Variable preemption in time sensitive networks using priority regeneration

Abstract

Embodiments of the present disclosure relate to variable pre-emption in time sensitive networks using priority regeneration. A method for operating a time sensitive network, TSN, wherein the TSN comprises at least a first high importance segment and a second low importance segment, such that traffic within the first segment on the one hand and traffic within the second segment on the other hand are transmitted over different sets of physical links in the TSN, the first segment being connected to a first port of a border network element, the border network element connecting the first segment and the second segment, and the second segment being connected to a second port of the border network element, the method comprising the steps of: at the border network element, using TSN priority regeneration per port, remapping priority tags attached to data flows to updated priority tags; at the border network element, splitting the data flows into a "pre-emptable" class and a "pre-empt" class; and forwarding the data flows from the border network element to at least one next hop network element.

Description

Technical Field

[0001] This invention relates to the field of Time-Sensitive Networks (TSNs), which can be used, for example, as a communication medium in a distributed control system (DCS) in an industrial plant. Background Technology

[0002] A distributed control system (DCS) for an industrial plant includes multiple controllers, sensors, and actuators. For example, sensors can deliver measurements from an industrial process being executed on the plant. The controllers can then communicate, for example, with actuators that physically act on the process to maintain the measurements (such as temperature or pressure) at a desired setpoint value.

[0003] Communication within a DCS requires fast and reliable delivery of data streams. Dedicated fieldbus networks are designed to provide the necessary low latency and reliability, but they aim to replace multiple proprietary fieldbus systems with standardized, high-performance networks. For this purpose, Time-Sensitive Networking (TSN) built on traditional Ethernet networks is well-known in the art. WO2020 / 136487A2 discloses a controller for process plants capable of communicating with a mix of TSN and non-TSN devices within a network.

[0004] Depending on the number of participants, configuring the entire TSN can be very complex and time-consuming. Summary of the Invention

[0005] Therefore, one object of the present invention is to facilitate and allow partial automation of TSN configuration.

[0006] This objective is achieved by the method for operating the TSN according to the first independent claim and by the method for configuring the TSN according to the second independent claim. Further advantageous embodiments are described in detail in the corresponding dependent claims.

[0007] This invention provides a method for operating a Time-Sensitive Network (TSN). The TSN comprises at least a first high-importance segment and a second low-importance segment. The terms "high importance" and "low importance" are relative to the specific application at hand. In each application, there will be some data streams whose timely and reliable delivery is more critical than the delivery of other data streams. For example, a measurement captured somewhere in an industrial process and displayed only somewhere in a control room is still important to some extent (otherwise it wouldn't have been measured in the first place). However, the time requirement for updating this measurement is lower compared to updating the measurement as part of a closed feedback loop.

[0008] The segments are independent because traffic within the first segment and traffic within the second segment are transmitted through different sets of physical links within the TSN. The first and second segments are connected via a border network element. The first segment is connected to the first port of the border network element, and the second segment is connected to the second port of the border network element. Through a third port and more, the border network element can connect the first and second segments to the outside world, namely other segments of the TSN or the Internet.

[0009] Data streams received at the first and second ports have priority labels attached to them. For example, these priority labels could be Time-Aware Traffic Shaping (TAS) priorities. According to the TSN standard, there are eight different TAS priorities. During this process, the boundary network elements remap these priority labels to updated priority labels, such that a data stream initially received at the first port does not have the same updated priority label as any data stream initially received at the second port.

[0010] Based on the mapping from updated priority labels to classes, boundary network elements split the data flow into "preemptive" and "preemptible" classes. For example, if there are eight priority levels from 0 to 7, priority levels 7 to 4 can be mapped to the "preemptive" class, and priority levels 3 to 0 can be mapped to the "preemptible" class.

[0011] Border network elements forward data streams to at least one next-hop network element. At least in the event of congestion on the link to the next-hop network element, the forwarding of "preemptive" data streams takes precedence over the forwarding of "preemptible" data streams. That is, frames from a "preemptible" data stream may have to wait in a queue to be transmitted until the transmission of frames belonging to a "preemptive" data stream is complete.

