the prior art, the present invention has the following obvious substantive features and significant advantages:

[0030] 1. This invention reduces the sensitivity to data stream set timeouts on the one hand, and greatly reduces the computational complexity of the maximum link time slot occupancy in time-sensitive networks on the other hand, thereby accelerating online scheduling and providing efficient conflict detection support for online scalable configuration.

[0031] 2. This invention has high scalability and high schedulability, meeting the real-time response requirements of dynamic changes in industrial time-sensitive networks.

[0032] 3. As a preliminary screening process, this invention helps to focus efforts on developing promising scheduling strategies, thereby avoiding unnecessary search and computational burdens and saving resource and time costs.

[0033] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of an online scalable configuration system structure according to a preferred embodiment of the present invention;

[0035] Figure 2 This is a flowchart of the operation of an online routing unit according to a preferred embodiment of the present invention;

[0036] Figure 3 This is a flowchart of the online scheduling unit operation according to a preferred embodiment of the present invention;

[0037] Figure 4 This is a schematic diagram of multi-level flow grouping of a flow conflict detection unit according to a preferred embodiment of the present invention. Detailed Implementation

[0038] The following description, with reference to the accompanying drawings, illustrates several preferred embodiments of the present invention to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0039] In the accompanying drawings, components with the same structure are indicated by the same numerical designation, and components with similar structures or functions are indicated by similar numerical designations. The dimensions and thicknesses of each component shown in the drawings are arbitrary, and the present invention does not limit the dimensions and thicknesses of each component. To make the illustrations clearer, the thickness of some components has been appropriately exaggerated in the drawings.

[0040] The schematic diagram of the online scalable configuration system architecture for dynamic changes in time-sensitive networks in this embodiment is shown below. Figure 1As shown, the system mainly consists of three parts: a network input module, an offline pre-routing module, and an online configuration module. The network input module is responsible for generating network topology, saving old routes and scheduling schemes, and inputting network parameters. It also updates dynamic network changes, including the addition and deletion of terminal devices and the replacement of data flow terminals, thereby updating the changed data flow set. The offline pre-routing module generates a set of candidate routing paths between network nodes based on the network topology from the network input module, serving as the prior routing input for the online configuration module. The online configuration module includes an online routing unit, an online scheduling unit, a schedulability analysis unit, an inter-flow conflict detection unit, and a deployment unit. The online routing unit generates a set of routing paths for new data flow sets based on the set of candidate paths between network nodes and inputs it into the online scheduling unit, which is responsible for generating a scheduling scheme for the new data flow set. The schedulability analysis unit and the inter-flow conflict detection unit are the foundational support for the online routing unit and the online scheduling unit. The deployment unit is responsible for converting the routing and scheduling schemes into configuration commands that terminal devices and network devices can recognize. The flowchart of the online routing unit's operation is shown below. Figure 2 As shown, the operation flowchart of the online scheduling unit is as follows: Figure 3 As shown, the data stream attributes are as follows: period range is {1-10}ms, data frame size range is {64-1500}B, cutoff time is equal to period, network bandwidth B is 100Mb / s, and time slot size T is 0.25ms.

[0041] Step 1: Initialize the offline pre-routing module, set the size of the number of candidate paths K, and calculate the set of K candidate paths between any two network nodes based on the network topology using classic loop-free routing algorithms such as Yen's and Lawler's algorithms. Output the results to the online configuration module.

[0042] Step 2: Based on dynamic network changes, update the information on terminal additions / deletions and data stream set changes. For newly added terminal devices, record the directly linked network nodes and corresponding data stream attributes, and add them to the new data stream set. For deleted terminal devices, add the corresponding data streams to the deleted data stream set. If the data stream sending and receiving terminal devices are changed, add them to the changed data stream set. Then, combine the new and changed data stream sets into a new data stream set. Finally, input the network parameters, including the new data stream set, deleted data stream set, existing data stream set, existing routing and scheduling scheme, network bandwidth, and time slot size, into the online configuration module.

[0043] Step 3: Invoke the schedulability analysis unit to determine the latency condition for the new data stream set. If the condition is met, activate the online routing unit to select the optimal path for the new data stream set; otherwise, adjust the time slot size to meet the latency condition. Deleting a data stream set does not require routing. The latency condition expression is as follows:

[0044]

[0045] in Represents data stream s i The shortest hop count of the candidate path, d i Represents data stream s i The transmission deadline.

[0046] In this example, the optimal path selection is implemented using an incremental algorithm. The routing order is given randomly, and the set of possible paths for each data flow is determined by the prior candidate path set of the network node pairs directly linked to the corresponding terminal device and real-time constraints. The real-time constraints are used to filter out candidate paths with excessively high hop counts. The routing optimization objective is to minimize the flow scheduling link load index based on secondary flow packets, specifically expressed as:

[0047]

[0048] Where M 2,w and ρ represents the second-level group of link w and the corresponding data stream set, respectively. w It equals the greatest common divisor of the integer periods of all data streams in link w. The m2th second-order packet cycle of link w is equal to the minimum processing cycle of the corresponding data stream set, b. i Represents data stream s i The frame size. Note: The integer period of the data stream in link w is equal to the quotient of the data stream period divided by the slot size, and the processing period is equal to the integer period divided by ρ. w The business.

