Method for configuring the TAS protocol of a TSN network and estimating the worst-case end-to-end latencies of transmitted data streams and associated devices

Abstract

Method for configuring the TAS protocol of a TSN network and estimating the worst-case end-to-end latencies of transmitted data streams and associated devices. The present invention relates to a method for estimating the worst-case end-to-end transmission latency of frames in a TSN communication network operating according to a TAS protocol, the method comprising: - for each data stream, the calculation of the worst-case end-to-end transmission latency value that a frame of the data stream may experience, the calculation step comprising, for each frame of the data stream and each output port of a node or switch through which a data stream may pass, the operations for estimating the worst-case disturbance for the transmission of the frame on the considered output port from a plurality of contributions.and - a step estimating the worst-case end-to-end transmission latency of a TSN network from the contributions for each data stream. Figure for the abbreviation: Figure 1,

Description

Title of the invention: Method for configuring the TAS protocol of a TSN network and estimating the worst-case end-to-end latencies of transmitted data streams and associated devices

[0001] The present invention relates to a method for estimating the worst case end-to-end transmission latency of frames in a TSN communication network operating according to the TAS protocol.

[0002] The present invention also relates to a method for configuring a TSN communication network and associated devices, namely a TSN communication network, a computer program product and a readable information medium.

[0003] Critical real-time systems, such as an aircraft or a satellite, rely on several communication networks in order to meet the heterogeneous needs, in terms of real-time criticality level, of the transmission of data streams.

[0004] Thus, one or more communication networks are dedicated to the exclusive transmission of critical data streams, while other communication networks are dedicated to the transmission - often large - of less critical, or even non-critical, data streams.

[0005] However, the proliferation of communication networks in critical real-time systems has significant impacts on the weight, size, and energy consumption of these systems. This problem is amplified by the number of communicating functions and the ever-increasing volume of data exchanged within these systems.

[0006] To address this problem, a mixed-criticality real-time communication network was developed; that is, a communication network enabling the transmission of data streams with varying levels of real-time criticality over the same physical links. This led to the development of the TSN mixed-criticality communication network.

[0007] The abbreviation TSN refers to the English name for "Timing Sensitive Network", which literally translates to time-sensitive network.

[0008] Indeed, in addition to offering high bandwidth (up to 400 GB / s), the TSN communication network implements a number of protocols to control end-to-end latencies and data rates in the network.

[0009] Among these protocols, the TAS protocol is intended to protect the transmission of critical data streams from disturbances that it may suffer in the network.

[0010] The TAS protocol is the IEEE 802.1Qbv protocol. The acronym TAS refers to the corresponding English name of "Time Aware Shaper", which can be literally translated as implementation with a consideration of time.

[0011] The TAS protocol ensures the proper transmission of critical data through mechanisms based on time windows, each exclusive to the transmission of certain streams in particular.

[0012] Thanks to the mechanisms of this protocol, it is therefore now possible to transmit in the same network, via the same physical links, both critical and less critical flows.

[0013] However, proper exploitation of the capabilities of the TAS protocol requires correctly configuring its mechanisms and carefully checking and validating the impact of this configuration on real-time constraints in the network (mainly the worst-case end-to-end latency of the transmission of critical data streams).

[0014] Given the size and complexity of the TSN communication network architecture (distributed network with nodes at the ends communicating via switches), the configuration of the TAS protocol, based on the synchronization of all network components, becomes a rather complex task.

[0015] To achieve such a configuration, it is known to use solvers to find a valid configuration of the TAS mechanisms, but the implementation of these solvers is complex in practice.

[0016] There is a need for a method to assist in the configuration of a TAS protocol of a TSN communication network.

[0017] To this end, the description relates to a method for estimating the worst-case end-to-end transmission latency of frames in a TSN communication network operating according to a TAS protocol,

[0018] the network comprising nodes, each node being connected to a switch, the switches being interconnected to ensure an exchange of data streams between the nodes, each data stream comprising a set of frames, the transmitted frames being critical or less critical and having a transmission priority, a frame with a higher transmission priority being transmitted before a frame with a lower transmission priority, the exchanges of data streams taking place during time windows, some time windows being critical windows exclusively dedicated to the transmission of critical streams and other time windows being less critical windows dedicated to the transmission of less critical streams,

[0019] the method being implemented by computer and comprising:

[0020] - for each data stream, a step of calculating the latency value of worst-case end-to-end transmission that a frame in the data stream can undergo,

[0021] the calculation step comprising, for each frame of the data stream and each output port of a node or switch through which a data stream can pass, the following operations:

[0022] - worst case disturbance estimation for frame transmission on the port output considered from a plurality of contributions, the plurality of contributions including:

[0023] - a first contribution related to other less critical frameworks,

[0024] - a second contribution related to the presence of guard intervals preventing the starting the transmission of a less critical stream within a predefined time interval just before a transmission window for critical streams,

[0025] - a third contribution related to the presence of critical windows, and

[0026] - calculation of the sum of the contributions, the sum of the contributions being the value of worst-case transmission latency of the frame,

[0027] the calculation step comprising, in addition:

[0028] - a worst-case transmission latency value selection operation the highest frame for all frames in the data stream for each output port, and

[0029] - an operation of summing the different values ​​selected for each output port, to obtain a summation result, the summation result value being the estimated value of the end-to-end worst-case transmission latency of the data stream, and

[0030] - a step for estimating the worst-case end-to-end transmission latency of a TSN communication network based on estimated values ​​for each data stream.

[0031] According to other advantageous aspects of the invention, the estimation method comprises one or more of the following features, taken individually or in all technically possible combinations:

[0032] - the operation of estimating the first contribution comprises:

[0033] - the estimation of a first sub-contribution related to less critical frames priorities,

[0034] - the estimation of a second sub-contribution related to less critical frames having the same priority and belonging to a different data stream,

[0035] - the estimation of a third sub-contribution related to less-critical frameworks of lower priority,

[0036] - the estimation of a fourth sub-contribution related to other frames of the stream of data preceding the frame under consideration, and

[0037]

[0038]

[0039]

[0040]

[0041]

[0042]

[0043]

[0044] - the addition of the first sub-contribution, the second sub-contribution, the third sub-contribution and the fourth sub-contribution, to obtain the first contribution. - The estimation of the first sub-contribution is implemented by calculating the equation: Ihp( Tj) = ^VLjehp(VL^j Or: hp(VL,) denotes the set of lower-priority, less-critical frames, • Ihp{ Lj) denotes the value of the first sub-contribution, • Cj denotes the transmission time of the largest frame in the data stream j, • Lât[ T,) denotes the longest time interval between the activation of the frame T, and the moment when the transmission of frame Tj starts, • Jj denotes the maximum activation jitter of a frame in a stream j, Activation is the moment when the frame enters a transmission queue, and • Pj designates the transmission period of frames of a stream j. - The estimation of the second sub-contribution is implemented by calculating the equation: (Lat(Th+J. \ Isp (T, ) = 2 VLj Cj pj I-1 j Or: sp(VL,) denotes the set of less critical frames having the same priority and belonging to another data stream, IspiT,) denotes the value of the second sub-contribution, • Cj denotes the transmission time of the largest frame in the data stream j, • Lât (Tj) denotes the longest time interval between the activation of the Tj frame and the moment when the transmission of the Tj frame starts, • Jj denotes the maximum activation jitter of a frame in a stream j, Activation is the moment when the frame enters a transmission queue in the output port, and • Pj designates the transmission period of frames of a stream j. - the third sub-contribution is calculated as equal to the maximum transmission time of a frame belonging to a less critical flow of lower priority than the flow for which the worst case end-to-end transmission latency is estimated.

