A method for scheduling time-sensitive flows with latency-redundant end-to-end cooperation synchronization

Abstract

This invention discloses an end-to-end collaborative delay-redundant time-sensitive flow scheduling method, belonging to the fields of optical networks and industrial internet. Based on a ring queuing and forwarding model, this invention expands the two queues used for transmitting time-triggered flows in the forwarding model to three, one of which is a backup queue. Time-triggered flows entering the backup queue will be transmitted in the next time slot. Priority switching of the time-triggered flow queues in the switch buffer is timed to ensure that data packets at the switch node are transmitted within two cycles. In an optical network scenario, end-to-end collaborative delay-redundant time-sensitive flow scheduling is achieved based on time-sensitive networking. By improving the traditional ring queuing and forwarding model, a dedicated backup queue is set up for time-triggered flows, providing a protected transmission window for them. While ensuring the transmission delay of time-triggered flows, this method maximizes the number of successfully scheduled time-triggered flows, thereby improving the overall throughput of time-triggered flows in the optical network.

Description

Technical Field

[0001] This invention belongs to the field of optical networks and industrial internet, and particularly relates to a traffic queuing method and resource scheduling method between optical network switching nodes. Background Technology

[0002] With the rapid development of information and technology, the Industrial Internet has attracted widespread attention from industry and academia. High-capacity processors, sensors, and actuators connect to these networks, where they can transmit data without human interaction. The Industrial Internet has stringent requirements for reliability and low latency. For these networks, ensuring timely delivery of critical control data and improving the reliability of delivery is a key issue. In recent years, one of the most groundbreaking technologies—Time-Sensitive Networking (TSN)—has offered significant opportunities to address these challenges. TSN is a set of IEEE 802 standards that provides deterministic connectivity over IEEE 802 networks, guaranteeing packet transmission with finite latency, low packet delay variation, and low packet loss. To enable the coexistence of various traffic flows with different priorities on the same network, TSN introduces traffic shaping and scheduling mechanisms. It allows configuring time slices (i.e., reserving transmission resources) to serve different types of traffic based on their arrival and departure times at each intermediate node. The IEEE 802.1 Time-Sensitive Networking Working Group standardized an enhancement to the scheduling of traffic: the addition of a time-aware shaper. The time-aware shaper creates a protected transmission window, guaranteeing bounded latency for time-triggered traffic through a gate-driven mechanism. This mechanism uses gate control lists to control and schedule frames in each output port of the switch. Each gate control list is connected to each output port via eight queues, which contain an ordered set of gate operations. Each operation controls the state of the corresponding queue gate. When the queue gate is open, queued frames can be transmitted. Otherwise, when the queue gate is closed, queued frames cannot be transmitted. To simplify configuration, the IEEE 802.1Qch standard employs a circular queuing and forwarding model. The circular queuing and forwarding model periodically switches queue transmission states to ensure deterministic delayed communication for time-triggered traffic.

[0003] A single switch's resources are insufficient to handle the real-time forwarding of large-scale time-sensitive services; inter-node collaboration can alleviate this limitation. Therefore, many collaborative methods are emerging to improve network reliability and determinism. Regarding reliability, redundant transmission is widely used in many optical networks. Furthermore, the stability of optical fiber—the transmission medium of optical networks—provides a good prerequisite for the implementation of time-sensitive networks, making the application of ring queuing and forwarding models in optical networks of significant value. Summary of the Invention

[0004] To address the issues of low node resource utilization and lack of inter-node cooperation in time-triggered flow forwarding in optical networks, the main objective of this invention is to provide an end-to-end collaborative and synchronous time-redundant time-sensitive flow scheduling method. This method, based on time-sensitive networking, enables end-to-end collaborative time-redundant time-sensitive flow scheduling in optical network scenarios. By improving the traditional cyclic queuing and forwarding model, a dedicated alternative queue is set up for time-triggered flows, providing a protected transmission window for them. This maximizes the number of successfully scheduled time-triggered flows while ensuring transmission latency, thereby improving the overall throughput of time-triggered flows in the optical network.

