Transmission window generation method for calculating TT flow based on qualified time in TSN network

Abstract

The invention discloses a transmission window generation method for calculating TT flow based on qualified time in a TSN network. The method comprises the following steps: determining a transmission time window of time-triggered flow according to the qualified time; the method comprises the following steps: calculating transmission time windows of different flows on different nodes based on a token bucket mechanism according to information such as topology network architecture, flow routing, period and priority; firstly, nodes are sorted according to the sequence of the nodes on a flow path; then, according to a node arrangement sequence, determining a starting point of a data frame transmission time window on each node through calculation of qualified time of a frame, and calculating an ending point of the transmission time window in combination with a frame length and a node bandwidth; according to the method, through calculation based on qualified time, a traffic transmission window which ensures complete transmission of data frames and does not affect each other in data frame transmission can be generated more efficiently, and scheduling solution can be carried out on a time-triggered scheduling scene with a large number of nodes, a large traffic scale or multi-priority coexistence.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network traffic transmission, in particular to a method for generating a transmission window of TT traffic based on eligible time in a TSN network. BACKGROUND

[0002] Time-Sensitive Networking (TSN) is a protocol system built by IEEE standardization organization to expand the application requirements and scope of deterministic Ethernet based on the inheritance and development of real-time transmission capabilities of Audio Video Bridging (AVB) technology. The core mission of TSN network is to provide unified and high-performance network infrastructure for fields such as industrial control, intelligent connected vehicles and aerospace, which have strict requirements on communication determinism, low delay and high reliability. In order to meet the requirements of TSN network on transmission delay, TSN working group specially defines Time-Aware Shaper (TAS) in IEEE 802.1Qbv standard. The transmission scheduling of the queue can be adjusted based on the preset time scale. To achieve this goal, each queue is associated with a transmission gate device, whose state determines whether the frame in the queue can be selected for transmission. An Evaluation of SMT-based Schedule Synthesis For Time-Triggered Multi-Hop Networks published in 2010 31st IEEE Real-Time Systems Symposium proposes to transfer the remaining time resources unused by TT traffic to RC traffic according to "schedule interpretation".

[0003] Time-Aware Shaper (TAS) divides different types of traffic (reference Traffic Type Classification in Time-Sensitive Networking Traffic Shaping Technology Review published in January 2022, Volume 39, Issue 1 of Microelectronics and Computer) into different transmission queues at the port of the switching node. Whether the transmission queue transmits traffic is controlled by the Gate Control List (GCL) mechanism, which is configured by the time transmission window of the traffic. Each data frame can only be transmitted within the specified transmission time window, thereby isolating the interference of other types of traffic on the current transmission traffic in the time dimension.

[0004] In a time-triggered communication system, the generation of transmission time windows needs to establish a constraint model, including timing constraints, resource constraints and logical transmission dependencies. The solving process uses satisfiability modulo theories (SMT) or mixed integer linear programming method (MIP) to find the traffic transmission time window configuration that meets the preset constraints. The problem belongs to the NP-hard complexity category in nature, which leads to the solving engine having to perform backtracking search and state space traversal operations. Specifically, when local constraints conflict, the SMT solver needs to back up to the previous decision point to re-explore the path, and the MIP solver needs to repeatedly perform branch cutting to explore the integer solution space.

[0005] To overcome the problem of limited solution size caused by backtracking search and state space traversal operations in the transmission time window generation method based on constraint solving, the present application proposes a transmission time window generation method for TT traffic based on eligible time calculation. SUMMARY

[0006] To overcome the problem of low efficiency of TT traffic transmission time window generation caused by dependence on backtracking search and state space traversal in the traditional time-triggered window planning method in TSN networks, the present application proposes a TT traffic transmission time window generation method based on eligible time calculation. The specific method is as follows: along the predetermined routing path of the data traffic, the transmission time window of the data frame is calculated on each forwarding node in sequence, thereby generating the transmission time window of the traffic.

[0007] In the present application, the transmission time window of different traffic at different nodes is calculated based on the token bucket mechanism, which is called time token bucket.

[0008] In the present application, the eligible time refers to the earliest time at which the data frame can be sent with sufficient tokens in the time token bucket algorithm.

[0009] The present application is a transmission window generation method for TT traffic based on eligible time calculation in a TSN network, which includes the following steps:

[0010] Step one, obtaining the topology structure of the time-sensitive network and the traffic information of each node;

[0011] In step one, on the one hand, a given topology network structure serving the TSN network is obtained; on the other hand, each node in the topology network structure is obtained; and on the third aspect, the traffic transmitted in the topology network and the traffic information of each node are obtained.