[0012] The first and second network segments can be managed independently. This means that the first data stream in the first network segment can be assigned a specific priority label (e.g., 5), and the second data stream in the second network segment can be assigned the same priority label. After remapping, the two data streams will have different updated priority labels. Therefore, when splitting into "preemptive" and "preemptible" classes, the updated priority labels reflect that the first data stream is more important than the second data stream because it originates from the more important first network segment. Thus, the first data stream can be mapped to the "preemptive" class, and the second data stream can be mapped to the "preemptible" class, in order to better deterministically deliver the first data stream at the expense of the second data stream.

[0013] On the condition that the data stream initially received on the first port does not have the same updated priority label as any data stream initially received on the second port, remapping and subsequent splitting can be performed based on any appropriate set of rules that takes into account the importance of each data stream relative to the application at hand.

[0014] In a simple example, all data streams from the first high importance segment can be assigned priority labels that will cause them to be mapped to the "preemptive" class of updates during the split, while all data streams from the second low importance segment can be assigned priority labels that will cause them to be mapped to the "preemptible" class of updates during the split.

[0015] In another example, remapping can increase the priority label of data streams from the first segment and decrease the priority label of data streams from the second segment. During the split, all data streams with priority labels above a predetermined threshold can be mapped to the "preemptive" class, while data streams with priority labels below that threshold can be mapped to the "preemptible" class. In this way, the original priority labels of the data streams (which reflect their relative priority within the corresponding network segment) still have some influence on the final decision of whether these data streams will be mapped to the "preemptive" or "preemptible" class.

[0016] Therefore, prioritization within each network segment and processing on boundary network elements can work together to improve the determinism of critical traffic transmitted through the TSN. However, this improvement in determinism can also be achieved by using the existing TSN as is and modifying only the processing on boundary network elements. That is, determinism can also be improved by making changes in only a single place. One advantage of making changes in only one place is that existing, tried-and-tested priorities and configurations within each network segment can remain unchanged. In some high-risk applications, such as chemical or nuclear processes, any configuration changes may depend on prior regulatory approval. Obtaining such approval for changes that only affect boundary network elements and thus keep communication within each network segment unchanged is much easier than obtaining approval for a complete redesign of the TSN.

[0017] In an advantageous embodiment, the next-hop network element can restore the original priority label of the data flow. In this way, the effect of remapping can be limited to the link between the boundary network element and the next-hop network element. For example, remapping and splitting on the boundary network element can be specifically used to alleviate congestion on the link between the boundary network element and the next-hop network element, which is a bottleneck within the TSN, but once the bottleneck is cleared, the original priority information can be reused. For example, if the link between the boundary network element and the next-hop network element must carry traffic to and from several network segments and it has a bandwidth smaller than the combined bandwidth of the several network segments, it may become prone to becoming a bottleneck.

[0018] In another advantageous embodiment, the data flow may have already been split into "preemptive" and "preemptible" classes on network elements in the first and second segments. These network elements can then forward the data flow to the corresponding next-hop network element, such that, at least in the event of congestion on the link to the corresponding next-hop network element, the forwarding of "preemptive" data flows takes precedence over the forwarding of "preemptible" data flows. The mapping from priority labels to classes differs between the first and second segments. That is, there is a segment-by-segment preemption configuration. In this case, the combination of remapping and splitting on the boundary network elements according to this method allows the use of the TSN preemption mechanism on the uplink from the boundary network element to the outside world, while preserving the existing preemption configuration in each network segment. However, any degree of coordination and cooperation can also exist between the configuration of each network segment and the configuration of the boundary network element.

[0019] In one example, the mapping between priority labels and classes within the first and / or second network segments, the remapping of priority labels on the boundary network element, and the mapping from updated priority labels to classes on the boundary network element can be coordinated so that data flows in the "preemptible" classes within the first and / or second network segments are not in the "preemptible" classes when forwarded from the boundary network element. In this way, the "waste" of privileged processing of the data flow by the boundary network element is avoided: the data flow is "contaminated" by uncertainty because it is already in the "preemptible" class in its originating network segment before reaching the boundary network element. The transmission of this data flow can no longer achieve high determinism through privileged processing on the boundary network element.