[0049] The secondary grouping is based on data flow cycle analysis. First, the processing cycle set of the corresponding link data flow is obtained. Then, the maximum processing cycle of the link data flow is calculated. Prime numbers within the range of 1 to the maximum processing cycle are identified, and each prime number represents a secondary group. If the prime number corresponding to a secondary group is the maximum divisible prime number of the data flow's processing cycle, this data flow is assigned to that secondary group. Furthermore, data flows with a processing cycle of 1 are assigned to a separate secondary group. For example, if the data flow processing cycle set of link w is... The maximum processing cycle is 10, and the set of prime numbers is {2,3,5,7}. Therefore, the processing cycle set of the second-level groups of link w is {1}, {2,4,8}, {3,6,9}, and {5,10}.

[0050] The real-time constraint is that the product of the data flow routing hop count and the time slot size is less than or equal to the deadline.

[0051] Step 4: Based on the optimal path set of data flows calculated by the online routing unit in Step 3, update the data flow set traversed by each link in the network. Call the schedulability analysis unit to determine the conflict conditions for each link's data flow set. If the conditions are met, return "yes"; otherwise, return "no". The mathematical expression for the conflict conditions is:

[0052]

[0053] Where, p i Represents data stream s i The period, S w The set of data flows that link w traverses includes both new and existing data flows.

[0054] Step 5: If step 4 yields a "yes" result, the online scheduling unit is invoked to generate the scheduling scheme; otherwise, the data stream set is unschedulable. There are two types of scheduling objects: new data streams and old data streams (existing data streams excluding deleted data streams). Old data streams retain their original scheduling schemes. New data streams are scheduled incrementally. The scheduling optimization objective is to minimize the maximum link time slot occupancy. The scheduling order of new data streams is in ascending order of offset range, calculated as follows: Where |R i | Represents a new data stream s i The hop count of the optimal routing path. For each new data flow scheduling step, the corresponding maximum link slot occupancy value is calculated using the inter-flow conflict detection unit based on multi-level flow packets. If the maximum link slot occupancy is less than or equal to the upper bound of network link resources, the data flow and scheduling scheme are synthesized; otherwise, scheduling fails.

[0055] The inter-flow collision detection unit mainly comprises two parts: link multi-level flow grouping and maximum time slot occupancy calculation. Link multi-level flow grouping is primarily based on data flow period and data flow offset analysis within the current link, as illustrated in the diagram below. Figure 4 As shown. The implementation method of two-level grouping has been described in step three, and the data flow s of link w. i The first-level and third-level group labels are calculated as follows:

[0056]

[0057] Where q i,w Represents data stream s i Transmission offset on link w.

[0058] For any two data streams belonging to different first-level groups, they do not occupy the same time slot, i.e., they do not conflict; for any two data streams belonging to different third-level groups, they also do not occupy the same time slot. For any two data streams belonging to different second-level groups, they occupy at least one time slot, i.e., they conflict. Based on this, the maximum time slot occupancy is calculated as follows:

[0059]

[0060] Among them, M 1,w , and These represent the first-level group, second-level group, and third-level group of link w, as well as the corresponding data stream set.

[0061] The mathematical expression for the upper bound of network link resources is: This represents the length of the network switching port queue.

[0062] Step 6: The deployment unit is invoked to convert the routing and scheduling schemes obtained in Steps 4 and 5 into configuration instructions based on XML or JSON format, and then distributed to network devices and terminal devices for updating and taking effect.

[0063] Step 7: Update the network link status and the new routing and scheduling scheme to the network input module.

[0064] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for dynamically changing online scalable configuration of a time-sensitive network, characterized in that, Includes the following steps: Step 1: Initialize the offline pre-routing module, set the number of candidate paths, and input the information into the online configuration module; Step 2: Based on dynamic network changes, update the terminal addition / deletion status and data stream set changes, and input the new data stream set, deleted data stream set, original data stream set, original routing and scheduling scheme, network bandwidth and time slot size network parameters into the online configuration module; Step 3: Call the schedulability analysis unit to determine the latency condition of the new data stream set. If the condition is met, activate the online routing unit to select the optimal path for the new data stream set; otherwise, adjust the time slot size to meet the latency condition. Step 4: Based on the optimal path set of data flow calculated by the online routing unit, update the data flow set traversed by each link in the network, call the schedulability analysis unit to perform conflict judgment on the data flow set of each link, and if the conflict is satisfied, return "yes"; otherwise, return "no". Step 5: If yes, call the online scheduling unit to generate a scheduling scheme; otherwise, the data stream set is not schedulable. Step 6: Invoke the deployment unit to convert the routing and scheduling scheme into configuration instructions, and send them to network devices and terminal devices to update and take effect; Step 7: Update the network link status and the new routing and scheduling scheme to the network input module; An efficient inter-flow conflict detection mechanism based on multi-level flow grouping avoids conflict detection methods based on link maximal cliques and link over-periods by using a correlation analysis of flow period and offset. The link data stream is grouped into multiple levels based on the data stream period and transmission offset. For any two data streams belonging to different first-level groups, there is no conflict as they do not occupy the same time slot. For any two data streams belonging to different third-level groups, there is no same time slot occupation; For any two data streams belonging to different secondary groups, there is at least one time slot occupied, i.e., a conflict.