[0012] Step two, node sorting based on dependency relationship;

[0013] In step two, the nodes are sorted according to the given routing path. The order of the nodes on the path for a certain traffic belonging to the routing path is consistent with the order of the nodes in the list. Therefore, in the process of traversing the sorted node list node by node and setting the node transmission time window, it can be ensured that the transmission time window of the node on each traffic path strictly follows the order.

[0014] Step three, calculate the transmission time window of each traffic at different nodes;

[0015] In step three, the frame eligible time of the data frame is calculated at each node as the start time of its transmission time window; for each data frame corresponding to the traffic, the token bucket algorithm is used to calculate the frame eligible time, realizing the preliminary transmission isolation between different frames of the traffic; wherein, the final calculated eligible time for each frame is determined by the maximum value of the frame arrival time, the eligible time of the previous frame and the frame eligible time calculated by the token bucket algorithm; in addition, the result of the frame eligible time calculated by the token bucket algorithm is determined by the values of the token recovery rate CIR, the bucket depth CBS and the frame length of the token bucket, and the traffic from different input nodes and belonging to different priority types will be calculated through different token buckets, if the frame length does not exceed the bucket depth CBS, the data frame is transmitted normally, if the frame length exceeds the bucket depth, the data frame is abandoned; finally, the end time of the transmission time window is calculated in combination with the length of the data frame and the bandwidth of the node.

[0016] Step four, generate the transmission time window of all data frames in the network;

[0017] In step four, after obtaining the transmission time window of the data frame at the current node, the end time of this transmission time window is taken as the arrival time of the next node on the path of the data frame, and the transmission time window of the data frame at each node from the source node to the destination node is obtained by this method.

[0018] In the present application, the TT traffic transmission time window generated by calculating the transmission time window based on the eligible time satisfies the following constraint conditions: the data frame can be completely transmitted within the transmission time window at the current node, which meets the complete transmission constraint; the transmission time window of the data frame on the path maintains the order, which follows the timing constraint; the data frames in the same queue are scheduled in the order of arrival in the transmission time window at the current node, which meets the first-in-first-out constraint; any data frame passing through the same node has a transmission time window that does not overlap with each other, which realizes the collision-free constraint.

[0019] The technical effects achieved by the method of the present application are:

[0020] (1) The present invention uses a token bucket algorithm to calculate the qualified time of the frame as the start time of the transmission window. Specifically, the qualified time that finally takes effect is the maximum of the following three: the time when the frame arrives at the current node, the qualified time of the previous frame, and the qualified time of the frame calculated by the token bucket algorithm. In contrast, the traditional time-triggered window planning method requires the construction of a global constraint model that includes timing, resources, and dependencies, and is solved with the help of satisfiability modulo theory (SMT) or mixed integer linear programming (MILP). This constraint modeling and solution process is essentially an NP-hard problem. In practical applications, it involves backtracking search and state space traversal operations, which limits the scale of schedulable traffic. To overcome the limitations of traditional methods, the present invention proposes a method for generating traffic transmission time windows based on qualified time calculation. This method does not require the construction of a global constraint model and the execution of complex state space search, thereby avoiding backtracking operations.

[0021] (2) The traffic transmission time window generated by the present invention fully complies with existing protocol standards and does not require modification of the existing time-sensitive network scheduling mechanism or hardware configuration. The output traffic transmission time window can be directly applied to switching devices that support the TAS mechanism.

[0022] (3) The present invention avoids the backtracking search process in traditional constraint solving and can support the generation of traffic transmission time windows for thousands of traffic flows in large-scale networks in a short period of time, meeting the real-time scheduling requirements of industrial Internet of Things and vehicle-mounted and airborne networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a flow chart of a method for generating a traffic transmission time window based on qualified time calculation.

[0024] Figure 2 This is a schematic diagram of the principle of the traffic transmission time window generation method based on qualified time calculation.

[0025] Figure 3 This is a topological diagram of the network used in Example 1 of the present invention.

[0026] Figure 4 This is an example diagram of the process of calculating the transmission time window in Example 1.

[0027] Figure 5 It is the transmission time window planned in each node of the network in Example 1.