[0020] In another example, the mapping between priority labels and classes within the first and / or second network segments, the remapping of priority labels on boundary network elements, and the mapping from updated priority labels to classes on boundary network elements are coordinated such that at least one system-critical data flow is in a "preemptive" class within the first and / or second network segments and during forwarding from the boundary network element. In this way, the particular system-critical data flow can be transmitted across the entire TSN in a highly deterministic manner. Privileged processing of the data flow within its originating network segment is not "wasted" due to non-deterministic processing on the boundary network element.

[0021] As previously mentioned, TSNs can be selected to include controllers, sensors, and actuators from a distributed control system (DCS) as participants in industrial plants. In this way, TSNs can replace previous proprietary fieldbus networks without sacrificing determinism.

[0022] In the context of DCS, at least one data stream that is part of a closed feedback loop in an industrial process can be selected as a system-critical data stream. The transmission of this data stream is time-critical because inappropriate delays can cause the process to escalate out of control. For example, if a pressure measurement in a vessel is delayed in the network and does not reach the controller in time to react, or if a command from the controller to open a pressure relief valve is lost in the network and the valve fails to open, the pressure in the vessel may exceed the vessel's physical limits within a very short time.

[0023] In another advantageous embodiment, memory for queuing frames of the received data stream is allocated on the boundary network element such that the queue for frames of a "preemptible" data stream can hold more frames than the queue for frames of a "preemptive" data stream. The boundary network typically stores received data stream frames in queues and forwards them according to a "first-in, first-out" (FIFO) principle. For example, there may be one queue for each possible priority label. If congestion exists and a "preemptible" data stream must wait for another "preemptive" data stream, the queue for that priority label can be filled. Once the queue is full, new arriving frames for that queue can be dropped, or these new arriving frames can cause the oldest frame at the front of the queue to be dropped. The larger the queue, the less likely data will be lost in this way. If the amount of queue space available for its operation is limited, it is advantageous to concentrate that queue space on data streams with a higher waiting tendency. In contrast, high-priority "preemptive" data streams do not require large queues because frames will not wait there for long periods. In addition, the size of the queue can be made dependent on the update priority label of the data stream, which is another indicator of the probability that a frame of a particular data stream will have to wait to be transmitted.

[0024] The present invention also provides a method for configuring a Time-Sensitive Network (TSN). The TSN includes multiple network elements interconnected by links. The network elements are configured to forward data flows to the appropriate next-hop network element.

[0025] The method begins by identifying at least one segment of the TSN as a congestion-prone segment, which provides connectivity to or from at least a first and a second additional segment of the TSN. Specifically, if the available bandwidth in this segment is less than the combined bandwidth of the first and second additional segments, simultaneous high activity in both additional segments can overload the congestion-prone segment.

[0026] Next, network elements connected to the first additional segment via the first port, to the second additional segment via the second port, and to the congestion-prone segment via the third port are identified as boundary network elements. These boundary network elements are then configured as follows:

[0027] • Using the TSN priority regeneration of each port, the priority labels of the data streams received on the first and second ports are remapped to updated priority labels, such that the data streams initially received on the first port do not have the same updated priority labels as any data streams initially received on the second port.

[0028] • Based on the mapping from updated priority labels to classes, the data flow is split into "preemptive" and "preemptible" classes; and

[0029] • The data streams received on the first and second ports are forwarded to the third port, where the forwarding of the "preemptive" data stream takes precedence over the forwarding of the "preemptible" data stream.

[0030] In this way, as described above, the TSN network is improved because data flows from more important network segments can take precedence over other data flows on congested segments, even if they are marked as "important" in the context of the corresponding first and second additional network segments.

[0031] Configuration can be performed fully automatically. Information about the network geometry is typically available electronically and can be resolved by a machine. Boundary network elements can be configured by software. No manual engineering of the entire TSN is required, and any engineering work already done on individual network segments is preserved.

[0032] In another advantageous embodiment, at least one key performance indicator of the TSN is monitored and / or simulated. Then, with the goal of improving the key performance indicator, the remapping of priority labels and / or the splitting of data flow into "preemptive" and "preemptible" classes are optimized. For example, multiple candidate configurations for remapping and / or splitting can be set, and for each such candidate configuration, a key performance indicator can be computed. The candidate configuration with the optimal key performance indicator value can then be implemented on the boundary network elements.

