A scheduling method and system for time-triggered reliable stream transmission

Abstract

This invention belongs to the field of TT stream transmission technology and discloses a scheduling method and system for reliable transmission of time-triggered streams. It utilizes exponential distribution to construct a fault model suitable for transient faults; calculates the minimum number of instance replicas required for each TT stream to tolerate transient faults; improves the original constraints and sets corresponding constraints for CRC-related streams and TT retransmission streams; and uses an improved ILP method to schedule TT streams, CRC-related streams, and TT retransmission streams. This invention achieves reliable transmission of TT streams through time redundancy; a time slice reservation method based on time redundancy, while ensuring data reliability requirements, can reduce the significant network resource waste caused by resource redundancy and improve resource utilization; and provides an effective solution for tolerating transient faults in factory intranets under TSN architecture.

Description

Technical Field

[0001] This invention belongs to the field of time-triggered (TT) stream transmission technology, and particularly relates to a scheduling method and system for reliable transmission of time-triggered streams. Background Technology

[0002] In the architecture of the Industrial Internet, network infrastructure is fundamental, primarily comprising two parts: the industrial control network and the industrial information network. Fieldbus control systems, as a fully distributed, fully digital system architecture, were widely used in the early days of industrial networks. However, due to high costs and incompatibility between different bus technologies, they have gradually been replaced by Industrial Ethernet technology. The biggest advantages of Industrial Ethernet are its low cost, high speed, and good openness. It can also unify the industrial control network and the industrial information network; however, this unification only achieves its purpose by transmitting control data and information data separately through real-time channels and standard channels.

[0003] Time-Sensitive Networking (TSN) is a general standard based on Ethernet technology that guarantees deterministic latency and bandwidth. As a standard, open Layer 2 technology, it also reduces costs and offers better interoperability. Secondly, it can meet the differentiated Quality of Service (QoS) requirements of different levels of service flows and enables the convergence of industrial control networks and industrial information networks, allowing production control data and office management information to be transmitted on the same network. Finally, it supports Software-Defined Networking (SDN), making TSN configuration more flexible and management more convenient, and providing possibilities for intelligent network operation and maintenance. There are three types of traffic in TSN: Time-Triggered (TT) streams, Audio / Video Bridging (AVB) streams, and Best-Effort (BE) streams. TT streams are used for hard real-time applications, requiring deterministic low latency and high reliability guarantees; AVB streams are used for soft real-time applications, requiring minimum end-to-end latency guarantees, but their latency constraints are more relaxed compared to TT streams; BE streams have no rigid latency requirements. To ensure the deterministic low latency and high reliability of TT streams, the IEEE 802.1 working group has developed several standards, including precise network time synchronization mechanisms, traffic shaping mechanisms, resource management mechanisms, and related reliability guarantee mechanisms.

[0004] For industrial systems, network failures can pose serious risks to personnel, equipment, or the environment. Common factors causing network unavailability include user error, packet loss due to network congestion, and network malfunctions. Network failures can be categorized into three types: permanent failures, transient failures, and intermittent failures. Permanent failures are continuous and may be caused by link or switch malfunctions, thus interrupting data transmission. Transient failures occur once and then disappear, and may be caused by electromagnetic interference or random hardware failures; they only affect the current data transmission and have no impact on subsequent data transmissions. Intermittent failures occur periodically and may be caused by a component that is about to fail.

[0005] Based on the above analysis, the problems and shortcomings of existing technologies are as follows: In the TSN standard, the main reliability-related standards are IEEE 802.1Qca and IEEE 802.1CB. They employ spatial redundancy to tolerate three common network faults: permanent faults, transient faults, and intermittent faults. In industrial applications, transient faults are more likely to occur than the other two types. Examples include mobile phones in buildings or vehicles, electrical equipment such as switches and relays, and environmental lightning. If only spatial redundancy is used, in addition to permanent and intermittent faults, extra redundant paths must be provided for each transient fault that needs to be tolerated, increasing system costs. Compared to spatial redundancy, temporal redundancy is more suitable for tolerating transient faults, i.e., transmitting multiple frame copies through the same path. In addition, Cyclic Redundancy Check (CRC) can detect data errors caused by transient faults and then return the detection result to the sender. The sender will retransmit the data that was just sent incorrectly. However, in TSN, since the occurrence of transient faults is random, the time for the sender to retransmit the data is also random. This will bring uncertainty to the deterministic transmission of TSN, and the CRC detection result message may also encounter transient faults during transmission. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention provides a scheduling method and system for reliable transmission of time-triggered streams.

[0007] This invention is implemented as follows: a scheduling method for reliable transmission of time-triggered streams. First, an exponential distribution is used to construct a fault model suitable for transient faults. Then, the TSRTR method is used. In the first part of this method, a formula for calculating the minimum number of replicas required for a TT stream is given, and the advantages and disadvantages of two different calculation methods are compared. In the second part, an improved ILP method is used for scheduling. This mainly involves improving the existing ILP scheduling constraints and adding constraints for Cyclic Redundancy Check (CRC) related streams and TT retransmission streams.

[0008] Furthermore, the scheduling method for reliable transmission of time-triggered streams includes the following steps:

[0009] Step 1: Construct a fault model suitable for transient faults using the exponential distribution;

[0010] Step 2: Calculate the minimum number of instance replicas that need to be transmitted for each TT stream to tolerate transient failures;

[0011] Step 3: Improve the original constraints and formulate corresponding constraints for CRC-related streams and TT retransmission streams. Use the improved ILP method to schedule TT streams, CRC-related streams, and TT retransmission streams.

[0012] Furthermore, in step one, the transient fault encountered by the flow instance is modeled as an exponential distribution, and the flow f is defined as follows: i Instance in link [v a ,v b The probability of uplink failure is:

[0013]

[0014] Where λ represents the expected number of failures per hour. For a TT stream f i Instance in link [v a ,v b ] corresponds to the transmission duration on the exit port, and d is the propagation delay of the link.

[0015] Furthermore, the fault mode defines the behavior of the switch when it encounters a fault; when the receiver correctly receives the data sent by the sender, it will return an acknowledgment message; when the receiver incorrectly receives the data sent by the sender, that is, the sender encountered a momentary fault during the data transmission process, the receiver will return a negative acknowledgment so that the sender can retransmit the data in time; if the receiver performs a CRC check and the returned check result message also encounters a momentary fault, the sender will perform a timeout retransmission.