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

[0051]

[0052]

[0053] - the estimate of the fourth sub-contribution is calculated as equal to the maximum transmission time of previous lagging frames of the same stream for which the worst case end-to-end transmission latency is estimated. - Critical and less critical windows alternate according to a cycle of a less critical window and a critical window, the estimation of the second contribution related to the presence of guard intervals being implemented by calculating the equation: / Lat(T^C i gb= l gb\~l^~ Or: * / GB denotes the value of the second contribution, • ^GB denotes the duration of a guard interval, • Cj denotes the longest maximum transmission time of a frame in the data stream j to which the frame belongs, Lat^) denotes the longest time interval between the activation of frame T and the moment when the transmission of frame T j starts, and • LCyC}e denotes the duration of a cycle. - Critical windows and less critical windows alternate according to a cycle of a less critical window and a critical window, and in which the estimation of the third contribution related to the presence of critical windows is implemented by calculating the equation: / Lat^T^C GC = ^GC \ Lcycle Or: * IGC refers to the value of the third contribution, • ^GC denotes the duration of a critical window, • Cj denotes the maximum transmission time of a frame in the data stream j to which the frame appears, • Tj) denotes the longest time interval between the activation of the Tj frame and the moment when the transmission of the Tj frame starts, and • LCyCie denotes the duration of a cycle. The description also relates to a method for configuring a TSN communication network operating according to a TAS protocol, The network comprises nodes, each node being connected to a switch, the switches being interconnected to ensure the exchange of data streams between the nodes, each data stream comprising a set of frames, the transmitted frames being either critical or less critical and having a transmission priority, Z + l) A frame with a higher transmission priority must be transmitted before a frame with a lower transmission priority; data exchanges occur during time windows, some time windows being critical windows exclusively dedicated to the transmission of critical data streams, and other time windows being less critical windows dedicated to the transmission of less critical data streams.

[0054] the configuration process being implemented by computer and comprising the steps of:

[0055] - obtaining an initial network configuration, the configuration comprising the network topology and the data flows intended to circulate on the network,

[0056] - application of a worst-case end-to-end transmission latency estimation method end-to-end on the network corresponding to the initial configuration to obtain the worst-case end-to-end network latency,

[0057] - calculation of a configuration validation criterion, the criterion being applied to the worst-case end-to-end latency value of the estimated network,

[0058] the final network configuration being the initial configuration if the validation criterion is met,

[0059] the process further comprising, if the validation criterion is not met, the steps of:

[0060] - modification of the network configuration to obtain a modified configuration,

[0061] - implementation of a worst-case transmission latency estimation method end-to-end on the network corresponding to the modified configuration to determine the worst-case end-to-end latency of the network,

[0062] - test of the validation criterion on the worst-case end-to-end network latency,

[0063] the modification, implementation and testing steps being repeated until to obtain a final modified network configuration of the network verifying the criterion, the final network configuration being the final modified configuration.

[0064] According to other advantageous aspects of the invention, the configuration method comprises one or more of the following features, taken individually or in all technically possible combinations:

[0065] - the process includes, at each iteration, a step of verifying a condition additional information relating to the size of the critical window and the size of the guard interval.

[0066] - the process includes, at each iteration, a step of verifying a condition additional information relating to the clock cycle period and the period of each data stream.

[0067] - the process includes, at each iteration, a step of verifying a condition additional information relating to the phase shift of each critical data stream and the associated critical window.

[0068] The description also relates to a TSN communication network operating according to a TAS protocol, the network comprising nodes, each node being connected to a switch, the switches being connected to each other to ensure an exchange of data streams between the nodes, each data stream comprising a set of frames, the transmitted frames being critical or less critical and having a transmission priority, a frame having a higher transmission priority being to be transmitted before a frame having a lower transmission priority, the exchanges of data streams taking place during time windows, some time windows being critical windows exclusively dedicated to the transmission of critical streams and other time windows being less critical windows dedicated to the transmission of less critical streams, the network being configured by implementing a configuration process as previously described.

[0069] The description also relates to a computer program product comprising a readable information carrier, on which is stored a computer program comprising program instructions, the computer program being loadable onto a data processing unit and adapted to cause the implementation of at least one step of a process as previously described when the computer program is implemented on the data processing unit.

[0070] The description also relates to a readable information carrier comprising program instructions forming a computer program, the computer program being loadable onto a data processing unit and adapted to cause the implementation of at least one step of a process as previously described when the computer program is implemented on the data processing unit.

[0071] In this description, the expression "specific to" means interchangeably "suitable for", "adapted to" or "configured for".

[0072] The invention will become clearer upon reading the following description, given solely by way of non-limiting example, and made with reference to the drawings in which:

[0073] - [Fig. 1] [Fig. 1] is a schematic representation of an example of a network of TSN communication having to operate with three critical data streams,

[0074] - [Fig.2] [Fig.2] is a diagram illustrating the time window mechanisms and TAS protocol guard interval,

[0075] - [Fig.3] [Fig.3] is a diagram illustrating the case where the activation of frames of a stream VL1 data is synchronized with the start of the critical time window, and

[0076] - [Fig.4] [Fig.4] is a diagram illustrating the case where the activation of frames of a stream VL1 data is not synchronized with the start of the critical time window.

[0077] An example of a TSN communication network is shown in [Fig.1].

[0078] A TSN communication network comprises a set of nodes and a set of switches.

[0079] According to the example in [Fig.1], the TSN communication network comprises five distinct nodes, respectively labeled ESI, ES2, ES3, ES4 and ES5.

[0080] For example, in the field of avionics, each node is a computer.

[0081] The role of switches is to route data flows from a sending node to a receiving node.

[0082] Thus, each node is connected to a switch.

[0083] In this case, the TSN communication network comprises two switches, namely a first switch SI and a second switch S2.

[0084] The first switch SI is connected to nodes ESI, ES2, ES4 and ES5 while the second switch S2 is connected to node ES3.

[0085] The SI and S2 switches are, moreover, connected to each other.

[0086] Data flows circulate between these different elements along a predefined path.

[0087] The term “data stream” refers to a set of frames.

[0088] A frame extends temporally between an activation time and an end time of transmission.

[0089] In the case described, the transmitted frames have a transmission priority, with a frame having a higher transmission priority being transmitted before a frame having a lower transmission priority.

[0090] In the example described, three critical data streams circulate in the TSN communication network, namely: • a first critical data stream flowing between the first node ES1 and the fourth node ES4 or the fifth node E5 through the first SI switch, • a second critical data stream flowing between the second node ES2 and the fourth node ES4 or the fifth node E5 through the first SI switch, and • a third critical data stream flowing between the third node ES3 and the fourth node ES4 through the first switch S1 and the second switch S2.