[0033] Specifically, key performance indicators may include one or more of the following:

[0034] • Data throughput of congested network segments;

[0035] • Delay in the delivery of at least one data stream;

[0036] • Frame loss rate of at least one data stream; and

[0037] • A measure of the determinism of the delivery of at least one data stream.

[0038] Similarly, optimization of one or more key performance indicators can be performed fully automatically. Therefore, fine-grained prioritization of the data stream into eight distinct priority tags based on the current TAS characteristics is no longer tied to the requirement for more manual configuration of the entire TSN. There is no longer a choice between eight priorities and extensive manual configuration, or only two preemptive classes and less manual configuration.

[0039] The method can be implemented entirely or partially by a computer. Therefore, the present invention also provides one or more computer programs having machine-readable instructions that, when executed on one or more computers, cause the one or more computers to perform one of the methods described above. In particular, the virtualization platform and one or more hardware controllers can be considered as computers.

[0040] The present invention also provides one or more non-transitory storage media having one or more computer programs and / or downloadable products. Downloadable products are those that can be sold in online stores for immediate use via download. The present invention also provides one or more computers having one or more computer programs and / or having one or more non-transitory machine-readable storage media and / or downloadable products. Attached Figure Description

[0041] The invention is illustrated below with reference to the accompanying drawings, but is not intended to limit the scope of the invention.

[0042] The attached image shows:

[0043] Figure 1An exemplary embodiment of the method 100 for operating TSN 1;

[0044] Figure 2 : An exemplary network geometry with congestion-prone segment 1c;

[0045] Figure 3 An exemplary embodiment of the method 200 for configuring TSN 1. Detailed Implementation

[0046] Figure 1 This is a schematic flowchart of an embodiment of method 100 for operating TSN 1. (As will be...) Figure 2 As shown in more detail, TSN 1 includes network elements 3a-3i and 4 interconnected via physical links 2a-2i. TSN 1 includes a first high-importance segment 1a with network elements 3a-3b and a second low-importance segment 1b with network elements 3c-3e.

[0047] In step 110, on network elements 3a-3e in segments 1a and 1b, based on the mapping from priority labels 6a-6g to classes 7a and 7b (this mapping is configured separately for each segment 1a and 1b), data streams 5a-5g carrying priority labels 6a-6g are split into "preemptive" class 7a and "preemptible" class 7b.

[0048] In step 120, data flows 5a-5g are forwarded to the corresponding next-hop network elements 3a-3e and 4, wherein the forwarding of data flows 5a-5g in the "preemptive" class 7a takes precedence over the forwarding of data flows 5a-5g in the "preemptible" class 7b.

[0049] In step 130, the boundary network element 4 uses TSN priority regeneration for each port to remap the priority labels 6a-6g of the data streams 5a-5g received on the first port 4a and the second port 4b to updated priority labels 6a*-6g*. This remapping is performed such that the data streams 5a-5c initially received on the first port 4a do not have the same updated priority labels 6a*-6g* as any data streams 5d-5g initially received on the second port 4b.

[0050] In step 140, on the border network element 4, regarding forwarding to the next-hop network element 8, data flows 5a-5g are split into "preemptive" class 7a and "preemptible" class 7b. This split is based on the mapping from the updated priority labels 6a*-6g* to classes 7a and 7b configured on the border network element 4.

[0051] According to box 141, on the boundary network element 4, memory can be allocated for queuing frames of the received data streams 5a-5g, such that the queue for frames of the "preemptible" data streams 5a-5g can hold more frames than the queue for frames of the "preemptive" data streams 5a-5g. According to box 141a, the size of the queue can also be made dependent on the updated priority tags 6a*-6g* of the data streams 5a-5g.

[0052] In step 150, data flows 5a-5g are forwarded from the border network element 4 to at least one next-hop network element 8. At least in the event of congestion on the link 2f to the next-hop network element 8, the forwarding of data flows 5a-5g in the "preemptive" class 7a takes precedence over the forwarding of data flows 5a-5g in the "preemptible" class 7b.