[0016] Furthermore, for If pfh xIf the following formula is satisfied, then all flows at level x meet the reliability requirements;

[0017] pfh x ≤PFH x

[0018] The same n is used for the flow in each reliability level. i , and different flow rd i 'and F i They are not the same. In order for every flow in level x to meet reliability requirements, rd needs to be used. i 'and F i The maximum value of ' is used to replace the value corresponding to each stream, so that when pfh x ≤PFH x At that time, each flow in level x can meet the reliability requirements.

[0019] Furthermore, regarding TT stream f i Instance transmission time and queue allocation These two variables are subject to the following constraints:

[0020] (1) Frame constraints

[0021] Each instance must complete the transfer within its cycle. have,

[0022]

[0023]

[0024] (2) Delay constraint

[0025] The following formula guarantees that both the TT stream and the TT retransmission stream will arrive at the destination node before their deadline; stream f i The path can be represented as for have:

[0026]

[0027] in, This indicates that the last TT retransmission stream instance copy is on path r. i The transmission time of the last link segment corresponding to the exit port.

[0028] (3) Streaming constraints

[0029] The following formula guarantees the flow f i The order in which instances are transmitted on the link must be determined by the preceding link [v] a ,v xOnly after the transmission on the [v] link is completed can the subsequent link be used for transmission. x ,v b Transmission is performed on [the platform]; for have:

[0030]

[0031] Where n i Indicates flow f i The number of copies that need to be transmitted, d represents the propagation delay of the stream on the link, and δ represents the maximum clock synchronization error.

[0032] (4) Link constraints

[0033] A physical link can only transmit one stream instance at a time. have:

[0034]

[0035]

[0036] in, σ∈{0,1}, M represents an infinite constant; only one of the two constraints in the above equation needs to be satisfied.

[0037] (5) Frame isolation constraint

[0038] Spatial isolation requires different TT stream instances to enter different queues; temporal isolation requires that when different TT stream instances enter the same queue, the next instance can only enter the queue after the previous instance has finished transmitting. have:

[0039]

[0040]

[0041] in, w∈{0,1}, w guarantees that only one of the two constraints in the above equation is satisfied, and M represents an infinitely large constant. The formula used to determine whether two flows enter different queues can be expressed as follows:

[0042]

[0043] Furthermore, CRC-related flow instances are in link [v] a ,v b The transmission times corresponding to the egress ports are as follows:

[0044]

[0045] CRC-related flow constraints:

[0046] for have,

[0047]

[0048]

[0049] Where σ1∈{0,1}, it is used to restrict that only one of the two constraints in the above equation can be satisfied; similarly, CRC-related streams need to be frame isolated to prevent crossover. have:

[0050]

[0051]

[0052] Where w1∈{0,1}, it is guaranteed that one of the two constraints in the above equation is satisfied. The formula used to determine whether two flows enter different queues can be expressed as follows:

[0053]

[0054] Furthermore, TT retransmission instances are on link [v] a ,v b The transmission time of the corresponding export port is represented as follows:

[0055]

[0056] TT retransmission constraints:

[0057] for have:

[0058]

[0059]

[0060] Where σ2∈{0,1}, it is used to restrict that only one of the two constraints in the above equation can be satisfied.

[0061] Furthermore, the objective function defined by the model minimizes the end-to-end latency of each flow, i.e.

[0062]

[0063] In the formula Ω i Indicates TT flow f i End-to-end delay.

[0064] Another object of the present invention is to provide a scheduling system for time-triggered reliable transmission of a scheduling method for time-triggered streams, the system comprising:

[0065] The fault model building module is used to construct fault models suitable for transient faults using the exponential distribution.

[0066] The minimum instance replica count calculation module is used to calculate the minimum number of instance replicas that need to be transmitted for each TT stream to tolerate transient failures.

[0067] The scheduling module is used to schedule TT streams, CRC-related streams, and TT retransmission streams using an improved ILP method.

[0068] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0069] First, this invention addresses the problem of transient faults caused by factors such as electromagnetic interference or random hardware failures by employing time redundancy to achieve reliable transmission of TT streams. The proposed time slice reservation (TSRTR) method based on time redundancy can reduce the waste of network resources caused by resource redundancy and improve resource utilization while ensuring data reliability requirements.

[0070] Extensive experiments and analyses have demonstrated the effectiveness of the algorithm proposed in this chapter, providing an effective solution for tolerating transient failures in factory intranets under the TSN architecture.

[0071] Secondly, to address transient faults caused by factors such as electromagnetic interference or random hardware failures, a time redundancy approach is adopted to achieve reliable TT stream transmission. Specifically, firstly, an exponential distribution is used to construct a fault model suitable for transient faults; then, the TSRTR method is proposed. In the first part of this method, a formula for calculating the minimum number of copies required for TT stream transmission is given, and the advantages and disadvantages of two different calculation methods are compared; in the second part, an improved ILP method is used for scheduling, mainly improving the existing ILP scheduling constraints and adding constraints for Cyclic Redundancy Check (CRC) related flows and TT retransmission flows. The effect is evident in TSN, where many applications cannot tolerate the long latency required for dynamic network control protocols to recover from network faults. Therefore, using dynamic redundancy to deal with network faults is not feasible, making static redundancy, i.e., time redundancy, a viable solution. This invention designs numerous test cases, and experimental results demonstrate that the proposed method is effective, providing an effective solution for reliable TT stream transmission during transient faults in factory intranets under TSN.

[0072] Third, the expected benefits and commercial value of the technical solution of this invention after transformation are as follows:

[0073] This paper proposes a scheduling method for reliable transmission of Time-Triggered Streams (TT streams) during transient faults. This method addresses transient faults caused by factors such as electromagnetic interference or random hardware failures by employing time redundancy to achieve reliable TT stream transmission. The proposed Time Slice Reservation (TSRTR) method based on time redundancy ensures data reliability while reducing the significant network resource waste caused by resource redundancy. This solution provides an effective solution for reliable TT stream transmission during transient faults in factory intranets under a TSN (Time-Triggered Network) architecture. It is expected to provide a better reliable transmission solution for factory intranets using TSN networks, thereby enhancing the competitiveness and profitability of enterprises.