[0028] Figure 6 This is a comparison chart of the transmission time window of each flow on each node based on the transmission window planning of the topology and traffic in Example 1. DETAILED DESCRIPTION

[0029] The present invention will be further described in detail below with reference to the accompanying drawings and examples. The parameters listed are merely exemplary embodiments of the present invention. It should be noted that those skilled in the art may make various improvements and modifications without departing from the principles of the present invention, and such improvements and modifications are also within the scope of protection of the present invention.

[0030] In the present invention, the network topology in the time-sensitive network is recorded as , and the network topology Contains a node set and connected edge sets ,Right now .

[0031] , subscript Indicates the identification number of the node in the network topology, subscript Represents the total number of nodes in the network topology, where:

[0032] Indicates the first node.

[0033] Indicates the second node.

[0034] Indicates the nodes.

[0035] Indicates the nodes.

[0036] Indicates the last node.

[0037] For the sake of convenience, Also called any node. and Not the same node.

[0038] In the present invention, a node on a traffic path may also be referred to as a node.

[0039] In the present invention, nodes can be distinguished by using consecutive numbers. In the network topology, a directed connection matrix is ​​used to represent the directed edges. , Assigning a value of 1 means there is a directed edge, and assigning a value of 0 means there is no directed edge. The assignment of to indicate whether there is a directed edge between two nodes.

[0040] Any node The node-comprehensive traffic characteristics of the same priority type traffic are recorded as ,and .

[0041] The bandwidth of a node is in Mbps.

[0042] The load rate of a node is The load rate of a node is , and , and wherein: represents the set of all flows passing through node is the flow rate passing through node is the bandwidth of node

[0043] The set of dependent nodes of a node is The set of dependent nodes of a node is composed of all nodes dependent on node

[0044] The in-degree of a node is The in-degree of a node represents the number of directed edges pointing to , i.e., the number of nodes depends on.

[0045] The set of network flows in a network topology is denoted as , and , wherein subscript represents the identification number of a flow, and subscript represents the total number of flows.

[0046] represents the first network flow.

[0047] represents the second network flow.

[0048] represents the network flow.

[0049] represents the network flow.

[0050] represents the last network flow.

[0051] For convenience of illustration, is also referred to as an arbitrary network flow.​​​​​ With Different same network traffic.

[0052] In this invention, a certain data frame belonging to the flow is recorded as .

[0053] In this invention, the traffic characteristic information of any network traffic is recorded as , and .

[0054] Frame length, unit: byte.

[0055] Traffic cycle, unit: ms.

[0056] Priority, value from 0 to 7. Refer to the article "Review of Traffic Shaping Technology in Time Sensitive Network" published in Microelectronics and Computer, Vol. 39, No. 1, January 2022, author: Zhang Lei, Wang Panpan.

[0057] Traffic rate, also the bandwidth occupied by traffic transmission, unit: Mbps. The traffic rate of network traffic is , and .

[0058] The source node of the flow.

[0059] The destination node of the flow.

[0060] The route of the flow, also represents the set of nodes passed through on the transmission path of network traffic .

[0061] The token bucket to which the priority queue of the flow belongs in the node. The priority to which belongs is the queue in the node Token bucket to which the computing node belongs. In the present application, the token bucket to which the computing node belongs adopts a token bucket algorithm, which is described in Video Traffic Analysis and QoS Management, Huang Tianyun, Chengdu: University of Electronic Science and Technology Press, 2013.03, pp. 94-96. The token bucket algorithm is the most commonly used algorithm in network traffic shaping (Traffic Shaping) and rate limiting (Rate Limiting). Tokens are saved in a queue, input data enters the token bucket, and flows out of the token bucket after obtaining sufficient tokens.

[0062] The flow eligibility time is calculated based on the token bucket mechanism in the present application The flow eligibility time .

[0063] The last bucket empty time of the token bucket.

[0064] The frame length of the flow.

[0065] The token growth rate per unit time in the token bucket, with units of Mbps. The flow The token growth rate per unit time in the token bucket to which the computing node belongs is .

[0066] The maximum allowed number of tokens in the token bucket. The frame length of the flow does not exceed the maximum allowed number of tokens in the token bucket to which the computing node belongs. . The unit is byte.

[0067] The present application needs to calculate the transmission time window of the data frame at the node, denoted as , and .

[0068] The data frame.

[0069] The node flag.

[0070] The start time of the transmission time window.

[0071] The end time of the transmission time window.