[0053] In step 160, the next-hop network element 8 may optionally restore the original priority labels 6a-6g.

[0054] Figure 2 An exemplary network geometry of TSN 1 is shown. TSN 1 includes network elements 3a-3i and 4 interconnected via physical links 2a-2i. Network elements 3a and 3b with links 2a and 2b form a first high-importance segment 1a. Network elements 3c, 3d, and 3e with links 2c, 2d, and 2e form a second low-importance segment 1b. The first segment 1a generates data streams 5a-5c with priority labels 6a-6c. The second segment 1b generates data streams 5d-5g with priority labels 6d-6g.

[0055] Both segments 1a and 1b are connected to border network element 4. Segment 1a is connected to the first port 4a of network element 4, and segment 1b is connected to the second port 4b of network element 4. Border network element 4 remaps priority labels 6a-6g to updated priority labels 6a*-6g*, and forwards all data flows 5a-5g to the next-hop network elements 3g and 8, which are connected via the third port 4c. Because all data flows 5a-5g must travel across a single physical link 2f between border network element 4 and the next-hop network element 8, segment 1c, which includes border network element 4, next-hop network element 8, and physical link 2f, is a congestion-prone segment.

[0056] When the next-hop network elements 3g and 8 forward the data flow to other network elements 3f and 3i, it does so via multiple links 2g and 2i, so that one of these links 2g and 2i is less prone to congestion.

[0057] Figure 3 It is used to configure TSN 1 (e.g.) Figure 2The schematic flowchart of method 200 of TSN 1) shown is shown.

[0058] In step 210, at least one segment 1c of TSN 1 is identified, which provides connectivity to or from at least a first additional segment 1a and a second additional segment 1b of TSN 1. That is, the constellation of the first additional segment 1a, the second additional segment 1b, and segment 1c, which is required by both segments 1a and 1b, is identified. As previously mentioned, segment 1c is a congestion-prone segment because high traffic volumes simultaneously entering and exiting both segments 1a and 1b can overload it.

[0059] In step 220, network element 4 is identified as boundary network element 4, which is connected to a first additional segment 1a via a first port 4a, to a second additional segment 1b via a second port 4b, and to a congested segment 1c via a third port 4c.

[0060] In step 230, the boundary network element 4 is configured as follows:

[0061] • Using the TSN priority regeneration of each port, the priority labels 6a-6g of the data streams 5a-5g received on the first port 4a and the second port 4b are remapped to the updated priority labels 6a*-6g*, so that the data streams 5a-5g initially received on the first port 4a do not have the same updated priority labels 6a*-6g* as any data streams 5a-5g initially received on the second port 4b.

[0062] Based on the mapping from the updated priority labels 6a*-6g* to classes 7a and 7b, the data flow 5a-5g is split into "preemptive" class 7a and "preemptible" class 7b; and

[0063] • Data streams received on port 4a and port 4b are forwarded to port 4c. Forwarding of data streams 5a-5g in "preemptive" class 7a takes precedence over forwarding of data streams 5a-5g in "preemptible" class 7b.

[0064] In step 240, at least one key performance indicator 1* of TSN 1 is monitored and / or simulated.

[0065] In step 250, with the goal of improving the critical performance indicator 1*, the remapping of priority labels 6a-6g and / or the splitting of data streams 5a-5g into "preemptive" class 7a and "preemptible" class 7b are optimized.

[0066] List of reference numerals

[0067] 1. Time-Sensitive Networking (TSN)

[0068] The first highly important paragraph of 1a TSN 1

[0069] The second least important segment of TSN 1b

[0070] 1c TSN 1 congestion-prone segment

[0071] 1* TSN 1 Key Performance Indicators

[0072] Physical link of 2a-2i TSN 1

[0073] Network elements in 3a-3i TSN 1

[0074] 4 Boundary Network Elements in TSN 1

[0075] 4a-4c Ports of Boundary Network Element 4

[0076] 5a-5g data stream

[0077] Priority tags for 6a-6g data streams and 5a-5g data streams

[0078] 6a*-6g* Updated priority tags generated by boundary network element 4

[0079] 7a Data Stream 5a-5g "Preemption" Class

[0080] 7b Data Stream 5a-5g "Preemptible" Class

[0081] 8 Next-hop network elements

[0082] 100 Methods for operating TSN 1

[0083] 110. Split the data stream 5a-5g into segments 1a and 1b, specifically classes 7a and 7b.