[0074] The technical solution of this invention fills a technological gap in the industry both domestically and internationally:

[0075] Reliable transmission of Time-Triggered (TT) streams is an important research area in Time-Slice Networking (TSN). Current research on TT stream transmission focuses on ensuring data latency requirements through different routing and scheduling methods, with limited research on reliable data transmission. However, for many industrial systems, network failures leading to the loss of critical data can cause serious harm to personnel, equipment, or the environment. In TSN, many applications cannot tolerate the long latency required for dynamic network control protocols to recover from network failures, making dynamic redundancy an impractical approach. Therefore, static redundancy technology becomes a viable solution. This invention proposes a Time Slice Reservation (TSRTR) method based on time redundancy, which, while ensuring data reliability, also reduces the significant network resource waste caused by resource redundancy. Specifically, firstly, an exponential distribution is used to construct a fault model suitable for transient failures. Then, the TSRTR method is proposed. In the first part of this method, a formula for calculating the minimum number of replicas required for TT stream transmission is given, and the advantages and disadvantages of two different calculation methods are compared. In the second part, an improved ILP method is used for scheduling, mainly by improving the existing ILP scheduling constraints and adding constraints for Cyclic Redundancy Check (CRC) related streams and TT retransmission streams. Experimental results show that the proposed method is effective and can provide an effective solution for reliable TT stream transmission when transient failures occur in the factory intranet under TSN.

[0076] The technical solution of this invention solves a technical problem that people have long desired to solve but have been unable to achieve:

[0077] In the field of Transmission-Side Networking (TSN), most current research focuses on ensuring data latency requirements through different routing and scheduling methods, with relatively little research on whether data can be reliably transmitted in the event of network failures. However, for industrial systems, network failures can cause serious harm to people, equipment, or the environment. For example, in an automated manufacturing plant, robots work on a production line. Each robot consists of several terminal systems, such as sensors, controllers, and actuators. To enable them to work together, these robots are connected to the same physical network and a central control unit. If a robot's sensor detects a value that requires all robots to stop immediately, this value is transmitted through the network to the central control unit, which then sends a stop message to each robot. In this scenario, a network failure could cause the transmission of this stop message to fail, resulting in equipment damage and even threatening the lives of workers near the production line. Therefore, solving the problem of reliable transmission during transient failures in TSN is a crucial issue that must be addressed.

[0078] Fourth, the scheduling method for reliable transmission of time-triggered streams provided by this invention has achieved remarkable results in solving the technical problems in the industrial application of existing technologies and has obtained outstanding technical effects.

[0079] First, to address the problem of transient failures, this invention utilizes an exponential distribution to construct a suitable failure model. This step effectively simulates transient failure scenarios that may occur in real-world industrial applications, providing an accurate foundation for subsequent scheduling strategies to predict failures.

[0080] Secondly, by calculating the minimum number of instance replicas required for each time-triggered stream (TT stream), this invention ensures reliable data transmission even in the event of transient failures. This strategy not only improves the reliability of data transmission but also avoids resource waste caused by unnecessary data redundancy.

[0081] Furthermore, this invention improves upon the original constraints and establishes corresponding constraints for CRC-related flows and TT retransmission flows. These improvements and constraints make the scheduling method more aligned with the needs of practical industrial applications, enhancing the flexibility and accuracy of scheduling.

[0082] Finally, by utilizing an improved integer linear programming (ILP) method to schedule the TT stream, CRC-related stream, and TT retransmission stream, this invention achieves reliable transmission scheduling for time-triggered streams. This method not only improves data transmission efficiency but also reduces the risk of transmission errors, providing a more reliable and efficient data transmission solution for industrial applications.

[0083] In summary, this invention represents a significant breakthrough in addressing the technical challenges of reliable transmission of time-triggered streams. It not only improves the reliability and efficiency of data transmission but also provides strong guarantees for the stability and security of industrial applications. Its innovation and practicality have important practical significance and application value for promoting technological progress in related fields. Attached Figure Description

[0084] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0085] Figure 1 This is a flowchart of a scheduling method for time-triggered reliable transmission of streams provided in an embodiment of the present invention;

[0086] Figure 2 This is a structural diagram of a scheduling system for reliable transmission of time-triggered streams provided in an embodiment of the present invention;

[0087] Figure 3 This is a schematic diagram of the TSRTR principle provided in the embodiment of the present invention; (a) no-fault scenario; (b) fault scenario 1; (c) fault scenario 2; (d) TSRTR principle;

[0088] Figure 4 This is a schematic diagram of experimental topology 1 provided in an embodiment of the present invention;

[0089] Figure 5 This is a schematic diagram of experimental topology 2 provided in an embodiment of the present invention;

[0090] Figure 6 This is a schematic diagram of experimental topology 3 provided in an embodiment of the present invention;

[0091] Figure 7 These are experimental topology diagrams illustrating the disadvantages of spatial redundancy tolerating transient faults, provided by embodiments of the present invention; (a) a topology with two redundant paths; (b) a topology with three redundant paths;

[0092] Figure 8 This is a schematic diagram illustrating the number of copies of the TT stream that need to be transmitted in reliability level A provided in this embodiment of the invention;

[0093] Figure 9 This is a schematic diagram illustrating the number of TT stream copies that need to be transmitted in other reliability levels provided in this embodiment of the invention; (a) the number of TT stream copies that need to be transmitted in reliability level B; (b) the number of TT stream copies that need to be transmitted in reliability level C;

[0094] Figure 10 This is a schematic diagram of the PFH of all flows in reliability level A provided in this embodiment of the invention;

[0095] Figure 11 This is a schematic diagram of the PFH of all flows in the reliability level provided in the embodiments of the present invention;

[0096] Figure 12 This is a schematic diagram of the PFH of all flows in the reliability level provided in the embodiments of the present invention;

[0097] Figure 13 The λ=10 provided in the embodiment of the present invention -5 A diagram illustrating the acceptability of TT streams;

[0098] Figure 14 The λ=10 provided in the embodiment of the present invention -3 A diagram illustrating the acceptability of TT streams; Detailed Implementation

[0099] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0100] The following are two industrial application examples based on the method, demonstrating how this scheduling method for time-triggered reliable stream transmission can be applied in real-world scenarios:

[0101] Example 1: Automated production line in the automobile manufacturing industry

[0102] In automated production lines for automobile manufacturing, precise time control and data transmission are crucial for ensuring production efficiency and product quality. This scheduling method can be applied to the control systems of automated production lines to ensure reliable data transmission between various robots and automated equipment during the production process.

[0103] 1. Deployment and scheduling methods:

[0104] Deploying this scheduling method in automated production lines uses an exponentially distributed fault model to predict and respond to potential communication failures.

[0105] Based on the specific requirements of the production line, calculate the minimum number of instance replicas required for each TT flow (e.g., robot control command flow, sensor data flow) and adjust the ILP method for optimized scheduling.