[0091] The direction of flow of critical data allows us to define input and output ports in this example.

[0092] The first SI switch here has three input ports and two output ports whose connections are as follows: the first input port is connected to the first node ESI (more precisely to its output port), the second input port is connected to the second node ES2 (more precisely to its output port), the third input port is connected to the second switch S2, the first output port is connected to the fourth node ES4 (more precisely to its input port) and the second output port is connected to the fifth node ES5 (more precisely to its input port).

[0093] The second switch S2 has an input port connected to the third node ES3 (more precisely to its output port) and an output port connected to the third input port of the first switch SI.

[0094] The TSN communication network of [Fig. 1] is designed to operate according to a TAS protocol

[0095] As previously stated, the TAS protocol of the TSN communication network provides a solution to the problem of blocking the transmission of a critical frame by the transmission of a less critical frame through the use of time windows, each exclusive to the transmission of certain flows in particular.

[0096] A frame is less critical than another when it is possible for it to be transmitted in a less critical window. Such a possibility depends on the application considered.

[0097] This less criticality characteristic can go as far as an absence of criticality, the less critical flow then being a non-critical flow.

[0098] In the rest of the description, the time window dedicated exclusively to the transmission of critical flows will be called the "critical window".

[0099] Similarly, the window dedicated to the exclusive transmission of less critical streams will be called the “less critical window”.

[0100] Critical data flows will be referred to as "VL critical flows". Less critical data flows will be designated as "less critical flows".

[0101] The term VL here refers to the English terminology of “Virtual link,” which literally means virtual link.

[0102] The TAS protocol allows for the configuration of critical and less critical windows. This configuration aims to prevent interference between critical VL flows and less critical flows.

[0103] The TAS protocol also defines the guard interval mechanism. This mechanism is often referred to as the "guard interval" mechanism, in reference to the corresponding English term.

[0104] The purpose of the guard interval is to prevent the start of transmission of a less critical frame in a less critical window when that frame risks "encroaching" on the next critical window reserved for the transmission of critical VL streams.

[0105] These mechanisms are schematically represented in [Fig.2].

[0106] It is these mechanisms of the TAS protocol (time windows and guard interval) which make it possible to "protect" the transmission of critical flows.

[0107] However, this protection can only be effective when these mechanisms are configured correctly.

[0108] Indeed, the network architect seeking to configure a TSN communication network faces two problems, namely how to configure the TAS protocol of the TSN communication network and how to obtain a guaranteed calculation of worst case end-to-end latencies.

[0109] The two problems are addressed successively in what follows by first presenting the problem and then by setting out the proposed solution.

[0110] Problem of configuring the TAS protocol of the TSN communication network

[0111] The question arises here of how to correctly configure the mechanisms of the TAS protocol in a TSN communication network?

[0112] Configuring the TAS protocol consists of determining the start times of critical windows and their size at each output port of a node or switch through which critical VL flows are emitted.

[0113] The complexity of configuring the TAS protocol arises in particular from the fact that the TSN communication network is distributed.

[0114] Thus, for the transmission of critical VL flows, it is necessary to synchronize the end time of a critical window at the level of an output port of a given node (or switch) with the start time of the next critical window at the level of an output port of the connected switch.

[0115] This synchronization of critical windows aims to streamline the transmission of critical frames across the network and to minimize, upon arrival at an output port, the time spent in the queue waiting for the start of the next critical window.

[0116] To solve such a problem, it is proposed to define a set of rules for the configuration of critical windows (the configuration of less critical windows is obtained de facto from the configuration of critical windows) as well as the configuration of guard intervals.

[0117] These rules also allow successive critical windows on the critical VL flow path to be synchronized with each other (from a given output port to the next output port) on the critical VL flow path through the TSN communication network operating according to the TAS protocol.

[0118] The rules differ depending on the type of output port concerned, namely whether the output port is an output port of a node or an output port of a switch.

[0119] The configuration rules to be applied to the output port of each node include the following configuration rules:

[0120] - a first configuration rule according to which the periods of all the flows critical VLs emitted by the output port of a node are harmonic, that is, the period of each critical VL stream is an integer multiple of the shortest of the periods of the critical VL streams emitted by that output port.

[0121] - a second configuration rule according to which the cycle period is the The shortest period among the critical VL flows emitted by the node's output port. A cycle is the set formed by a less critical window and the following critical window at an output port (node ​​or switch).

[0122] In the following, it is assumed that there is only one critical window (for the transmission of critical VL flows emitted by the output port) and only one less critical window (for the transmission of less critical VL flows emitted by the output port) per cycle.

[0123] It would be possible to develop the same considerations for a cycle case involving several critical or less critical windows.

[0124] Furthermore, a realistic assumption is also used according to which the frames of the critical VL streams are periodically emitted by their emitting node.

[0125] - a third configuration rule according to which the size of a critical window is greater than or equal to the sum of the maximum times required to transmit a frame of each critical VL stream emitted by the node's output port.

[0126] - a fourth configuration rule according to which the size of the guard interval corresponds to the maximum size of a less critical frame.

[0127] In some cases, this set of configuration rules is supplemented by additional configuration rules.

[0128] Indeed, the TAS protocol of the TSN communication network allows the network architect, if desired, to synchronize the sending of critical VL flow frames at the output port of a node with the start of critical windows at the port. This choice is relevant since it minimizes the waiting time of critical VL flow frames before they are sent within the window. Synchronization is configured through the definition of time phases, which will be referred to as "offsets".

[0129] Thus, the synchronization of two elements consists of defining an offset for each of these elements with respect to a common time reference.

[0130] For such a choice of synchronization of critical VL flows and their critical window, the preceding set of configuration rules further includes a fifth configuration rule according to which the offset of each critical VL flow at the output port of a node is the offset of the critical window associated with the output port of the node.

[0131] In the case where synchronization is not desired, this results in a higher worst case transmission delay of the frames as illustrated in [Fig.3].

[0132] Furthermore, for this case, the fifth configuration rule is expressed differently.

[0133] According to this fifth modified configuration rule, the offset of each critical VL flow at the output port of a node is less than or equal to the offset of the critical window associated with the output port of the node or the offset of each critical VL flow at the output port of a node is greater than or equal to the offset of the critical window associated with the output port of the node.

[0134] The rules to be applied to the output port of a switch are now examined.

[0135] For the remainder, a predecessor output port is defined as follows.

[0136] For a given output port of a switch and through which critical flows VL are emitted, a predecessor output port designates an output port that lies in the path of some of these critical VL flows and is directly connected to the switch of the considered output port.

[0137] For example, in the case of [Fig. 1], the output port of switch SI to node ES4 can emit critical VL flows. Its predecessor output ports are the output port of node ESI, the output port of node ES2, and the output port of switch S2.

[0138] Similarly, a predecessor critical window can be defined as follows. For a given critical window associated with an output port of a switch, a predecessor critical window refers to a critical window associated with a predecessor output port of the output port of the window under consideration.