[0084] 120 Forward data streams 5a-5g within segments 1a and 1b.

[0085] 130 Remap priority labels 6a-6g on boundary network element 4

[0086] 140. Based on the new tags 6a*-6g*, the data stream 5a-5g is split into classes 7a and 7a.

[0087] 141. Allocate queue memory based on classes 7a and 7b.

[0088] 141a Based on the new tags 6a*-6g*, allocate queue memory

[0089] 150 forwards data stream 5a-5g to the next-hop network element 8

[0090] 160 Restore original priority tags 6a-6g

[0091] 200 How to configure TSN 1

[0092] 210 identifies the congested segment 1c serving segments 1a and 1b.

[0093] 220 Identifying Boundary Network Element 4

[0094] 230 Configure boundary network element 4

[0095] 240 Monitor and / or simulate key performance indicators 1*

[0096] 250 Optimizations were performed with the goal of improving the key performance indicator 1*.

Claims

1. A method (100) for operating a Time-Sensitive Network (TSN) (1), wherein the TSN comprises at least a first high-importance segment (1a) and a second low-importance segment (1b) managed independently, wherein traffic in the first segment (1a) and traffic in the second segment (1b) are transmitted through different physical link sets (2a-2b; 2c-2e) in the TSN (1), the first segment (1a) is connected to a first port (4a) of a border network element (4), the border network element (4) connecting the first segment (1a) and the second segment (1b), and the second segment (1b) is connected to a second port (4b) of the border network element (4), the method (100) comprising the following steps: • On the boundary network element (4), using the TSN priority regeneration of each port, the priority labels (6a-6g) of the data streams (5a-5g) received on the first port (4a) and the second port (4b) are remapped (130) to updated priority labels (6a*-6g*), thereby causing the data streams (5a-5c) initially received on the first port (4a) to not have the same updated priority labels (6a*-6g*) as any data streams (5d-5g) initially received on the second port (4b); • On the boundary network element (4), based on the mapping from the updated priority labels (6a*-6g*) to classes (7a, 7b), the data stream (5a-5g) is split (140) into a "preemptive" class (7a) and a "preemptible" class (7b); and • The data stream (5a-5g) is forwarded (150) from the boundary network element (4) to at least one next-hop network element (8), wherein, at least in the event of congestion on the link (2f) to the next-hop network element (8), the forwarding of the "preemptive" data stream (5a-5g) takes precedence over the forwarding of the "preemptible" data stream (5a-5g).

2. The method (100) according to claim 1 further includes: The original priority labels (6a-6g) of the data stream (5a-5g) are recovered (160) by the next-hop network element (8).

3. The method (100) according to claim 1 or 2, further comprising: • On the network elements (3a-3e) in the first segment (1a) and the second segment (1b), based on the mapping from priority labels (6a-6g) attached to the data streams (5a-5g) to classes (7a, 7b), these data streams (5a-5g) are split (110) into "preemptive" classes (7a) and "preemptible" classes (7b); and • The data streams (5a-5g) are forwarded (120) to the corresponding next-hop network elements (3a-3e, 4), wherein, at least in the event of congestion on the link (2a-2e) to the corresponding next-hop network element (3a-3e, 4), the forwarding of "preemptive" data streams takes precedence over the forwarding of "preemptible" data streams. The mapping from priority labels (5a-5g) to classes (7a, 7b) differs between the first segment (1a) and the second segment (1b).

4. The method (100) according to claim 3, wherein the mapping between priority labels (5a-5g) and classes (7a, 7b) in the first segment (1a) and / or the second segment (1b), the remapping of priority labels (6a-6g) on ​​the boundary network element (4), and the mapping from updated priority labels (6a*-6g*) to classes (7a, 7b) on the boundary network element (4) are coordinated so that, when forwarded from the boundary network element (4), the data streams (5a-5g) in the "preemptible" classes (7b) in the first segment (1a) and / or the second segment (1b) are not in the "preemptible" classes (7a).