[0072] Problems in constraint solving method

[0073] ​The non-uniform time slot arrangement problem of the front and rear stage transmission link without conflict overlap under the time-aware shaper (TAS) is one of the key bottlenecks restricting the scheduling solution efficiency. The problem requires ensuring that the transmission between the front and rear stage links does not conflict when allocating time slots for multiple transmission links, while also meeting the arrangement requirements of non-uniform time slots. Such constraint conditions make the problem complexity rise sharply. Because the solution space increases exponentially with the number of links and the time slot scale, the traditional method often has difficulty finding a feasible solution within a reasonable time when facing a large-scale network.

[0074] When using satisfiability modulo theories (SMT) for solving, the problem is transformed into a satisfaction problem of logical constraints, which needs to find a feasible solution through a large number of logical reasoning and constraint solving. However, due to the complexity of non-uniform time slot arrangement and the strictness of non-conflict constraints, the SMT solver often needs to perform multiple iterations and backtracking searches to verify the feasibility of different time slot allocation schemes. This process not only has huge computational overhead, but also may lead to too long solving time, making it difficult to meet the real-time scheduling requirements.

[0075] Similarly, the mixed integer linear programming (MILP) method models the problem as a linear programming model and introduces integer variables to represent time slot allocation decisions. Although ILP can find the optimal solution in theory, due to the NP-hard nature of the problem, a large number of possible time slot combinations still need to be traversed during the solving process, resulting in extremely high computational complexity. Especially in the case of a large number of links and large time slot scale, the computation time of the MILP solver will increase significantly, and it may even be unable to get a feasible solution within a limited time.

[0076] Therefore, the non-uniform time slot arrangement problem of the front and rear stage transmission link without conflict overlap under the TAS is not only theoretically challenging, but also puts high requirements on the scheduling solution method in practical applications.

[0077] Referring to Figure 1 The present application calculates the transmission time window of each flow in the time-sensitive network (TSN) at different nodes by calculating the eligible time of the frame, so as to generate the flow transmission time window of the time-sensitive network. Specifically, it includes the following steps:

[0078] Step one, obtain the topology structure of the time-sensitive network and the flow information of each node;

[0079] Step 101, obtain the topology structure of the time-sensitive network and each node;

[0080] In the present application, the network topology structure in the time-sensitive network is denoted as , the network flow set is denoted as , and the flow characteristic information of any flow is denoted as .

[0081] Network topology contains node set information and edge set information . The network topology uses a directed connection matrix to represent a directed edge, denoted as , which is assigned a value of 1 if there is a directed edge, and a value of 0 if there is no edge. Whether there is a directed edge between two nodes is represented by the assignment of .

[0082] Step 102, traffic marking;

[0083] Any traffic contains a variety of feature information, so the traffic feature information is constructed using a structure in the present application , and

[0084] .

[0085] Similarly, the traffic feature information of traffic is obtained, and . .

[0086] Similarly, the traffic feature information of traffic is obtained, and . .

[0087] Similarly, the traffic feature information of traffic is obtained, and . .

[0088] Similarly, the traffic feature information of traffic is obtained, and . .

[0089] Then, the traffic-flow feature set .

[0090] Step 103, node marking;

[0091] Any node contains a variety of feature information, so the node feature information is constructed using a structure in the present application , and . The load rate of the node is denoted as , and in the initial state, the initial load rate of all nodes , the dependent node set is empty, and the in-degree is 0. ​

[0092] Similarly, the node The node feature information is .node The node feature information is .node The node feature information is .node The node feature information is .

[0093] Then: All node feature sets are recorded as ,and .

[0094] Step 2: Sort nodes based on dependency relationships;

[0095] The node dependencies are confirmed and sorted according to the transmission path of each flow.

[0096] Step 201, building dependencies between nodes;

[0097] The dependency relationship between nodes refers to searching along a certain routing path and finding the next node among the two adjacent nodes on the path. Determined to be dependent on the previous node , recorded as ; At the same time join in The set of dependent nodes During the retrieval process, if the node set Any node in Each time the node is added to the dependent node set of another node, the node Indegree .

[0098] According to this principle, all paths are searched and the dependent node sets and in-degrees of all nodes are updated.

[0099] Step 202, sorting the nodes;

[0100] The first cache queue is denoted as .

[0101] The second cache queue is denoted as .