[0139] The configuration rules to be applied to the output ports of each switch include the following configuration rules:

[0140] - a first configuration rule according to which the periods of the cycles of all The critical windows preceding a given critical window linked to an output port of a switch are harmonic, that is, the period of each of these cycles is an integer multiple of the shortest among the periods of all the cycles.

[0141] - a second configuration rule according to which the cycle period of a The critical window associated with the output port of a switch is the shortest period among the cycle periods of all predecessor critical windows of the critical window under consideration.

[0142] - a third configuration rule according to which the size of a critical window linked to an output port of a switch is greater than or equal to the sum of the sizes of all predecessor critical windows of the critical window under consideration.

[0143] - a fourth configuration rule according to which the size of the guard interval corresponds to the maximum size of a frame that a less critical frame can have.

[0144] - a fifth configuration rule according to which there is no intersection in The time between a critical window associated with a switch's output port and all its predecessor critical windows

[0145] Compliance with the preceding configuration rules ensures the following: • Any critical frame emitted by an output port of a node or switch is guaranteed to be directly transmitted during the next critical window associated with the output port. • For each critical VL stream, a maximum of one frame is transmitted per critical window. This ensures that the transmission of a given frame is not disrupted by the transmission of delayed frames from the same critical VL stream. • For each critical VL flow, the end-to-end worst-case latency does not depend on the traffic in the TSN communication network (see latency calculation below). This latency depends solely on the size, period, and start times of the critical windows along its path. • Any less critical frame whose transmission has been authorized by the output port is guaranteed to complete its transmission before the start of a critical window. A less critical frame can therefore never "stuck" in a critical window. Similarly, a critical frame will complete its transmission within the critical window before the start of the next less critical window.

[0146] In other words, compliance with the configuration rules described above ensures that the transmission of a critical frame will not be disrupted by that of less critical frames. Nor is the transmission disrupted by the transmission of lagging critical frames from the same critical VL stream.

[0147] The use of configuration rules also makes it possible to limit the transmission time of any critical frame at the level of a given output port: this transmission time is guaranteed to be less than or equal to the size of the critical window in this output port during which this frame will be transmitted.

[0148] Adherence to the configuration rules also ensures that two frames of the same critical VL flow will never be transmitted within the same critical window. This is obviously contingent upon the frames being sent periodically from their sending node (which is generally the case in industrial communication networks).

[0149] This guarantee is very important because it excludes the formation of "bursts" of critical frames in the network and considerably reduces the jitter of the critical VL stream concerned at the receiving node (the jitter being often a real-time constraint in critical systems, this jitter can thus be conceived as a delay).

[0150] Problem of guaranteed calculation of worst-case end-to-end latencies

[0151] The aim here is to determine how to calculate, for a given configuration of the mechanisms of the TAS protocol, the worst case latencies of the critical VL streams transmitted?

[0152] The calculated worst case latencies are to be guaranteed, that is to say it is necessary to prove that for a critical VL flow, the worst case latency calculated on the basis of the TSN communication network model and the TAS protocol configuration is greater than or equal to the worst latency that any frame of this critical VL flow may experience during its transmission.

[0153] To this end, it is proposed to use the calculation rules which will be set out after the introduction of some definitions and notations: • All the less critical VL flows mentioned include less critical data whose transmission takes place during less critical windows. • VL notation; designates a less critical flow for which the worst-case latency of end-to-end transmission across the TSN communication network (operating according to the TAS protocol) is calculated, • A Ti frame corresponds to a frame of the less critical VL stream; • The activation of a frame at an output port refers to the moment when the frame enters its transmission queue at the output port. • Pj is the transmission period of a frame containing the less critical data stream VLj, • for any less critical flow VLj, Cj denotes the maximum transmission time of the largest frame of the less critical flow VLj through the output port of the node or switch in question (i.e., the transmission time assuming the frame is not disturbed), • for any less critical flow VLj, Jj denotes the maximum jitter (maximum delay) that the activation of a frame Tj of the less critical flow VLj can have, • Lat(VLi) denotes the local worst-case latency that a frame of the less critical VL stream can experience during its transmission through an output port. This worst-case latency is the longest duration of the time interval that begins with the activation of a frame of the less critical VL stream and ends when the transmission of the frame through the considered output port is completed. • hp(VLi) designates the set of less critical VL flows with higher priorities than VL; which are emitted by the same considered output port, • sp(VLi) denotes the set of less critical VL flows with a priority equal to VU that are emitted by the same considered output port, and • lp(VL) denotes the set of less critical VL flows with lower priorities than the VL flow; which are emitted by the same considered output port.

[0154] The worst case end-to-end latency that a Ti frame of the VL stream could experience is calculated in "slices".

[0155] Each slice corresponds to the worst-case transmission latency of the frame Ti through one of the output ports traversed during its path. This latency is called a worst-case local latency.

[0156] The worst case end-to-end latency of the Ti frame is therefore obtained by adding all the worst case local latencies through all the output ports traversed by the Ti frame, to which must be added the technological latencies of all the switches traversed.

[0157] The aim here is therefore to calculate the local worst case latency that the Ti frame can experience.

[0158] To this end, the worst-case “local” scenario that a could experience is first constructed A Ti frame at the output port of a node or switch. Once this worst-case scenario is constructed, we can calculate the local worst-case latency of the Ti frame.

[0159] To determine the local worst-case scenario of the Ti frame at a given output port, the set of potential sources of disturbance that the Ti frame could experience includes the following sources of disturbance: • a first source of disruption corresponding to the disruption of frame Ti by lower priority frames, • a second source of disturbance corresponding to the disturbance of the Ti frame by less critical frames having the same priority and belonging to VL flows other than the VL flows; • a third source of interference corresponding to the interference of frame Ti by less critical frames of lower priority, • a fourth source of disturbance corresponding to the disturbance of the Ti frame by guard intervals, • a fifth source of disturbance corresponding to the disturbance of the Ti frame by critical windows, and • a sixth source of disturbance corresponding to the disturbance of frame Ti by other frames of the VL stream; which precede it.

[0160] In what follows, the scenario allowing the considered disturbance to be maximized is identified and its contribution to the worst case local latency of frame Ti is calculated.

[0161] Calculation of the first source of disturbance, namely the worst-case local disturbance of the transmission of latrameTi by frames of priority VL streams

[0162] The calculation of the worst-case contribution to local latency of the first source of disturbance is based on several considerations.

[0163] According to a first consideration, the maximum disruption of frame Ti by a priority frame Tj belonging to the VLj stream occurs when frame Ti and frame Tj are activated at the same instant. This results in frame Ti being blocked for the entire duration of frame Tj's transmission.

[0164] However, the Ti frame can be blocked by several priority frames of the VLj stream. These frames are designated frame Tj i, frame Tj>2, ..., frame Tj>n.

[0165] Thus, according to a second consideration, the maximum contribution of the frames of the VLj stream to the local latency of the Ti frame is obtained when the Tj i frame, a first frame of the VLj stream, is activated with maximum jitter at the same time as the Ti frame. All subsequent frames of the VLj stream (the Tj2, Tj3, ..., Tjn frames) are activated as soon as possible after the activation instant of the Tj i frame of the Ti frame.