5. The method (100) according to claim 3, wherein the mapping between priority labels (5a-5g) and classes (7a, 7b) in the first segment (1a) and / or the second segment (1b), the remapping of priority labels (6a-6g) on ​​the boundary network element (4), and the mapping from updated priority labels (6a*-6g*) to classes (7a, 7b) on the boundary network element (4) are coordinated, with the aim of ensuring that at least one system-critical data flow (5a-5g) is in the "preemptive" class (7a) within the first segment (1a) and / or the second segment (1b) and when forwarded from the boundary network element (4).

6. The method (100) according to claim 1 or 2, wherein the TSN (1) is selected to include controllers, sensors and actuators of a distributed control system (DCS) as participants in an industrial plant.

7. The method (100) according to claim 6, wherein at least one data stream (5a-5g) as part of a closed feedback loop of an industrial process performed on the industrial plant is selected as a system-critical data stream.

8. The method (100) according to claim 1 or 2, further comprising: On the boundary network element (4), a memory is allocated (141) for queuing frames of the received data stream (5a-5g) to allow the queue for frames of the "preemptible" data stream (5a-5g) to hold more frames than the queue for frames of the "preemptive" data stream (5a-5g).

9. The method (100) according to claim 8, further comprising: The size of the queue (141a) also depends on the updated priority label (6a*-6g*) of the data stream (5a-5g).

10. A method (200) for configuring a Time-Sensitive Network (TSN) (1), wherein the TSN (1) comprises a plurality of network elements (3a-3i, 4) interconnected by links (2a-2i), and the network elements (3a-3i, 4) are configured to forward data streams (5a-5g) to the corresponding next-hop network elements (3a-3i, 4), the method (200) comprising the following steps: • Identify (210) at least one segment (1c) of the TSN (1) as a congestion-prone segment, the at least one segment (1c) providing connectivity to or from at least a first additional segment (1a) and a second additional segment (1b) of the TSN (1), wherein the first additional segment (1a) and the second additional segment (1b) are managed independently; • Identify (220) network element (4) as a boundary network element (4), said network element (4) being connected to the first additional segment (1a) via a first port (4a), to the second additional segment (1b) via a second port (1b), and to the congested segment (1c) via a third port (4c); and • The boundary network element (4) is configured (230) as follows: • Using the TSN priority regeneration for each port, the priority tags (6a-6g) appended to the data streams (5a-5g) received on the first port (4a) and the second port (4b) are remapped to the updated priority tags (6a*-6g*). This results in the data stream (5a-5g) initially received on the first port (4a). It does not have the same updated priority label (6a*-6g*) as any data stream (5a-5g) initially received on the second port (4b); Based on the mapping from the updated priority tags (6a*-6g*) to classes (7a, 7b), the data stream (5a-5g) is split into a "preemptive" class (7a) and a "preemptible" class (7b); and The data streams received on the first port (4a) and the second port (4b) are forwarded to the third port (4c), wherein the forwarding of the "preemptive" data streams (5a-5g) takes precedence over the forwarding of the "preemptible" data streams (5a-5g).

11. The method (200) according to claim 10, further comprising: • Monitor and / or simulate (240) at least one key performance indicator (1*) of the TSN (1); as well as • With the goal of improving the key performance indicator (1*), optimize (250) the remapping of the priority labels (6a-6g), and / or the splitting of the data stream (5a-5g) to the "preemptive" class (7a) and the "preemptible" class (7b).

12. The method (200) of claim 11, wherein the key performance indicator (1*) comprises one or more of the following: • Data throughput in the congested segment (1c); • Delay in the delivery of at least one data stream (5a-5g); • Frame loss rate of at least one data stream (5a-5g); and • A measure of the determinism of the delivery of at least one data stream (5a-5g).

13. A computer program product comprising machine-readable instructions that, when executed by one or more computers, cause the one or more computers to perform the method (100, 200) according to any one of claims 1 to 12.

14. A non-transitory storage medium comprising the computer program product according to claim 13.

15. One or more computers having a computer program product according to claim 13 and / or having a non-transitory storage medium according to claim 14.

Images