[0106] 2. Effects and Benefits:

[0107] By enabling accurate data retransmission and timely fault handling, production line downtime is significantly reduced, and production efficiency is improved.

[0108] It improves the automation level and reliability of the production line, and reduces the costs caused by human error and downtime.

[0109] Example 2: Real-time monitoring and control of power systems

[0110] In power systems, real-time monitoring and control are crucial for maintaining stable grid operation and responding to emergencies. This dispatching method can be applied to the power system's communication network to ensure the reliable transmission of critical monitoring data and control commands.

[0111] Deploy this scheduling method in the power system's communication network and establish fault models and retransmission mechanisms for critical data streams (such as substation monitoring data streams and remote control command streams).

[0112] By using an improved ILP method to schedule different data streams, a rapid response can be achieved in the event of transient faults, data retransmission can be performed, and the stable operation of the power grid can be guaranteed.

[0113] This method makes data transmission in the power system more reliable and reduces misoperations and delayed responses caused by communication failures.

[0114] It enhances the power grid's ability to respond to emergencies, improves the safety and stability of power grid operation, and plays an important role in preventing major accidents such as large-scale power outages.

[0115] These two examples demonstrate the application potential and practical benefits of a scheduling method for time-triggered reliable streaming in various industrial sectors, and have significant value in improving system reliability and optimizing operational efficiency.

[0116] like Figure 1 As shown, this embodiment of the invention provides a scheduling method for reliable transmission of time-triggered streams, including the following steps:

[0117] Step 1: Construct a fault model suitable for transient faults using the exponential distribution;

[0118] Step 2: Calculate the minimum number of instance replicas that need to be transmitted for each TT stream to tolerate transient failures;

[0119] Step 3: Improve the original constraints and formulate corresponding constraints for CRC-related streams and TT retransmission streams. Use the improved ILP method to schedule TT streams, CRC-related streams, and TT retransmission streams.

[0120] Assume that the terminal systems and TSN switches in the TSN network comply with the IEEE 802.1AS and IEEE 802.1Qbv standards. The entire network adopts a fully centralized configuration mode based on SDN. First, the flow sets joining the network will send a request to the controller. Then, the controller will use the proposed method to calculate and generate a static scheduling table. Finally, the scheduling table will be distributed to the TSN switches and terminal systems.

[0121] The network model is a directed graph G(V,E), where V represents the set of all communication nodes in the network, consisting of terminal systems and TSN switches, i.e., V = ES∪B, es represents a terminal system in the set ES, b represents a TSN switch in the set B, and E represents the set of all links. For any two devices v a and v b The full-duplex physical links between them each correspond to two independent unidirectional links [v] a ,v b ] and [v b ,v a ], each unidirectional link [v a ,v b All of these can be defined using a triplet, namely (c, d, n), where c represents the link's transmission rate, d is the link's propagation delay, and n represents the number of queues at the outgoing port. Switch processing delay is not considered.

[0122] In TSN, a stream refers to time-sensitive data from a source node to one or more destination nodes.

[0123] Define the relevant properties of the flow: Definition The set of all streams in TSN. It can be represented as in and Let them represent the sets of TT streams and non-TT streams, respectively. This represents a set of CRC-related flows. When the receiving end receives an instance, it performs a CRC check and returns the result, either an acknowledgment message or a negative acknowledgment message. This result can then be used... To express. This represents the set of TT streams that need to be retransmitted. Similarly, It can be defined as being in link [v] a ,v b The set of all streams on [the given list]. in and These respectively represent the links [v] a ,v b The set of TT streams, CRC-related streams, TT retransmission streams, and non-TT streams.

[0124] TT flow f i It can be derived from tuples (src) i ,dst i ,T i ,sz i ,dl i ,x i ,r i ) indicates that src i and dst i Let T represent the source node and destination node of the stream, respectively. Let T represent the stream period size and the amount of data sent per period, respectively. i and sz i It means, dl i This indicates the deadline for the flow. Different flows have different reliability requirements, x i Indicates the reliability level. i Indicates flow f i The chosen path is a collection of multiple links. To simplify the model, it is assumed that each stream instance contains only one frame.

[0125] Table 1. Relevant Attributes of Flow

[0126]

[0127] In TSN, TT streams are transmitted periodically, therefore a super-period is defined. Indicates in link [v a ,v b The least common multiple of all TT stream cycles transmitted on the network. The calculation only needs to be performed within one supercycle; the next supercycle will use the previous calculation result. Links in the network [v a ,v b Each port corresponds one-to-one with an exit port, therefore the definition is... For a TT stream f i Instance in link [v a ,v b The corresponding transmission duration on the export port is similar. This indicates the transmission duration of a CRC-related stream instance. Indicates TT flow f i Instance in link [v a ,v b The transmission time at the corresponding output port is relative to the start time of a supercycle. Similarly... Indicates the transmission time of the TT retransmission stream instance. Indicates the transmission time of the CRC-related stream instance. Indicates TT flow f i In relation to link [v] a ,vb Queue allocation on connected output ports. A summary of flow-related attributes is shown in Table 1. Unless otherwise specified, all flows mentioned refer to TT flows.

[0128] To properly design fault tolerance mechanisms, it is essential to first define the fault models and fault modes encountered in the network. The most common type of fault in communication systems is the transient fault. A flow instance encountering a transient fault can typically be modeled as an exponential distribution; therefore, a flow f is defined as follows. i Instance in link [v a ,v b The probability of transmission failure is,

[0129]

[0130] Where λ represents the expected number of failures per hour.

[0131] On the other hand, fault modes define the behavior of the switch when it encounters a fault. When the receiver correctly receives data sent by the sender, it returns an acknowledgment message. When the receiver incorrectly receives data sent by the sender, i.e., the sender encountered a transient fault during data transmission, the receiver returns a negative acknowledgment so that the sender can retransmit the data promptly. If the receiver performs a CRC check and the returned check result message also encounters a transient fault, the sender performs a timeout retransmission. Note that the sender and receiver refer to two TSN devices connected on the same link, and should be distinguished from the concepts of source node and destination node.

[0132] Time slice reservation method based on time redundancy:

[0133] First, make the following assumptions.

[0134] (1) A fault can only affect one instance at a time.

[0135] (2) Repeating the TT stream instance multiple times is sufficient to overcome the impact of transient failures.

[0136] (3) The devices in the network have fault detection capabilities and can detect faults within a specified time, such as using CRC, timers, etc.