[0166] The preceding activation scheme ensures maximum contribution of frames from the VLj stream to the local latency of frame Ti.

[0167] For the entire less critical priority flow VL disrupting the transmission of frame Ti, the guard interval mechanism means that their contribution to the latency of frame Ti will depend on the transmission order of their respective frames within the less critical window.

[0168] Given the complexity of testing all possible combinations of frame transmission orders, according to a third consideration, the following approximation is used: any ongoing transmission of a priority frame Tj within the less critical window that does not complete before the start of the guard interval is preempted by the guard interval. Transmission resumes at the start of the next less critical window.

[0169] The three considerations allow us to calculate an upper bound (denoted by ^hp) of the contribution to the local latency of the VL flow; by the less critical priority VL flows at the output port level of the node or switch considered.

[0170] The equation for calculating this upper bound is the following equation 1:

[0171] , x (LattT^J \ Ihp( P]

[0172] Where: • hp[VLj] designates the set of lower priority frames, * Ihp(Tj) denotes the value of the contribution, • Cj denotes the transmission time of the largest frame of the data stream j through the output port, • Lât[ Tj) denotes the longest time interval between the activation of frame Tj and the moment when the transmission of that frame starts, • Jj denotes the maximum activation jitter of a frame in a stream j, Activation is the moment when the frame enters a transmission queue in the output port, and • Pj designates the transmission period of frames of a stream j.

[0173] Calculation of the second source of disturbance, namely the disturbance of frame Ti by less critical frames having the same priority and belonging to VL flows other than the VL flow t

[0174] The calculation of the worst-case contribution to local latency of the second source of disturbance is based on several considerations.

[0175] According to a first consideration, the transmission of a frame Tj cannot disrupt that of a frame Ti having the same priority unless it precedes it in time.

[0176] The maximum perturbation of the Tipar frame by the Tj frame takes place when the Tj frame is activated at a delay e before the Ti frame such that £ -> 0.

[0177] According to a second consideration, it is possible that several frames of the VLj stream may disrupt the transmission of frame Ti if these frames are activated before frame Ti.

[0178] However, not all of these frames belonging to the VLj stream can be activated at a delay e before frame Ti such that £ ^ 0, the aim being to maximize their disruption in the transmission of frame Ti.

[0179] It is therefore necessary to consider several possible cases of activation of these frames. In each case, the activation of frame Ti is preceded by an activation of a given frame of the VLj stream by a delay e before the activation of frame Ti such that £ ^ 0.

[0180] The set of possible cases and the number of combinations to be established taking into account all less critical VLs of the same priority level as the VL flow; being high, an approximation of this calculation consists of considering frames of equal priority to frame Ti as being higher priority frames.

[0181] Taking into account the two preceding considerations leads to the following formula 2:

[0182] , , \ Ihp( Ti) ~ ^VL^hp^VL^ J\ P] M

[0183] Where: sptVLi) designates the set of less critical frames having the same priority and belonging to another data stream, designates the value of the contribution,

[0184]

[0185]

[0186]

[0187]

[0188]

[0189]

[0190]

[0191]

[0192]

[0193] - + 1) • Cj denotes the transmission time of the largest frame of the data stream j through the output port, • Lât (Tj) denotes the longest time interval between the activation of frame Tj and the moment when the transmission of that frame starts, • Jj denotes the maximum activation delay of a frame in a stream j, Activation is the moment when the frame enters a transmission queue in the output port, and • Pj designates the transmission period of frames of a stream j. Equation 1 can therefore be extended to include less critical frames of equal priority according to the following equation 3: / LatiT^J Isp (Tj)+ Ihp (T'2JVLj&sp^VL.phpçVL^Cj p- Calculation of the third source of disturbance, namely the disturbance of frame Ti by less critical frames of lower priority The calculation of the worst-case contribution of the third source of disturbance to local latency is based on several considerations. According to a first consideration, the blocking of frame Ti by a less critical frame of lower priority Tl can only take place if the transmission of Tl by the output port of the node or switch in question started before the activation of frame Ti. According to a second consideration, the Ti frame can only be blocked by a single, less critical frame of lower priority. Taking into account these two considerations, the longest perturbation that frame Ti can suffer from a less critical frame of lower priority at the output port of the node or switch in question is caused by a less critical frame of lower priority Ti whose transmission started with a delay e before the activation of frame Ti such that £ 0 and Ci is longer than the transmission time of any other less critical frame of lower priority than frame Ti transmitted by the output port in question. It then occurs that Iip(T for any frame Tk with a priority lower than frame Ti. CkzCt Calculation of the fourth source of disturbance, namely the disturbance of the Ti tram by guard intervals s The worst-case local disturbance of Ti frame transmission by guard intervals is obtained when the activation of the Ti frame coincides with the beginning of a guard interval.

[0194]

[0195]

[0196]

[0197]

[0198]

[0199]

[0200]

[0201]

[0202]

[0203]

[0204] The following equation 4 then allows us to calculate the maximum contribution of the guard intervals (denoted by IGB) to the worst-case local latency of frame Ti transmission: _ / j. i ] L^cle + Or: • Lqb is the duration of a guard interval, and • Lcvcie is the length or period of the cycle at the output port of the node or switch in question. The calculation in equation 4 ignores the maximization of the perturbation of the transmission of a T frame by a lower priority frame (see equation 3). Indeed, assuming that frame Ti is activated at the beginning of a guard interval, this is equivalent to activating the lower priority blocking frame as described above with a delay e before the start of the guard interval such that £ -* 0. The lower priority blocking frame will therefore be transmitted during the guard interval without disturbing the Ti frame. In order to reconcile the maximization of the perturbation of the Ti frame by a lower priority blocking frame and the maximization of the perturbation of the Ti frame by the guard intervals, it is appropriate that the transmission of the lower priority blocking frame, which started with a delay e before the activation of the Ti frame and the activations of all less critical higher priority frames such as £ ^ 0, ends just at the beginning of the guard interval. Taking this consideration into account, we arrive at the following equation 5: / Lat(TfCj \ IGB=LGB\ Lcyd / +1J Or: * IGB refers to the contribution related to the presence of guard intervals, * ^gb denotes the duration of a guard interval, • Cj denotes the maximum transmission time of a frame in the data stream to which the frame belongs, • Lat(Tj) denotes the longest time interval between the activation of frame T and the moment when the transmission of this frame starts, and • LCyCje denotes the time of the cycle. Calculation of the fifth source of perturbation, namely the perturbation of a train T by critical windows

[0205] The worst-case local disturbance of the transmission of the Ti frame through critical windows is obtained when the activation of the T frame corresponds to the beginning of a guard interval.

[0206] This consideration stems from the fact that every critical window is preceded by a guard interval.