[0005] The objective of this invention is achieved through the following technical solution.

[0006] This invention discloses a delay-redundant time-sensitive flow scheduling method for end-to-end cooperative synchronization, comprising the following steps:

[0007] Step 1: Based on the ring queuing and forwarding model, the two queues used for transmitting time-triggered streams in the forwarding model are expanded to three, one of which is a backup queue. Time-triggered streams that rush into the backup queue will be transmitted in the next time slot. At the source node, the set of time-triggered streams and the transmission path of the time-triggered streams are confirmed so that the SDN (Software-Defined Network) controller can coordinate the transmission of time-triggered streams.

[0008] Step 1.1: Set up a backup queue for each node switch for time-triggered flow scheduling. Define a simple topology with four nodes. Two time-triggered flows, f1 and f2, arrive at switch SW2 simultaneously. The switch's queue can only buffer two packets per cycle, while f1 and f2 will send two packets per cycle. Due to the limited buffer capacity, some packets from the two time-triggered flows f1 and f2 will be lost, resulting in packet loss. By adding packets to the backup queue, packets that could not be buffered in the main queue in the current cycle are pushed into the backup queue and forwarded in the next cycle, reducing the packet loss rate of the optical network system and improving the throughput of time-triggered flows in the optical network system.

[0009] Step 1.2: The SDN controller collects statistics on the set F of time-triggered flows transmitted across the entire network, calculates the set of routing paths for each time-triggered flow and the available bandwidth of that path, so that the SDN (Software Defined Network) controller can coordinate the transmission of time-triggered flows.

[0010] Step 2: Time-triggered streams often arrive before the latest arrival time slot, while some time-triggered streams fail to arrive before the latest arrival time slot, resulting in packet loss. Time-sensitive scheduling based on end-to-end collaborative synchronization using time-sensitive networking (TSN) comprehensively utilizes the allowed latency redundancy of time-triggered streams. While meeting the latest delivery deadline for time-triggered streams, it sacrifices some latency to increase the number of successfully scheduled time-triggered streams, thereby improving the throughput of time-triggered streams in the optical network.

[0011] Step 2.1: The source node sending the time-triggered stream determines the maximum tolerable delay redundancy f of the time-triggered stream based on its latest arrival time and transmission path. i * .ext:

[0012] f i * .ext = f i * .ddl-|f i * .path| (1)

[0013] Where f i * .ddl represents the stream f i * The latest arrival time and |f i * .path| represents the stream f i * The number of jumps.

[0014] Step 2.2, for flow f i * For each hop on the path, the SDN controller calculates the total bandwidth requirement for all time-triggered flows according to formula (2):

[0015]

[0016] in f represents the total bandwidth of all time-triggered streams in the network. i * .bond represents a time-triggered stream f i * The transmission bandwidth.

[0017] Step 2.3: For this jump, the bandwidth of the time-triggered stream in the standby queue is shown in equation (2):

[0018]

[0019] in f represents the bandwidth requirement of time-triggered streams in the candidate queue.i * .slot(i) represents the stream f i * Whether to buffer this node into the candidate queue, the value is 0 or 1.

[0020] Step 2.4, for the time-triggered stream transmission path [v i v j For each hop, define a weighting factor. Used to determine whether to utilize time slot redundancy in this hop, pushing excess time trigger streams into the candidate queue:

[0021]

[0022] According to formula (4), a set of weight factors W is determined on the transmission path. After obtaining W, the SDN controller triggers the flow f according to the time. i * Maximum delay redundancy f i * The .ext file determines the number of delayed transmission nodes, based on... The size of the time-triggered stream is output, along with an indicator f that determines whether each node should add the time-triggered stream to the candidate queue. i * .slot(i) selects the node with high load and buffer congestion and sends the request to inject the time trigger stream into the candidate queue to each node.