[0137] (4) A data retransmission mechanism exists. When the receiving end returns a negative acknowledgment message, the sending end will retransmit the data.

[0138] The pseudocode for the TSRTR method is shown in Algorithm 2. It consists of two parts. The first part calculates the minimum number of instance replicas required for each TT stream to tolerate transient failures. The second part schedules the streams added to the network. In this part, time slices are reserved in advance for the retransmission of TT streams based on the calculation results of the first part. When the receiver correctly receives the data sent by the sender and returns an acknowledgment message, the reserved time slice is used to transmit non-TT streams. If the receiver incorrectly receives the data sent by the sender and returns a negative acknowledgment message, the reserved time slice is used to retransmit TT streams.

[0139] Specifically, the first line uses Dijkstra's algorithm to determine the path for each stream; lines 2 to 5 calculate the minimum number of replicas required for each TT stream, where n i Indicates flow f i The number of copies to be transmitted, N refers to the set of copies transmitted for each stream; Line 6 uses the formula to calculate the bandwidth utilization u, where HP represents the time-over-period, which is the least common multiple of the periods of all TT streams; Lines 7 to 15 use the scheduling algorithm for scheduling. When u>1, the scheduling will definitely fail, so FALSE is returned directly. When u<1, if the scheduling algorithm succeeds, TRUE is returned, otherwise FALSE is returned.

[0140]

[0141]

[0142] like Figure 3 As shown, the principle of the TSRTR method is as follows: Assume there are three flows in total, two TT flows, f1 and f2, and one non-TT flow; the transmission delay of all flows is two time units, and the periods of flows f1 and f2 are 20 and 10 respectively; all flows share the switch's egress port v. a However, in the three different queues, at time 0, all flows arrive from the three different ingress ports of the switch.

[0143] exist Figure 3 In (a), no transient faults occur during the transmission of any stream. In this scenario, the remaining time slices of the TT stream transmission can be used to transmit non-TT streams. However, if a transient fault occurs at t = 1, 3, 7, 11, 13, such as... Figure 3 (b) requires the additional transmission of multiple TT stream instance replicas to tolerate transient failures and improve the reliability of TT stream transmission. In this case, the time allocation of the egress port is as follows: TT stream f1 is transmitted on [6,10), TT stream f2 is transmitted on [0,6) and [10,16), and non-TT streams are transmitted in the remaining time slice. Note Figure 3All transmissions experiencing transient failures are represented by dashed boxes. For a TT stream f1, assume its probability of failure on a certain link is F1 = 0.1; and its reliability requirement is a probability of failure of 0.05 every ten time units. Therefore, if two copies of the TT stream instance are transmitted, F1 = 0.1. 2 =0.01 < 0.05, which meets the reliability requirements. However, it should be noted that transient failures occur randomly. If a transient failure only occurs at t=1, transmitting multiple instance replicas will waste bandwidth and cause network congestion. Figure 3 (c)

[0144] Figure 3 (d) illustrates the principle of the proposed TSRTR method, where the TT flow f1 instance moves along the link [v] at time t1. a ,v b Transmission; at time t2, the last bit of this instance was also transmitted; at time t3, the receiver v b The instance was successfully received; after passing the CRC check, the receiver v b To the sender v a Transmit CRC-related messages and complete the transmission at time t4; sender v a A CRC-related message is received at time t5; the previous operation is then repeated.

[0145] The advantage of the TSRTR method lies in its ability to reserve the required time slices in advance after obtaining the number of transmissions needed for the TT stream. For example, if the sending end v at time t5... a If an acknowledgment message is received, the time slice [t5, t6) can be used to transmit non-TT streams. However, if a negative acknowledgment message is received, or a CRC-related message is not received on time, or an incorrect CRC-related message is received, then the time slice [t5, t6) will be used to retransmit the TT stream instance. Based on the above analysis, the following conclusions can be drawn:

[0146] (1) All TT streams, TT retransmission streams, and CRC-related streams will pass through link [v] during the time interval [t1, t9). a ,v b ] or [v b ,v a Deterministic transmission completed.

[0147] (2) TT stream f1 will flow from v at time t1. a Transmission begins. Since it's uncertain whether a retransmission of the TT stream will occur at time t5, no other streams are allowed to arrive at v during the time interval [t2, t5). a If a flow arrives at v during the time interval [t2, t5). aTherefore, the TT retransmission stream may fail to transmit normally at time t5, thereby increasing the end-to-end latency of the TT stream f1.

[0148] (3) In this example, because stream f1 in TT may be retransmitted, it can only be transmitted from v at least at time t7. b Transmission begins, which of course depends on the number of copies that need to be transmitted for each stream.

[0149] (4) During the time interval [t3, t4), v b Will to v a Transmit CRC-related messages; therefore, to avoid collisions, v... b Cannot send to v a Transmit other streams. The same applies to the time interval [t7,t8).

[0150] (5) If v a If an acknowledgment message is received at time t5, then non-TT streams can be transmitted during the subsequent time interval [t5, t9).

[0151] The core of the TSRTR method is to meet the reliability requirements of each flow by transmitting multiple instance replicas, using a time-slice reservation method for transmission, rather than transmitting multiple instance replicas simultaneously. A crucial step in the TSRTR method is calculating the minimum number of replicas required for a TT flow to tolerate transient failures. The authors proposed two methods for calculating the number of TT stream replicas. In the first method, only the edge switches of the network have frame replication functionality deployed; in the second method, all switches in the network have frame replication functionality deployed. The first method will be referred to as Method A, and the second method as Method B.

[0152] Based on the fault model proposed in the previous section, the following lemma is proposed.

[0153] Lemma 1: For flow f i The instance is in its selected path r i The probability of a failed transmission can be expressed as,

[0154]

[0155] in Indicates flow f i Instance in link [v a ,v b The probability of transmission failure. Indicates flow f i Instance in link [v a ,v b The probability of successful transmission on stream f iInstance in path r i The probability of successful transmission on each link segment is multiplied to obtain the stream f. i The instance is in its selected path r i The probability of successful transmission is calculated by subtracting the stream f from 1. i The instance is in its selected path r i The probability of successful transmission can be used to obtain the stream f. i The probability of instance transmission failure.