[0102] initialization and . Set the node All in-degrees The node with value 0 is added to the queue The first update queue is formed in .from The head node is taken out from the head and stored in the head node The dependent node set is retrieved from the head of the queue The corresponding dependent node set The in-degree of all nodes contained If the in-degree of a node in the dependent node set is reduced to 0, the node with in-degree 0 is added to the tail of the queue After all nodes in the dependent node set are retrieved and the state of each node is updated, the head node is taken out from the head of the queue and added to the tail of the queue to form a second update queue After all nodes in the dependent node set are retrieved and the state of each node is updated, the head node is taken out from the head of the queue and added to the tail of the queue to form a second update queue Then is taken as the head of the queue and the tail of the queue is taken as According to this principle, the operation is repeated until there are no remaining nodes in the queue At this time, the queue obtained is the sorted node list, denoted as The first node in the list is denoted as , and also referred to as any node in the node list.

[0103] After obtaining , for a certain flow , the nodes belonging to the routing path of the flow are in the same order on the path as in . Therefore, in the process of traversing node by node and setting the transmission time window of the node, it can be ensured that the transmission time window of the node on each flow path strictly follows the order.

[0104] Step three, calculate the transmission time window of each flow at different nodes;

[0105] The transmission time of the data frame is denoted as .

[0106] The transmission start time of the data frame is denoted as .

[0107] The transmission end time of the data frame is denoted as .

[0108] The arrival time of the data frame is denoted as .

[0109] The qualified time of the data frame is denoted as​​ .

[0110] The flow qualification time is recorded as .

[0111] The group qualifying time is recorded as .

[0112] The last empty time of the token bucket is recorded as .

[0113] To facilitate the description of the transmission time window planning process, it is assumed that the traversal to the node Time flows through this node flow Data frame Plan the transmission time window. Data frame At the node The transmission time window is determined by the transmission start time and the transfer end time Composition, represented as a data frame At the node The transmission time window on .

[0114] The schematic diagram of planning the transmission time window is as follows: Figure 2 As shown. At the node Multiple data frames ( ) Record the arrival time of each data frame in order of arrival time, and record them as ; Then calculate the flow qualification time of the data frame corresponding to the traffic according to the token bucket corresponding to each data frame, which are recorded as ; Finally, calculate the frame qualification time of each data frame and record it as ; The frame qualification time will be used as the start time of the transmission time window.

[0115] For nodes The first data frame on . For nodes The second data frame above. For nodes On the data frames, which are also nodes The last data frame on .

[0116] For data frame Arrival time. For data frame Arrival time. For data frame Arrival time.

[0117] For flow Flow eligibility time of the corresponding traffic. For data frame Flow eligibility time of the corresponding traffic. For data frame Flow eligibility time of the corresponding traffic.

[0118] For data frame Frame qualification time. For data frame Frame qualification time. For data frame Frame qualification time.

[0119] Step 301, calculating the start time of the transmission time window;

[0120] Data Frame At the node The start time of the transmission on Frame Qualification Time Determine the frame qualification time By the arrival time of the data frame , Flow Qualification Time Combined qualifying time Composition, that is .

[0121] Step 301A, confirming the arrival time of the data frame;

[0122] Case 1, traversed nodes For flow The source node, , then the data frame Arrival time The creation time of the data frame.

[0123] Case 2, traversed nodes Not a flow Assume that the source node In the flow path The previous node on , then the data frame Arrival time The transmission end time of this data frame at the previous hop node .

[0124] Step 301B, calculation of flow qualified time;

[0125] Data frame in the present invention At the node Flow qualified time Obtained by the token bucket algorithm. Data frame At the node Whether to send depends on the bucket As long as there are enough tokens in the bucket, the transmission starts. The length is not sent until the appropriate amount of tokens is added to the data frame. length.

[0126] flow At the node Flow qualified time The calculation method is .

[0127] For flow At the node The corresponding token bucket The last time the bucket was empty.

[0128] For flow If the stream Frame length The maximum number of tokens allowed in the token bucket is not exceeded Normal transmission, if the flow Frame length The maximum number of tokens allowed in the token bucket has been exceeded. The data frame will be discarded.

[0129] Token Bucket The token growth rate per unit time.

[0130] Step 301C, calculation of qualifying time;

[0131] For Node The data frames in the data frame are calculated based on the time when each data frame arrives. Sorting is performed to ensure that data frames arriving earlier are placed at the front of the queue.

[0132] Case 1, data frame If it is the first frame in the queue, the group qualifying time ms.