[0023] Step 2.5: Based on steps 2.1 to 2.4, time-sensitive scheduling is performed for end-to-end collaborative synchronization using time-sensitive networking. The allowed latency redundancy of time-triggered streams is utilized in a coordinated manner. Under the premise of meeting the latest deadline for time-triggered stream transmission, some latency of time-triggered streams is sacrificed to increase the number of successfully scheduled time-triggered streams, thereby increasing the throughput of time-triggered streams in the optical network.

[0024] Step 3: Following the time-sensitive scheduling for end-to-end collaborative synchronization based on time-sensitive networking in Step 2, the next cycle begins, and the priority of the node switch buffer queues is changed. In the enhanced circular queuing and forwarding model, there are three queues transmitting time-sensitive streams. In each time slot, two open queues are used to receive data packets. One of the two open queues is a strict time-triggered stream candidate queue, which can only receive data packets and not send any data packets; the other queue can both receive and send data packets; the third queue does not receive data packets, but it can ensure that the data packets in its queue are completely forwarded. The current cycle ends, and the next cycle begins. In one cycle, the third queue is empty because it only forwards data packets and does not receive them, so its priority drops to the lowest. In this cycle, it only receives data and does not forward it, becoming a candidate queue. Meanwhile, the candidate queues that only received data packets in the previous cycle will become crowded and their priority will increase. In this cycle, they will both forward and receive data. The queues that both received and forwarded services in the previous cycle will become the highest priority and will forward all the data in their queues in this cycle. Every three cycles, a priority transition cycle is performed to change the queue priority. This queue priority transition ensures that time-triggered streams are forwarded within two time slots, guarantees the upper limit of the latency of time-triggered streams, and improves the latency performance of the optical network system.

[0025] Based on steps 2 and 3, in the optical network scenario, end-to-end collaborative time-sensitive flow scheduling with latency redundancy is implemented using time-sensitive networking. By improving the traditional circular queuing and forwarding model, a dedicated alternative queue is set up for time-triggered flows, providing a protected transmission window for time-triggered flows. Under the premise of ensuring the transmission latency of time-triggered flows, the number of successfully scheduled time-triggered flows is maximized, thereby improving the throughput of time-triggered flows in the entire optical network.

[0026] Beneficial effects:

[0027] 1. This invention discloses an end-to-end cooperative synchronization delay-redundant time-sensitive flow scheduling method. Based on a cyclic queuing and forwarding model, the two queues used for transmitting time-triggered flows in the forwarding model are expanded to three, one of which is a backup queue. Time-triggered flows that are pushed into the backup queue will be transmitted in the next time slot. That is, based on the enhanced cyclic queuing and forwarding model of the backup queue, data packets that cannot be buffered in the current cycle are pushed into the backup queue and forwarded in the next cycle. Under the premise of meeting the latest arrival deadline of the time-triggered flows, the delay characteristics of some time-triggered flows are sacrificed to increase the number of successfully scheduled time-triggered flows, thereby improving the throughput of the optical network.

[0028] 2. The present invention discloses an end-to-end cooperative synchronization delay redundancy time-sensitive flow scheduling method. By calculating the delay tolerance of time-triggered flows and the bandwidth of nodes, the weight factor of each node is calculated, and then the delay tolerance of time-triggered flows is allocated on the transmission path of time-triggered flows, thereby increasing the number of successfully scheduled time-triggered flows and thus improving the throughput of optical networks.

[0029] 3. The present invention discloses an end-to-end collaborative synchronization delay redundancy time-sensitive flow scheduling method, which realizes the transmission of data packets of switch nodes within two cycles by converting the priority of the time-triggered flow queue in the switch buffer and converting the time-triggered flow buffer queue at regular intervals, thereby ensuring the upper limit of the delay of the time-triggered flow and improving the delay performance of the optical network system.