[0156] Lemma 2: Flow f i At least n needs to be transmitted i Only a few instance replicas can meet its reliability requirements. Here, the flow f is... i Complete n i The transfer of one instance replica is defined as one round of transfer. Then, for stream f... i The maximum number of rounds that can be transmitted within [0, t] can be expressed as:

[0157]

[0158] Among them Λ i Indicates flow f i To obtain the maximum number of rounds, the end-to-end latency of the instance is calculated by determining the shortest end-to-end latency, i.e., ignoring queuing latency. This can be expressed by the following formula.

[0159]

[0160] Equation (3) can be derived from equation (5).

[0161] t≥(rd′ i -1)*T i +Λ i (5)

[0162] Among them Λ i Indicates flow f i The time required for the first round of instance transfer, (rd′) i -1)*T i This indicates the time required for the subsequent rounds after the first round.

[0163] To describe network reliability, refer to the DO-178B security standard, as shown in Table 2. In Table 2, x represents the reliability level, and PFH represents the probability of a system failure after one hour of operation. Let the set of all flows at reliability level x be denoted as... So when it passes The failure rate of stream computing in PFH is less than that in PFH. x At that time, it can be called The reliability requirements are met. The standard defines five reliability levels, each with different reliability requirements. Levels A, B, and C correspond to TT (Time-To-Time) flows, while levels D and E correspond to non-TT flows.

[0164] Table 2 DO-178B Safety Standards

[0165]

[0166] Lemma 3: At reliability level x, the probability of a network failing per hour can be expressed as,

[0167]

[0168] Where n i Indicates flow f i To tolerate transient failures, the minimum number of copies required for transmission Indicates flow f i n i Each instance copy is located at its selected path r i The probability of transmission failure. Thus, stream f i The probability of a transmission failure within one hour is For each flow in reliability level x The summation is pfh' x Furthermore, the calculations here only consider the TT flows corresponding to the three levels: A, B, and C.

[0169] Lemma 4: Based on the above lemmas, we can deduce pfh' x The upper bound, that is

[0170]

[0171] in

[0172] Theorem: For If pfh x If equation (4.8) is satisfied, then all flows at level x are said to meet the reliability requirements.

[0173] pfh x ≤PFH x (8)

[0174] To simplify the problem, the same n is used for the flow in each reliability level. i , and different flow rd′ i and F′ i They are not the same, so in order for every flow in level x to meet reliability requirements, rd′ needs to be used. i and F′ iThe maximum value is used to replace the value corresponding to each stream, so that when pfh x ≤PFH x At that time, each flow in level x can meet the reliability requirements.

[0175] The main difference between Method B and Method A is whether all switches in the network have deployed frame duplication functionality.

[0176] Lemma 5: Corresponding to Lemma 1 in method A, for flow f i n of the instances i One copy is on path r i The probability of a failed transmission can be expressed as,

[0177]

[0178] Because each switch has frame replication functionality deployed, as long as one instance successfully reaches the switch, the switch will transmit n again. i One instance copy.

[0179] Lemma 6: Corresponding to Lemma 3 in Method A, at reliability level x, the probability of a network failure per hour can be expressed as,

[0180]

[0181] Among them, rd′ i The calculation method is the same as that of method A.

[0182] Lemma 7: Corresponds to Lemma 4 in method A, pfh' x The upper bound can be represented as,

[0183]

[0184] in The final theorem is consistent with method A.

[0185] Regarding the two methods above, Method A utilizes some commercially available off-the-shelf (COTS) switches, significantly reducing network costs. Method B, however, requires each switch in the network to have frame replication functionality deployed, inevitably increasing costs. Furthermore, to meet the same reliability requirements, Method A requires a greater number of transmission instance replicas than Method B. Therefore, using Method A can lead to network congestion, thus affecting the schedulability of TT flows.

[0186] The scheduling method based on ILP addresses the question of whether there is a feasible solution for scheduling TT streams, CRC-related streams, and TT retransmission streams in the context of transient failures. This invention employs the simple and easy-to-implement ILP method. Since this invention aims to schedule not only TT streams but also CRC-related streams and TT retransmission streams, the original constraints are improved, and corresponding constraints are established for CRC-related streams and TT retransmission streams.

[0187] The core of the TSN scheduling problem is to find a feasible scheduling schedule, which can be transformed into a problem involving TT streams f. i Instance transmission time and queue allocation The solution for these two variables is based on the following constraints:

[0188] (1) Frame constraint: Each instance must complete the transmission within its cycle.

[0189] (2) Delay constraint: Ensure that the TT stream and the TT retransmission stream can reach the destination node before their deadline.

[0190] (3) Stream transport constraints: guarantee the flow f i The order in which instances are transmitted on the link must be determined by the preceding link [v] a ,v x Only after the transmission on the [v] link is completed can the subsequent link be used for transmission. x ,v b Transmission is performed on [the platform].

[0191] (4) Link constraint: Only one flow instance can be transmitted on a physical link at a time.

[0192] (5) Frame isolation constraint: In order to prevent different TT stream instances from crossing in the queue, they must be isolated in space or time. On the one hand, spatial isolation requires different TT stream instances to enter different queues; on the other hand, temporal isolation requires that when different TT stream instances enter the same queue, the next instance can only enter the queue after the previous instance has finished transmitting.

[0193] (6) CRC related flow constraint: Whenever the sender transmits an instance to the receiver, the receiver will return a CRC detection message. Based on this detection message, it is determined whether the next reserved time slice is used to retransmit the TT stream or transmit a non-TT stream.

[0194] like Figure 3 As shown, this embodiment of the invention provides a scheduling system for time-triggered reliable transmission of time-triggered streams, which includes a scheduling method for time-triggered reliable transmission of streams.

[0195] The fault model building module is used to construct fault models suitable for transient faults using the exponential distribution.

[0196] The minimum instance replica count calculation module is used to calculate the minimum number of instance replicas that need to be transmitted for each TT stream to tolerate transient failures.

[0197] The scheduling module is used to schedule TT streams, CRC-related streams, and TT retransmission streams using an improved ILP method.

[0198] The effectiveness of the TSRTR method is demonstrated through experiments. First, a simple experiment is used to illustrate the disadvantages of using spatial redundancy to tolerate transient failures. Then, two methods proposed in calculating the number of TT stream transmission replicas are compared. Finally, the TSRTR method of this invention is analyzed from three perspectives: reliability, schedulability, and resource waste rate.