[0133] Case 2, data frame If it is not the first frame in the queue, the group qualifying time is equal to the transmission end time of the previous frame. Assume that the previous frame in the queue is , then the qualifying time Equal to the previous frame The end time of the transmission .

[0134] Step 301D, obtaining the start time of the transmission time window of the data frame;

[0135] Data Frame Frame qualification time Arrival time of data frame , Flow Qualification Time Combined qualifying time The relationship is . This results in a data frame The start time of the transmission time window ,and .

[0136] Step 302, calculating the end time of the transmission time window;

[0137] Data Frame The end time of the transmission time window By data frame The start time of the transfer time window Heliu At the node Transmission time on ( The calculation method for the end time of the transmission time window is: ,and .

[0138] In the present invention, the data frame At the node The transmission time window on ,and .

[0139] Step 4: Generate a transmission time window for all data frames in the network;

[0140] Complete the traversal according to the method in step 3 For each node in the process, a transmission time window is planned for the data frames passing through the node until the transmission time windows of all nodes passed by all data frames on the path are obtained. Then, the transmission time of the traffic is delimited according to the generated transmission time windows.

[0141] In step 4, after obtaining the transmission time window of the current node of the data frame, the end time of this transmission time window is used as the arrival time of this data frame at the next node on the path. This method can obtain the transmission time window of the data frame from the source node to each node in the destination node. Example

[0142] like Figure 3A typical network topology of time sensitive network is shown, each node in the network performs traffic shaping through TAS. A traffic transmission time window in a super cycle is generated by using a method based on eligible time calculation.

[0143] Table 1 is a case traffic configuration. In this embodiment, a super cycle is 4ms, and the node bandwidth of all nodes is set to 36 Mbps.

[0144] Table 1 sets the network traffic attribute configuration parameters of embodiment 1

[0145]

[0146] Based on the traffic configuration parameters in Table 1, the following will specifically illustrate how to generate a traffic transmission time window in this embodiment according to the method of the present application, which includes the following steps:

[0147] Step (1). According to the routing information of each traffic, search along the path of the traffic 、 、 、 Add the next node in the adjacent nodes on the path to the dependent node set of the previous node , and make the in-degree of the next node ; respectively obtain the dependent node set of each node.

[0148] The , , of node .

[0149] The , , of node .

[0150] The , , of node .

[0151] The , , of node .

[0152] The , , of node .

[0153] The , , of node .

[0154] The of node is empty, .

[0155] node of is empty, .

[0156] Step (2). Initialize two cache queues and ,Will and Join the queue Update the updated queue From the updated queue Remove the node Join the queue Update the tail and get the updated queue and make The in-degree of the two inner nodes , get the updated indegree , ,queue Updated to ,queue Updated to ;

[0157] Step (3). From the queue Remove the node Join the queue tail, and make The in-degree of the two inner nodes , get the updated indegree , , the in-degree is 0 and Join the queue tail, queue Updated to ,queue Updated to ;

[0158] Step (4). Repeat steps (2) and (3) until the last updated queue Contains all nodes, the queue in this embodiment ,Right now Too ;

[0159] Step (5). Take The first node ,exist There are two flows and The number of frames transmitted in one super cycle is 3 frames, which are arranged in the following order: 、 and ;

[0160] Step (6). Calculation Due to the qualified time for The source node, so The arrival time is the creation time of the frame, which is 0ms; exist Flow qualified time , the last empty bucket time ms, so exist Flow qualified time ms; because is the first frame, so exist The qualifying time of the above combination ms; The qualifying time is ms;

[0161] Step (7). Calculation The transmission time window of Frame At the node The start time of the transfer time window ms, the end time of the transmission time window ms;

[0162] Step (8). Calculation Due to the qualified time for The source node, so The arrival time is the creation time of the frame, which is 0ms; Flow qualified time , the last empty bucket time ms, so the flow qualified time ms; because For the second frame, the group format time ms; The qualifying time is ms;

[0163] Step (9). Calculation The transmission time window of Frame At the node The start time of the transfer time window ms, the end time of the transmission time window ms;

[0164] Step (10). Calculation Due to the qualified time for the source node, so the arrival time of the frame is the creation time of the frame, taking 2 ms; the flow eligibility time on the flow eligibility time on the last empty bucket time ms, so the flow eligibility time ms; the group eligibility time of ms; the eligibility time of ms;