[0207] _ pm-c 1GC~LGC\ Lcvcle + 1

[0208] Where: * ^GC denotes the contribution related to the presence of critical windows, • ^GC denotes the duration of a critical window, • Cj denotes the maximum transmission time of a frame in the data stream to which the frame belongs, • Ld.t ( Tj) denotes the longest time interval between the activation of frame Tj and the moment when the transmission of that frame starts, and • LCyC}e denotes the duration of a cycle.

[0209] Equation 5 can therefore be extended to include the maximum contribution of critical windows to the worst-case local latency of T-frame transmission; as follows:

[0210] z (Lat(TfC Igb + IgC = (LGB + Lgc) L^le 1

[0211] Calculation of the sixth source of disturbance, namely the disturbance of frame T, by other frames of the VL i stream which precede it.

[0212] It is possible to calculate the worst-case local latency of frame T transmission via an output port of a given node or switch according to the following equation 7:

[0213] Lat^T^ = Ci+Ilp(Ti)+Isp( <Ti')+Ibp(Ti)+IGB+IGC

[0214] Based on the previous equations, it then follows that: (Lat^T^J, \ (Lat(T,)-C. \ Lat( = Q+ C1+'^VLjesp^VLy)bp(yL^Cj[ p, I-1] + (Lgb+ Lgc) [ L^!fi +1 j

[0215] It can be noted that the term Lat( ) is present on both sides of equation 7.

[0216] This is therefore an equation that can be solved using the fixed-point algorithm by initiating the calculation with a value Lat{ Tj) = Gy

[0217] At each iteration of the algorithm, a new value of Lat( Tj) is calculated according to equation 7.

[0218] As long as this new value is greater than that of the previous iteration, the algorithm continues to iterate.

[0219] The algorithm stops when two successive iterations provide the same value of Lat( T^.

[0220] Thus, equation 7 can be redefined as follows in order to take into account the iterative aspect necessary for its solution: Laü (T}) = C.+ C}+^VLjesp(VLphl^v^^ p': l-1] + (Lgb+Lgc) (F1 j

[0222] This preceding expression does not take into account the sixth source of disturbance, that is to say the impact of the frames of the VL stream; which precede the Ti frame in the calculation of its worst case local latency.

[0223] Indeed, the transmission of the T frame; may be blocked by an ongoing transmission of a previous frame that is behind the VL stream; (a frame that has suffered a blockage which has caused it to be caught up by the T frame;).

[0224] The transmission of these frames therefore has priority over that of frame Ti.

[0225] In order to maximize the perturbation of frame Ti by these frames, the activation scheme of all preceding and lagging frames of the VL stream therefore corresponds to that of frames of higher or equal priority of the VL stream, as described previously.

[0226] We can note the first frame T; idu flux VL; whose activation takes place with a delay e after the start of the transmission of the longest least priority frame such as £^0.

[0227] It is noted Ti>2, Ti>3, ..., T; nTset of frames of the VL stream; such that the end of the transmission of frame Tiq^a takes place after the activation instant of frame Tiq.

[0228] Based on equation 8, it is then possible to calculate for any frame Tiq (with Q > 1) the duration of the time interval W( T ^g] starting with the activation of frame Ti 4 and ending with the end of the transmission of frame Ti q.

[0229] This can be written mathematically according to the following equation 9:

[0230] ^i(T,ç)=q*Ci+Ci+2vl^l^^ 1)+ ( 7 gj +

[0231] The worst-case local latency of Tig is thus obtained by taking into account its activation time using equation 10:

[0232] La t (Ti q) = W (Ti g) - Max [(q -1) Pr0]

[0233] Thus, to obtain the worst-case local latency of the VL stream, the worst-case local latency is calculated for all frames Ti i, Ti 2, Ti3, ..., Tindu of the VL stream. The longest of the calculated latencies is the one that will determine the worst-case local latency of the VL stream.

[0234] This allows an operator wishing to configure a TSN communication network to operate according to the TAS protocol to do so by implementing a worst-case end-to-end frame transmission latency estimation method in a TSN communication network operating according to a TAS protocol,

[0235] the network comprising nodes, each node being connected to a switch, the switches being interconnected to ensure the exchange of data streams between the nodes, each data stream comprising a set of frames, the transmitted frames being critical or less critical and having a transmission priority, a frame with a higher transmission priority being transmitted before a frame with a lower transmission priority, the exchange of data streams taking place during time windows, some time windows being critical windows exclusively dedicated to the transmission of critical streams and other time windows being less critical windows dedicated to the transmission of less critical streams,

[0236] the process being implemented by computer and comprising:

[0237] - for each data stream, a step to calculate the latency value of worst-case end-to-end transmission that a frame in the data stream can undergo,

[0238] the calculation step comprising, for each frame of the data stream and each output port of a node or switch through which a data stream can pass, the following operations:

[0239] - worst case disturbance estimation for frame transmission on the port output considered from a plurality of contributions, the plurality of contributions including:

[0240] - an initial contribution related to other, less critical frameworks,

[0241] - a second contribution related to the presence of guard intervals preventing the starting the transmission of a less critical stream within a predefined time interval just before a transmission window for critical streams,

[0242] - a third contribution related to the presence of critical windows, and

[0243] - calculation of the sum of contributions, the sum of contributions being the value of worst-case transmission latency of the frame,

[0244] the calculation step comprising, in addition:

[0245] - a worst-case transmission latency value selection operation the highest frame for all frames in the data stream for each output port, and

[0246] - an operation of summing the different values ​​selected for each output port, to obtain a summation result, the summation result value being the estimated value of the end-to-end worst-case transmission latency of the data stream, and

[0247] - a step for estimating the worst-case end-to-end transmission latency of a TSN communication network based on estimated values ​​for each data stream. + 11

[0248] According to a particular example, the operation of estimating the first contribution comprises:

[0249] - the estimation of a first sub-contribution related to less critical frames priorities,

[0250] - the estimation of a second sub-contribution related to less critical frames having the same priority and belonging to a different data stream,

[0251] - the estimation of a third sub-contribution related to less-critical frameworks of lower priority,

[0252] - the estimation of a fourth sub-contribution related to other frames of the stream of data preceding the frame under consideration, and

[0253] - the addition of the first sub-contribution, the second sub-contribution, of the third sub-contribution and the fourth sub-contribution, to obtain the first contribution.

[0254] Preferably, the estimation of the first sub-contribution is implemented by calculating the equation:

[0255] / T,^ ^pk1 i) 2j VLfihpÇVQ CJ \ Pj

[0256] Where: hp(VL,) denotes the set of lower-priority, less-critical frames, • ^hp( Tj) denotes the value of the first sub-contribution, • Cj denotes the transmission time of the largest frame in the data stream j, • Lât (Tj) denotes the longest time interval between the activation of the Tj frame and the moment when the transmission of the Tj frame starts, • Jj denotes the maximum activation jitter of a frame in a stream j, Activation is the moment when the frame enters a transmission queue, and • Pj designates the transmission period of frames of a stream j.