[0199] The experimental simulation was performed using Python on an Intel(R) Core(TM) i5-8250U processor with 1.60GHz and 8GB RAM. In the experiment, three topologies were first generated using the network modeling tool NetworkX, and flows were randomly generated on the Python-based experimental platform. Then, Dijkstra's algorithm was used to generate a shortest path for each flow. Finally, the TSRTR method was used for scheduling. The TSRTR method can be divided into two steps: the first step calculates the number of replicas for the TT flow. Experiments showed that, under the same reliability requirements, method B requires fewer instance replicas than method A. Therefore, the result calculated by method B was used for the second step, where scheduling was performed on the Gurobi solver using the ILP method. The topology used in the experiment is shown in Figure 1. Figure 4 As shown, the topology 2 used in the experiment is as follows: Figure 5 As shown, the topology 3 used in the experiment is as follows: Figure 6 As shown, the number of TSN switches in these three topologies are 6, 8, and 10, respectively; the number of terminal nodes is set to 6; the link transmission rate c is uniformly set to 1Gbps (link bandwidth); and the link propagation delay d is uniformly set to 1μs. Switch processing delay is ignored. For each experiment, one of these three topologies is randomly selected as the experimental topology. The expected number of faults per hour λ in the fault model can be set to 10. -3 Or 10 -5These two values. For the flow parameter settings, the source and destination nodes for each flow are randomly selected from the Elasticsearch (ES), while the flow period set is set to {400μs, 800μs, 1600μs}, the data length set per period is set to {100B, 200B, 400B, 800B, 1500B}, and the reliability level set for each flow is set to {A, B, C}. The period, data length, and reliability level of each flow are randomly selected from the corresponding sets. Additionally, the deadline for each flow is set to its period. The deadline is not explicitly defined; instead, all frames must be transmitted within their respective periods. The effectiveness of the TSRTR method of this invention is independent of the selection of network parameters.

[0200] Comparing the use of spatial redundancy to tolerate transient faults with the use of temporal redundancy in this invention: To illustrate that temporal redundancy is more suitable for tolerating transient faults than spatial redundancy, the topology used is as follows: Figure 7 As shown, Figure 7 (a) shows a topology with two redundant paths, which was used in the first four experiments. Figure 7 (b) illustrates a topology with three redundant paths, the topology used in experiment 5. The port connecting terminal node es1 to path l1 is designated as port 1, and so on for ports 2 and 3. In each experiment, 100 frames are transmitted from es0 to es1 on each path. Then, based on experimental requirements, different packet loss rates are set on the terminal node es0 using the netem module to simulate failures. Finally, the network analysis tool Wireshark is used on the terminal node es1 to capture traffic received from different ports.

[0201] Table 3 shows the number of frames received on each port in each experiment, as well as the number of frames lost in that experiment. Experiment 1 illustrates a fault-free network scenario: terminal node es0 transmits 100 frames on each of two different paths, which es1 receives without error; no frames are lost. Experiment 2 simulates a momentary fault by setting the packet loss rate of the port on es0 connected to the l1 path to 5%, while data transmission on the l2 path is fault-free, thus using two redundant paths to tolerate a momentary fault; no frames are lost. Experiment 3 simulates a permanent fault by setting the packet loss rate of the port on es0 connected to the l1 path to 100%, while data transmission on the l2 path is fault-free, thus using two redundant paths to tolerate a permanent fault; no frame loss is observed. Experiment 4 simulates a momentary fault by setting the packet loss rate of the port on es0 connected to the l1 path to 5%, and a permanent fault by setting the packet loss rate of the port on es0 connected to the l2 path to 100%. In this scenario, 95 frames were received at port 1, but no frames were received at port 2, resulting in a loss of 5 frames. This demonstrates that two redundant paths cannot simultaneously tolerate a permanent fault and a transient fault. Experiment 5 added a path l3 to Experiment 4, and no frames were lost. This shows that additional paths are needed to tolerate simultaneous permanent and transient faults using spatial redundancy. Therefore, spatial redundancy increases network costs, while it is necessary to use it to tolerate permanent faults. Thus, to reduce network costs and improve the efficiency of spatial redundancy in tolerating permanent faults, temporal redundancy is more suitable for tolerating transient faults.

[0202] Table 3 illustrates the experimental results demonstrating the disadvantages of spatial redundancy in tolerating transient faults.

[0203]

[0204] Comparison of two methods for calculating the number of TT stream replicas: Method A and Method B were compared to calculate the number of TT stream replicas under the premise of the same reliability. In this experiment, the expected number of failures λ per hour in the fault model was set to 10. -3 To make the experiment more convincing, the results of this experiment were taken as the average of 50 experiments, and for results that were not integers, a rounding-up method was used.

[0205] Figure 8 This shows the number of TT stream replicas that need to be transmitted, as included in reliability level A. Figure 8The horizontal axis n_l represents the maximum number of links selected by the TT flow in reliability level A; the vertical axis n represents the number of copies that the TT flow needs to transmit. When the number of links is n_l = 1, the number of TT transmission copies calculated by both methods is the same. However, as the number of links increases, the number of copies required for transmission calculated by method B gradually becomes less than that of method A, and this gap widens. In method A, only edge switches have frame replication capabilities. For a given TT flow, the longer the path it chooses, the more likely it is to encounter transient failures. Therefore, the number of copies required to meet its reliability requirements is greater. In method B, all switches in the network have frame replication capabilities. This means that as long as one copy successfully arrives at a switch, the switch can replicate multiple copies at the corresponding egress port. Therefore, for a given TT flow, the longer the path it chooses, the more times the copy is replicated by the switches, the stronger its tolerance for transient failures, and the fewer copies need to be transmitted. Figure 9 The document shows the number of TT stream replicas that need to be transmitted, as included in reliability levels B and C. Figure 9 The trends in the line charts in (a) and (a) are basically the same. Figure 8 Maintain consistency, with Figure 8 The comparison shows that the lower the reliability level, the fewer copies of the TT stream it contains need to be transmitted.

[0206] Furthermore, Method B can be easily upgraded to a dynamic replication scheme, where each switch adjusts the number of replicas according to different situations. Given the advantages of Method B, the experimental results from Method B will be used to complete the following experiments.

[0207] Reliability Analysis: 500 test scenarios were generated to perform reliability analysis on our proposed TSRTR method. Figure 10 The PFH of all flows in reliability level A is shown, where the horizontal axis u represents bandwidth utilization, defined by the following formula, and the vertical axis PFH represents the probability of network failure per hour calculated by formula (6).