[0165] Step (11). The transmission time window planning process of node is shown in Figure 4 , the process is step (6) to step (10), according to the order of the queue , the transmission time window of each node in a super cycle is divided according to the above process;

[0166] Step (12). Calculate the transmission time window of all data frames, get the transmission time window of each flow on each node, as shown in Figure 5 ;

[0167] The topology and traffic used in this embodiment are transmitted window planning, the time used for planning time window, and the comparison chart of the transmission time window of each flow on each node as shown in Figure 6 , compared with ls method, smt_wa method, smt_nw method, ilp_nw method and ilp_nw method in the case of increasing the number of configured flows, the method (referred to as ubtts) still can plan the time window when the number of flows is greater than 160, and the time window planning efficiency of the method is improved by more than 300% compared with the control group method when the number of flows is greater than 60, so the method has higher transmission time window planning efficiency.

[0168] ls method: M. Pahlevan, N. Tabassam, and R. Obermaisser, “Heuristic list scheduler for time triggered traffic in time sensitive networks,” ACM Sigbed Review, vol. 16, no. 1, pp. 15–20, 2019.

[0169] smt_wa method: S. S. Craciunas, R. S. Oliver, M. Chmelik, and W. Steiner, “Scheduling real-time communication in IEEE 802.1 Qbv time sensitive networks,” in Proceedings of the 24th International Conference on Real-Time Networks and Systems, 2016, pp. 183-192.

[0170] smt_nw method with ILP_nw method: F. Durr and N. G. Nayak, “No-wait packet scheduling for IEEE time-sensitive networks (TSN),” in Proceedings of the 24th International Conference on Real-Time Networks and Systems, 2016, pp. 203-212.

[0171] ilp_nw method: E. Schweissguth, P. Danielis, D. Timmermann, H. Parzyjegla, and G. Muhl, “ILP-based joint routing and scheduling for time-triggered networks,” in Proceedings of the 25th International Conference on Real-Time Networks and Systems, 2017, pp. 8-17.

Claims

1. A method for generating a transmission window for calculating TT traffic based on qualified time in a TSN network, characterized in that The steps are as follows: Step 1: Obtain the time-sensitive network topology and traffic information of each node; On the one hand, it is used to obtain a given topological network structure serving the TSN network; On the other hand, it is used to obtain each node in the topological network structure; The third aspect is used to obtain the traffic transmitted in the topology network and the traffic information of each node; Step 2: Sort nodes based on dependency relationships; Sort the nodes according to the routing path obtained in step 1; For any flow, the order of nodes on its routing path is consistent with the order of nodes in the list; Therefore, in the process of traversing the sorted node list node by node and setting the node transmission time window, it can be ensured that the transmission time window of the node on each traffic path strictly follows the order; Step 3: Calculate the transmission time window of each flow at different nodes; Calculate the frame qualification time of the data frame at each node as the start time of its transmission time window; For each data frame corresponding to the traffic flow, the token bucket algorithm is used to calculate the qualified time of the frame, thus achieving preliminary transmission isolation between different frames of the traffic flow; The final calculated combined time for each frame is determined by the maximum of the frame arrival time, the previous frame's qualified time, and the frame qualified time calculated by the token bucket algorithm. The frame eligibility time calculated by the token bucket algorithm is determined by the token recovery rate (CIR), bucket depth (CBS), and frame length. Traffic from different input nodes and with different priority types is calculated using different token buckets. If the frame length does not exceed the bucket depth (CBS), the data frame is transmitted normally. If the frame length exceeds the bucket depth (CBS), transmission is abandoned. Finally, the end time of the transmission time window is calculated based on the data frame length and the node's bandwidth. Step 4: Generate a transmission time window for all data frames in the network; After obtaining the transmission time window of the current node of the data frame, the end time of this transmission time window is used as the arrival time of this data frame at the next node on the path. This method can obtain the transmission time window of the data frame from the source node to the destination node at each node.