[0030] 4. The present invention discloses an end-to-end cooperative synchronization delay redundancy time-sensitive flow scheduling method. By improving the traditional ring queuing and forwarding model, a dedicated alternative queue is set up for time-triggered flows, providing a protected transmission window for time-triggered flows. Under the premise of ensuring the transmission delay of time-triggered flows, the number of successfully scheduled time-triggered flows is maximized, thereby improving the throughput of time-triggered flows in the entire optical network. Attached Figure Description

[0031] Figure 1 This is a flowchart of a time-sensitive flow scheduling method with delay redundancy based on end-to-end collaborative synchronization using time-sensitive networks;

[0032] Figure 2 This is a thumbnail of the 60-node network built in the experiment;

[0033] Figure 3 An enhanced circular queuing and forwarding model based on alternative queues. Figure 3 (a) is an illustration of packet loss caused by the traditional circular queuing and forwarding model. Figure 3 (b) is the process of improving the model to correct for packet loss;

[0034] Figure 4 This is a schematic diagram of the priority conversion mechanism of the time-sensitive flow scheduling method;

[0035] Figure 5 , 6 The results show the performance of the proposed method in terms of service timeout rate in simulations. Detailed Implementation

[0036] like Figure 1 As shown in the figure, the specific implementation steps of the time-sensitive flow scheduling method with end-to-end cooperative synchronization and latency redundancy disclosed in this embodiment are as follows:

[0037] Step 1: Based on the cyclic queuing and forwarding model, the two queues used for transmitting time-triggered streams in the forwarding model are expanded to three, one of which is a backup queue. Time-triggered streams that rush into the backup queue will be transmitted in the next time slot. At the source node, the set of time-triggered streams and the transmission paths of the time-triggered streams are confirmed so that the SDN (Software Defined Network) controller can coordinate the transmission of time-triggered streams.

[0038] Step 1.1: Configure a backup queue for each node switch for time-triggered flow scheduling. Build a 60-node network topology, its thumbnail is shown below. Figure 2 The maximum bandwidth of the link channel is 1G. The size of the node switch buffer varies from 400-1200K depending on the importance of the node. Two time-triggered flows, f1 and f2, arrive at switch SW2 simultaneously. Each packet of f1 and f2 is 200-600K in size. Since f1 and f2 send two data packets in each cycle, the switch queue can only buffer two data packets in one cycle. Due to the limited buffer capacity, such as Figure 3 As shown in (a), some data packets in the two time-triggered streams f1 and f2 will inevitably be lost, resulting in packet loss. For example... Figure 3 As shown in (b), by adding packets that could not be cached in the main queue during the current period to the alternative queue, and then forwarding them in the next period, the packet loss rate of the optical network system is reduced and the event-triggered flow throughput of the optical network system is increased.

[0039] Step 1.2: The SDN controller counts the set F of time-triggered flows transmitted across the entire network and calculates the set of routing paths for each time-triggered flow and the available bandwidth of that path.

[0040] Step 2: Time-triggered streams often arrive before the latest arrival time slot, while some time-triggered streams fail to arrive before the latest arrival time slot, resulting in packet loss. Time-sensitive scheduling based on end-to-end collaborative synchronization using time-sensitive networking (TSN) comprehensively utilizes the allowed latency redundancy of time-triggered streams. While meeting the latest delivery deadline for time-triggered streams, it sacrifices some latency to increase the number of successfully scheduled time-triggered streams, thereby improving the throughput of time-triggered streams in the optical network.

[0041] Step 2.1: The source node sending the time-triggered stream determines the maximum tolerable delay redundancy f of the time-triggered stream based on its latest arrival time and transmission path. i * .ext:

[0042] f i * .ext = f i *.ddl-|f i * .path| (1)

[0043] Where f i * .ddl represents the stream f i * The latest arrival time and |f i * .path| represents the stream f i * The number of hops, and the size of each time interval in CQF is 100-1000ms.

[0044] Step 2.2, for flow f i * For each hop on the path, the SDN controller in step 1, as shown in formula (2), calculates the total bandwidth requirement for all time-triggered flows:

[0045]

[0046] in f represents the total bandwidth of all time-triggered streams in the network. i * .bond represents a time-triggered stream f i * The transmission bandwidth.