[0208]

[0209] Where HP represents the supercycle, note that each result in the graph is an average value over a range. For example, the PFH value when u = 0.6 is the average value of u between [0.55, 0.65]. Figure 10 It can be seen that regardless of whether the expected number of failures λ per hour in the failure model is set to 10... -3 Still 10 -5 The PFH value will increase as u increases. The larger u is, the more TT flows are added to the network, and the PFH value calculated according to equation (6) will also increase. Figure 10 It can also be observed that all flows at reliability level A meet its reliability requirements, meaning the calculated PFH value is less than the PFH value required for level A in the DO-178B security standard. This is reflected in the graph as the blue and yellow lines being below the red line. Furthermore, the larger λ is, the greater the probability of network failures. This is further relevant when solving for the number of TT flow replicas n. i When solving a problem, we always want to minimize λ, so a larger λ will result in a larger PFH value. Figure 8 The yellow line is above the blue line. Figure 11 , Figure 12 The PFH of all flows in reliability levels B and C are shown respectively, and it can be found that all flows in them can also meet their reliability requirements.

[0210] Five hundred test scenarios were generated to perform schedulability analysis on the TSRTR method. The experiments simulated the worst-case scenario: if a TT stream requires three replicas, it will encounter two transient failures. To better illustrate the experimental results, the acceptability rating (AR) is defined here.

[0211]

[0212] Where #sched represents the number of schedulable TT streams, #F TT This represents the total number of TT streams. We compared the AR of several methods.

[0213] (1) NORMAL1: Without considering reliability, only one TT stream instance is transmitted per cycle.

[0214] (2) NORMAL2: Calculate the number of copies n that need to be transmitted for the TT stream using method B. i Then, in each cycle, directly transmit these n... i One copy

[0215] (3)OUR: Using the TSRTR method proposed in this invention

[0216] like Figure 13 and 14 The expected number of failures per hour, λ, is shown separately, with λ set to 10. -5 and 10 -3The acceptability of TT flows under two scenarios. The horizontal axis (u) represents bandwidth utilization, and the vertical axis (AR) represents acceptability. As shown in the graph, the acceptability AR decreases with increasing bandwidth utilization (u) for both methods. The NORMAL2 and OUR methods require multiple additional replicas to tolerate transient failures, resulting in more scheduled instances compared to the NORMAL1 method. Therefore, their acceptability AR is naturally lower than the AR calculated using the NORMAL1 method. Additionally, the OUR method also needs to schedule CRC-related flows, so its acceptability AR is slightly lower than the AR calculated using the NORMAL2 method. (Comparison) Figure 13 and 14 It was found that for larger λ values, the AR calculated using both NORMAL2 and OUR methods would be smaller. This is because the larger λ is, the greater the possibility of transient failures, which would require transmitting more instance replicas to meet reliability requirements.

[0217] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A scheduling method for time-triggered flow reliable transmission, characterized in that, The method comprises the following steps: Step one, using exponential distribution, build a fault model suitable for transient fault; Step two, calculate the minimum instance copy number each TT flow needs to transmit in order to tolerate transient fault; Step three, improve the original constraint condition, and make corresponding constraints for CRC related flow and TT retransmission flow, and use improved ILP method to schedule TT flow, CRC related flow and TT retransmission flow; In step one the flow instance encounters transient failures modeled as an exponential distribution, define flow The instance fails to transmit on a link The probability of a transmission failure is: ， wherein represents the expected number of failures per hour, is a TT flow Instance at link corresponding to the transmission duration on the egress port, d is the propagation delay of the link; The fault mode defines the behavior of the switch when it encounters a fault; When the receiving end correctly receives the data sent by the sending end, an acknowledgement message will be returned; when the receiving end incorrectly receives the data sent by the sending end, i.e. the sending end encounters a transient fault in the process of sending data, the receiving end will return a negative acknowledgement so that the sending end can timely perform data retransmission, and if the returned detection result message of the receiving end after CRC detection also encounters a transient fault, the sending end will perform timeout retransmission; For If All flows in the x level meet the reliability requirement if the following is satisfied: ， The same is used for all streams in each reliability level and are different for different streams The maximum of is used for each stream instead of the value corresponding to each stream When For TT flows Example transmission instants And queue allocation These two variables are constrained as follows: (1) Frame constraint Each instance must complete its transmission within its period, for Yes, ， ， (2) Delay constraint The following formula guarantees that both TT flows and TT retransmission flows can reach the destination node before their deadlines; flow The path of a flow can be represented as For We have: ， wherein, represents the transmission time instant of the last segment link on the path corresponding to the egress port; wherein and represent the set of TT flows and non-TT flows, respectively (3) Flow transmission constraint The following formula guarantees the flow The order in which the instances are transmitted over the links must be such that the predecessor link transmission ends before the successor link transmission can begin; for there are: ， wherein denotes a stream the number of copies needed for transmission, d denotes the propagation delay of the stream over the link, denotes the maximum clock synchronization error; (4) Link constraint A physical link can only transmit one flow instance at a time, and for , there are: ， ， wherein M represents a constant of infinity; only one of the two constraints of the above equation needs to be satisfied; (5) Frame isolation constraint Spatial isolation requires different TT flow instances to enter different queues; temporal isolation requires that when different TT flow instances enter the same queue, the latter instance can enter the queue only after the transmission of the former instance is completed; for There are: ， wherein, , ensuring that only one of the two constraints is satisfied, M representing a constant of infinity, to determine whether two flows enter different queues, which can be expressed by the following formula, ； CRC-related flow instance in the link The transmission time of the corresponding egress port, as follows: ， CRC related flow constraint: For Yes, ， ， wherein, to restrict that only one of the two constraints of the above equation can be satisfied; similarly, CRC-related streams need to be frame-isolated to prevent crossing, for , there are: ， wherein , ensure that the two constraints of the formula meet one, To determine whether the two flows into different queues, can be represented by the following formula: ； The TT retransmission flow instance is in the link The transmission instant of the corresponding egress port is denoted as: ， TT retransmission flow constraint: For , there are: ， ， wherein to restrict that only one of the two constraints of the above equation can be satisfied; The objective function defined by the model minimizes the end-to-end delay of each flow, i.e. ， In the formula represents the end-to-end delay of the TT flow .

2. A scheduling system for time-triggered flow reliable transmission according to claim 1, characterized in that, The system comprises: A fault model construction module, configured to use exponential distribution to build a fault model suitable for transient fault; A minimum instance copy number calculation module, configured to calculate the minimum instance copy number each TT flow needs to transmit in order to tolerate transient fault; A scheduling module, configured to use improved ILP method to schedule TT flow, CRC related flow and TT retransmission flow. ​

Images