[0257] Advantageously, the estimation of the second sub-contribution is implemented by calculating the equation:

[0258] T ix / Lat(T^-J Isp(Ti) —^vL^s^VL^j\ Pj

[0259] Where: • Sp(VLj) designates the set of less critical frames having the same priority and belonging to another data stream, IspiT;) denotes the value of the second sub-contribution, Cj denotes the transmission time of the largest frame in the data stream j, Lât (Tj) denotes the longest time interval between the activation of the Tj frame and the moment when the transmission of the Tj frame starts, Jj denotes the maximum activation jang of a frame in a stream j, activation being the instant when the frame enters a transmission queue in the output port, and Pj denotes the transmission period of frames in a stream j.

[0260] According to an embodiment that can be combined with the preceding ones, the third sub-contribution is calculated as equal to the maximum transmission time of a frame belonging to a less critical flow of lower priority than the flow for which the worst case end-to-end transmission latency is estimated.

[0261] Similarly, the estimate of the fourth sub-contribution is calculated as equal to the maximum transmission time of previous lagging frames of the same stream for which the end-to-end worst-case transmission latency is estimated.

[0262] According to an advantageous embodiment, the critical windows and the less critical windows alternate according to a cycle of a less critical window and a critical window, the estimation of the second contribution related to the presence of guard intervals being implemented by calculating the equation:

[0263] _ / Lat{T^Cj \ i GB - Lgb + 1 /

[0264] Where: * GB denotes the value of the second contribution, • LGb denotes the duration of a guard interval, • Cj denotes the longest maximum transmission time of a frame in the data stream j to which the frame belongs, • Lât(Tj) denotes the longest time interval between the activation of frame Tj and the moment when the transmission of frame Tj starts, and • Lcvcle denotes the duration of a cycle.

[0265] Preferably, the critical windows and the less critical windows alternate according to a cycle of a less critical window and a critical window, and wherein the estimation of the third contribution related to the presence of critical windows is implemented by calculating the equation:

[0266] _ / LatiT^C . \ IGC = LGC LcycIr + 1 /

[0267] Where: * -^GC denotes the value of the third contribution, • ^GC denotes the duration of a critical window, • Cj denotes the maximum transmission time of a frame in the data stream j to which the frame appears, • Ld.t[ Tj) denotes the longest time interval between the activation of frame T and the moment when the transmission of frame T starts, and • Lcycie refers to the duration of a cycle.

[0268] The process just described makes it possible to calculate for each VL flow (critical or less critical) of the TSN communication network operating according to the TAS protocol the worst case end-to-end latency that one of its frames could suffer.

[0269] The calculated latency is guaranteed to be correct in the sense that its value is always greater than or equal to the actual latency.

[0270] The calculation is based on a model of the TSN communication network architecture, its TAS configuration and the flows transmitted there.

[0271] This guaranteed end-to-end worst-case latency calculation of VLs streams (including less critical streams) is extremely fast (on the order of a few seconds) compared with other known methods.

[0272] The present method enables the network architect to explore several configuration solutions and make the necessary corrections if required, and then to validate the real-time requirements in the TSN communication network before proceeding to the target implementation.

[0273] In other words, an application of the calculation method described above is a method for configuring the TSN communication network operating according to the TAS protocol.

[0274] This process is implemented by computer and comprises the following steps:

[0275] - obtaining an initial network configuration, the configuration comprising the network topology and the data flows intended to circulate on the network,

[0276] - application of a worst-case end-to-end transmission latency estimation method end-to-end on the network corresponding to the initial configuration to obtain the worst-case end-to-end network latency,

[0277] - calculation of a configuration validation criterion, the criterion being applied to the worst-case end-to-end latency value of the estimated network,

[0278] the final network configuration being the initial configuration if the validation criterion is met,

[0279] the process comprising, furthermore, if the validation criterion is not met, the steps of:

[0280] - modifying the network configuration to obtain a modified configuration,

[0281] - implementation of a worst-case transmission latency estimation method end-to-end on the network corresponding to the modified configuration to determine the worst-case end-to-end latency of the network,

[0282] - test of the validation criterion on the worst-case end-to-end network latency,

[0283] the modification, implementation and testing steps being repeated until to obtain a final modified network configuration of the network verifying the criterion, the final network configuration being the final modified configuration.

[0284] Advantageously, the method includes, at each iteration, a step of verifying an additional condition relating to the size of the critical window and the size of the guard interval.

[0285] Preferably, the method includes, at each iteration, a step of checking an additional condition relating to the period of the clock cycle and the period of each data stream.

[0286] According to one embodiment, the method includes, at each iteration, a step of verifying an additional condition relating to the phase shift of each critical data stream and the associated critical window

[0287] The method also has the advantage of providing evidence of the validity of the real-time behavior of critical flows (which represents a strong argument for the certification of critical systems).

[0288] By deriving this calculation, it is possible to define rules to ensure that the worst case end-to-end transmission latency of a critical frame Ti through the TSN communication network operating according to the TAS protocol is bounded by the distance between the start time of its critical window of transmission of the frame Ti linked to the output port of the transmitting node and the end time of the critical window of the frame Ti linked to the output port of the last switch located on its path.

[0289] This worst-case end-to-end latency is therefore totally independent of the traffic in the TSN communication network. It depends solely on the configuration of the critical windows.

[0290] When these rules are respected by the architect, he obtains the following guarantees: • any critical frame emitted by an output port of a node or switch is guaranteed to be directly transmitted during the next critical window associated with the output port. • For each critical VL stream, a maximum of one frame is transmitted through a critical window. This ensures that the transmission of a given frame is not disrupted by the transmission of delayed frames from the same VL stream. • For each critical VL flow, the end-to-end worst-case latency does not depend on the traffic in the TSN communication network (see the calculation of the latency below). It depends solely on the size, period and start times of critical windows in its path. • Any less critical frame whose transmission has been authorized by the output port is guaranteed to complete its transmission before the start of a critical window. A less critical frame can therefore never "stuck" in a critical window. Similarly, a critical frame will complete its transmission within the critical window before the start of the next less critical window.

[0291] The processes described thus make it possible to provide a solution to the two problems mentioned above, namely how to configure the TAS protocol of the TSN communication network and how to obtain a guaranteed calculation of worst case end-to-end latencies.

Claims

1. Demands Method for estimating worst-case end-to-end frame transmission latency in a TSN communication network operating according to a TAS protocol, the network comprising nodes, each node being connected to a switch, the switches being interconnected to ensure data stream exchange between the nodes, each data stream comprising a set of frames, the transmitted frames being critical or less critical and having a transmission priority, a frame with a higher transmission priority being transmitted before a frame with a lower transmission priority, the data stream exchanges taking place during time windows, some time windows being critical windows exclusively dedicated to the transmission of critical streams and other time windows being less critical windows dedicated to the transmission of less critical streams,The process being implemented by computer and comprising: - for each data stream, a calculation step of the worst-case end-to-end transmission latency value that a frame of the data stream may experience, the calculation step comprising, for each frame of the data stream and each output port of a node or switch through which a data stream may pass, the following operations: - estimation of the worst-case disturbance for the transmission of the frame on the considered output port from a plurality of contributions, the plurality of contributions comprising: - a first contribution related to other less critical frames, - a second contribution related to the presence of guard intervals preventing the start of the transmission of a less critical stream in a predefined time interval just before a transmission window of critical streams, - a third contribution related to the presence of critical windows,and - calculation of the sum of contributions, the sum of contributions being the worst-case transmission latency value of the frame, the calculation step further comprising: - an operation to select the highest worst case transmission latency value for all frames in the data stream for each output port, and - an operation to sum the different values ​​selected for each output port, to obtain a summation result, the value of the summation result being the estimated value of the end-to-end worst case transmission latency of the data stream, and - a step to estimate the end-to-end worst case transmission latency of a TSN communication network from the estimated values ​​for each data stream.