2. The online scalable configuration method for dynamically changing time-sensitive networks as described in claim 1, characterized in that, The multi-level grouping principle is as follows: The first-level grouping label of the data stream is equal to the remainder of the link offset divided by the greatest common divisor of the periods of all data streams in the link; the second-level grouping of the link is implemented based on the data stream period analysis. First, the processing period set of the corresponding link data stream is obtained, then the maximum processing period of the link data stream is calculated, and prime numbers in the range of 1 to the maximum processing period are found. Each prime number represents a second-level group. If the prime number corresponding to the second-level group is the maximum divisible prime number of the data stream processing period, the data stream is assigned to the second-level group. Data streams with a processing period of 1 are assigned to a separate second-level group; the third-level grouping label of the data stream is equal to the remainder obtained by dividing the link offset by the quotient of the greatest common divisor of the periods of all data streams in the link and then dividing by the processing period of the second-level group in which it belongs.

3. The online scalable configuration method for dynamically changing time-sensitive networks as described in claim 1, characterized in that, An online incremental approach based on decoupling routing and scheduling, as well as offline pre-routing, accelerates the computation speed of routing and scheduling schemes while ensuring scheduling space.

4. The online scalable configuration method for dynamically changing time-sensitive networks as described in claim 3, characterized in that, The routing and scheduling are decoupled by constructing a link load balancing index that reflects flow scheduling to guide the online optimal route. The link load balancing index is based on period analysis and secondary grouping, which aggregates data flows with multiples of each other in period on the same link and distributes data flows with coprime periods on different links.

5. The online scalable configuration method for dynamically changing time-sensitive networks as described in claim 3, characterized in that, The offline pre-routing provides the shortest candidate path set between network nodes for online routing. For new data streams, it is not necessary to re-invoke the loop-free routing algorithm to traverse the network topology nodes. Instead, it is only necessary to check the directly linked network nodes of the terminal devices corresponding to the new data streams and quickly filter out the set of selectable paths based on real-time constraints.

6. The online scalable configuration method for dynamically changing time-sensitive networks as described in claim 1, characterized in that, The schedulability analysis conditions are constructed from the perspectives of transmission delay and flow conflict, including delay judgment conditions and conflict judgment conditions. The analysis is conducted to determine whether the data stream set can be scheduled under the best case conditions to meet the requirements of real-time performance and conflict-free operation.

7. The online scalable configuration method for dynamically changing time-sensitive networks as described in claim 6, characterized in that, The latency judgment condition is whether the minimum network forwarding latency of the new data stream is less than or equal to the corresponding deadline, where the minimum network forwarding latency is equal to the product of the minimum routing hop count and the time slot size. If the data stream set does not meet the latency judgment condition, the time slot size can be adjusted downward. The conflict judgment condition is whether the maximum network link utilization of all data streams is less than or equal to the network bandwidth, where the network link utilization is equal to the sum of the ratios of the frame size to the period of all data streams on the link.

8. An online scalable configuration system for time-sensitive network dynamic changes using the method described in any one of claims 1-7, characterized in that, It includes a network input module, an offline pre-routing module, and an online configuration module. The network input module is responsible for generating network topology, saving old routes and scheduling schemes, and inputting network parameters. It is also responsible for updating dynamic network changes, including adding and deleting terminal devices, changing data flow terminals, and thus updating the changed data flow set. The offline pre-routing module generates a set of candidate routing paths between network nodes based on the network topology of the network input module, which serves as the prior routing input for the online configuration module. The online configuration module includes an online routing unit, an online scheduling unit, a schedulability analysis unit, an inter-flow conflict detection unit, and a deployment unit. The online routing unit generates a set of routing paths for the new data flow set based on the set of candidate paths between network nodes and inputs it into the online scheduling unit. The online scheduling unit is responsible for generating a scheduling scheme for the new data flow set. The schedulability analysis unit and the inter-flow conflict detection unit are the basic support for the online routing unit and the online scheduling unit. The deployment unit is responsible for converting the routing and scheduling schemes into configuration instructions that can be recognized by terminal devices and network devices.

Images