2. The method for generating a transmission window for calculating TT traffic based on qualified time in a TSN network according to claim 1, characterized in that Step 1 includes: Step 101, obtaining a time-sensitive network topology and each node; The network topology in the time-sensitive network is denoted as , the network traffic set is recorded as Any flow The traffic characteristic information is recorded as ; Network topology Contains node set information and connected edge set information ; In the network topology structure, the directed connection matrix is ​​used to represent the directed edges. , Assigning a value of 1 means there is a directed edge, and assigning a value of 0 means there is no directed edge; The assignment of to indicate whether there is a directed edge between two nodes; Step 102, traffic marking; flow Traffic characteristics information ,and ; flow Traffic characteristics information ,and ; flow The traffic characteristic information is ,and ; flow Traffic characteristics information ,and ; flow Traffic characteristics information ,and ; Then there is: flow-flow feature set ; Step 103: Node marking; node The node feature information is ,and ; node The node feature information is ; node The node feature information is ; node The node feature information is ; node The node feature information is ; node The node feature information is ; Then: All node feature sets are recorded as ,and ; The load rate of the node is recorded as , in the initial state, the initial load rate of all nodes , the set of dependent nodes Empty, in-degree is 0.

3. The method for generating a transmission window for calculating TT traffic based on qualified time in a TSN network according to claim 1, characterized in that Step 2 includes: Step 201, building dependencies between nodes; The dependency relationship between nodes refers to searching along a certain routing path and finding the next node among the two adjacent nodes on the path. Determined to be dependent on the previous node , recorded as ; At the same time join in The set of dependent nodes In the retrieval process, if the node set Any node in Each time the node is added to the dependent node set of another node, the node Indegree ; According to this principle, all paths are searched and the dependent node sets and in-degrees of all nodes are updated; Step 202, sorting the nodes; The first cache queue is denoted as ; The second cache queue is denoted as ; initialization and ; Set the node All in-degrees The node with value 0 is added to the queue The first update queue is formed in ;from The head takes out a node and stores it in the head node and the The corresponding dependent node set The in-degree of all nodes included ;like There are nodes with in-degree in If it is reduced to 0, the in-degree The node with value 0 is added to the queue The tail of the node; after retrieving the dependent node set After updating the status of each node, the head node From the queue Take out the head and add it to the queue The second update queue is formed in ; then As a queue According to this principle, the operation is repeated until the queue There are no remaining nodes in the queue. It becomes a sorted node list, recorded as ; The node list is represented as ; List The nodes are recorded as , Also known as any node in the node list.

4. The method for generating a transmission window for calculating TT traffic based on qualified time in a TSN network according to claim 1, characterized in that Step three includes: Step 301, calculating the start time of the transmission time window; Data Frame At the node The start time of the transmission on Frame Qualification Time Determine the frame qualification time The arrival time of the data frame , Flow Qualification Time Combined qualifying time Composition, that is ; Step 301A, confirming the arrival time of the data frame; Case 1, traversed nodes For flow The source node, , then the data frame Arrival time The creation time of the data frame; Case 2, traversed nodes Not a flow Assume that the source node In the flow path The previous node on , then the data frame Arrival time The transmission end time of this data frame at the previous hop node ; Step 301B, calculation of flow qualified time; Data Frame At the node Flow qualified time Obtained by the token bucket algorithm; data frame At the node Whether to send depends on the bucket As long as there are enough tokens in the bucket, the transmission begins; if the number of tokens in the bucket is less than the number of data frames The length is not sent until the appropriate amount of tokens is added to the data frame length; flow At the node Flow qualified time The calculation method is ; For flow At the node The corresponding token bucket The last time the bucket was empty; For flow Frame length; If the flow Frame length The maximum number of tokens allowed in the token bucket is not exceeded Normal transmission, if the flow Frame length The maximum number of tokens allowed in the token bucket has been exceeded. Then the data frame will be discarded; Token Bucket The token growth rate per unit time; Step 301C, calculation of qualifying time; For Node The data frames in the data frame are calculated based on the time when each data frame arrives. Sorting is performed to ensure that the data frames arriving earlier are placed at the front of the queue; Case 1, data frame If it is the first frame in the queue, the group qualifying time ms; Case 2, data frame If it is not the first frame in the queue, the group qualification time is equal to the transmission end time of the previous frame; assuming that the previous frame in the queue is , then the qualifying time Equal to the previous frame The end time of the transmission ; Step 301D, obtaining the start time of the transmission time window of the data frame; Data Frame Frame qualification time Arrival time of data frame , Flow Qualification Time Combined qualifying time The relationship is ; Thus we get the data frame The start time of the transmission time window ,and ; Step 302, calculating the end time of the transmission time window; Data Frame The end time of the transmission time window By data frame The start time of the transmission time window Heliu At the node Transmission time on ( ); the calculation method of the transmission time window end time is: ,and ; Data Frame At the node The transmission time window on ,and .

Images