2. A calculation method according to claim 1, wherein the first contribution estimation operation comprises: - estimating a first sub-contribution related to lower priority less critical frames, - estimating a second sub-contribution related to less critical frames having the same priority and belonging to another data stream, - estimating a third sub-contribution related to lower priority less critical frames, - estimating a fourth sub-contribution related to other frames in the data stream that precede the frame under consideration, and - adding the first sub-contribution, the second sub-contribution, the third sub-contribution and the fourth sub-contribution, to obtain the first contribution.

3. A calculation method according to claim 2, wherein the estimation of the first sub-contribution is implemented by calculating the equation: (Lat(T^J. \ Ihp( Ti) Pj Where:

4. hp( VL^ denotes the set of lower-priority critical frames, * Ihp(Ti) denotes the value of the first sub- contribution, • Cj denotes the transmission time of the largest frame in the data stream j, • Lat( T^) denotes the longest time interval between the activation of frame Tj and the moment when the transmission of frame Tj starts, • Jj designates the maximum activation jitter of a frame of a stream j, activation being the moment when the frame enters a transmission queue, and • Pj designates the transmission period of frames of a stream j. Calculation method according to claim 2 or 3, wherein the estimation of the second sub-contribution is implemented by calculating the equation: = ^VLj&sp(VL^j Or: Pi • Sp(VLj) denotes the set of less critical frames having the same priority and belonging to another data stream, • ISp(Tj) denotes the value of the second sub-contribution, • Cj denotes the transmission time of the largest frame of data stream j, • Lât(Tj) denotes the longest time interval between the activation of frame Tj and the moment when the transmission of frame Tj starts, • Jj denotes the maximum activation jitter of a frame of a stream j, activation being the moment when the frame enters a transmission queue in the output port, and • Pj denotes the transmission period of frames of a stream j.

5. A calculation method according to any one of claims 2 to 4, wherein the third sub-contribution is calculated as equal to the maximum transmission time of a frame belonging to a less critical stream of lower priority than the stream for which the worst-case end-to-end transmission latency is estimated.

6. A calculation method according to any one of claims 2 to 5, wherein the estimation of the fourth sub-contribution is calculated as equal to the maximum transmission time of previous delaying frames of the same stream for which the end-to-end worst-case transmission latency is estimated.

7. A calculation method according to any one of claims 1 to 6, wherein critical and less critical windows alternate according to a cycle of a less critical window and a critical window, the estimation of the second contribution related to the presence of guard intervals being implemented by calculating the equation: / Lat^T^-C. \ ^GB ~ ^GB ( Lcyc!e + 1 / Where: * IGB denotes the value of the second contribution, • -^GB denotes the duration of a guard interval, • Cj denotes the longest maximum transmission time of a frame in the data stream j to which the frame belongs, • Lât{ Tj) denotes the longest time interval between the activation of frame Tj and the moment when the transmission of frame Tj starts, and • Lcycie denotes the duration of a cycle.

8. A calculation method according to any one of claims 1 to 7, wherein critical windows and less critical windows alternate according to a cycle of a less critical window and a critical window, and wherein the estimation of the third contribution related to the presence of critical windows is implemented by calculating the equation: / LatiT^-C. \ ^GC - ^GC ( LcycSe + 1 ) Where: * GC denotes the value of the third contribution, • ^GC denotes the duration of a critical window, • Cj denotes the longest maximum transmission time of a frame of the data stream j to which the frame appears, • Ld.t( Tj) denotes the longest time interval between the activation of frame Tj and the moment when the transmission of frame Tj starts, and • LCyCje denotes the duration of a cycle.

9. A method for configuring a TSN communication network operating according to a TAS protocol, the network comprising nodes, each node being connected to a switch, the switches being interconnected to ensure the exchange of data streams between the nodes, each data stream comprising a set of frames, the transmitted frames being either critical or less critical and having a transmission priority, a frame with a higher transmission priority being transmitted before a frame with a lower priority lower transmission, data flow exchanges taking place during time windows, some time windows being critical windows exclusively dedicated to the transmission of critical flows and other time windows being less critical windows dedicated to the transmission of less critical flows, the configuration process being implemented by computer and comprising the steps of: - obtaining an initial network configuration, the configuration including the network topology and the data flows intended to circulate on the network, - application of a worst-case end-to-end transmission latency estimation process on the network corresponding to the initial configuration to obtain the worst-case end-to-end latency of the network, - calculation of a configuration validation criterion, the criterion being applied to the estimated worst-case end-to-end latency value of the network,The final network configuration being the initial configuration if the validation criterion is met, the process further comprising, if the validation criterion is not met, the steps of: - modifying the network configuration to obtain a modified configuration, - implementing a process for estimating the end-to-end worst-case transmission latency on the network corresponding to the modified configuration to determine the end-to-end worst-case latency of the network, - testing the validation criterion on the end-to-end worst-case latency of the network, the modification, implementation, and testing steps being repeated until a final modified network configuration is obtained that satisfies the criterion, the final network configuration being the final modified configuration.

10. A configuration method according to claim 9, wherein the method includes, at each iteration, a step of verifying an additional condition relating to the size of the critical window and the size of the guard interval.

11. A configuration method according to claim 9 or 10, wherein the method comprises, at each iteration, a verification step of an additional condition relating to the clock cycle period and the period of each data stream.

12. A configuration method according to any one of claims 9 to 10, wherein the method includes, at each iteration, a step of verifying an additional condition relating to the phase shift of each critical data stream and the associated critical window.

13. TSN communication network operating according to a TAS protocol, the network comprising nodes, each node being connected to a switch, the switches being connected to each other to ensure an exchange of data streams between the nodes, each data stream comprising a set of frames, the transmitted frames being critical or less critical and having a transmission priority, a frame having a higher transmission priority being to be transmitted before a frame having a lower transmission priority, the exchanges of data streams taking place during time windows, some time windows being critical windows exclusively dedicated to the transmission of critical streams and other time windows being less critical windows dedicated to the transmission of less critical streams, the network being configured by implementing a configuration method according to any one of claims 9 to 12.

14. Product computer program comprising a readable information carrier, on which is stored a computer program comprising program instructions, the computer program being loadable onto a data processing unit and adapted to drive the implementation of at least one step of a process according to any one of claims 1 to 12 when the computer program is implemented on the data processing unit.

15. Readable information carrier comprising program instructions forming a computer program, the computer program being loadable onto a data processing unit and adapted to drive the implementation of at least one step of a method according to any one of claims 1 to 12 when the computer program is implemented on the data processing unit.