[0047] Step 2.3: For this jump, the bandwidth of the time-triggered stream in the standby queue is shown in equation (2):

[0048]

[0049] in f represents the bandwidth requirement of time-triggered streams in the candidate queue. i * .slot(i) represents the stream f i * Whether to buffer this node into the candidate queue, the value is 0 or 1.

[0050] Step 2.4, for the time-triggered stream transmission path [v i v j For each hop, define a weighting factor. Used to determine whether to utilize time slot redundancy in this hop, pushing excess time trigger streams into the candidate queue:

[0051]

[0052] According to formula (4), a set of weight factors W is determined on the time-triggered stream transmission path. After obtaining W, the SDN controller determines the weight factor set W based on the time-triggered stream f.i * Maximum delay redundancy f i * The .ext file determines the number of delayed transmission nodes, based on... The size of the time-triggered stream is output, along with an indicator f that determines whether each node should add the time-triggered stream to the candidate queue. i * .slot(i) selects the node with high load and buffer congestion and sends the request to inject the time trigger stream into the candidate queue to each node.

[0053] Step 3: Following the time-sensitive scheduling for end-to-end collaborative synchronization based on time-sensitive networking in Step 2, the next cycle begins, and the priorities of the node switch buffer queues are switched. In the enhanced ring queuing and forwarding model, there are three queues transmitting time-sensitive flows. In each time slot, two open queues are used to receive data packets. One of the two open queues is a strict time-triggered flow candidate queue, which can only receive data packets and not send any data packets; the other queue can both receive and send data packets. The third queue does not receive data packets, but it ensures that the data packets in its queue are completely forwarded.

[0054] Step 3: Following the time-sensitive scheduling for end-to-end collaborative synchronization based on time-sensitive networking in Step 2, the next cycle begins, and the priority of the node switch buffer queues is changed; for example... Figure 4 As shown, in the enhanced cyclic queuing and forwarding model, there are three queues transmitting time-sensitive flows. In each time slot, two open queues are used to receive data packets. One of the two open queues is a strict time-triggered flow candidate queue, which can only receive data packets and not send any data packets. The other queue can both receive and send data packets. The third queue does not receive data packets, but it can ensure that the data packets in its queue are completely forwarded. When the current cycle ends and the next cycle begins, the third queue is empty because it only forwards data packets and its priority drops to the lowest. In this cycle, it only receives and does not forward data packets, becoming a candidate queue. Meanwhile, the candidate queue that only received data packets in the previous cycle will become crowded and its priority will increase. In this cycle, it will both forward and receive data. The queue that both received and forwarded data in the previous cycle will become the highest priority and will forward all the data in its queue in this cycle. Every three cycles, the queue priority is transformed as a priority transformation cycle. Through the transformation of queue priority, it is ensured that the time-triggered flow is forwarded within two time slots, ensuring the upper bound of the latency of the time-triggered flow and improving the latency performance of the optical network system.

[0055] The proposed algorithm was simulated in the 60-node network described in step 1, and the results are as follows: Figure 5 , 6As shown, under the conditions of fixed channel buffering or node switch load, the method outperforms traditional priority-based scheduling algorithms and circular queuing forwarding models in terms of service timeout ratio.

[0056] The above detailed description further illustrates the purpose, technical solution, and beneficial effects of the invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for end-to-end cooperative latency-redundant time-sensitive stream scheduling, the method comprising: Comprising the following steps, Step 1, based on the cyclic queuing and forwarding model, the two queues in the forwarding model for transmitting time-triggered flow are expanded to three, one of which is an alternative queue; the time-triggered flow in the alternative queue will be transmitted in the next time slot; on the source node, the set of time-triggered flows and the transmission path of the time-triggered flow are confirmed for the SDN (Software Defined Network) controller to coordinate the transmission of the time-triggered flow; Step 2, the time-triggered flow often arrives before the latest arrival time slot, while some time-triggered flows cannot arrive before the latest arrival time slot, causing packet loss; based on the time-sensitive network, the time-sensitive scheduling of end-to-end cooperative synchronization is carried out, and the time delay redundancy allowed by the time-triggered flow is utilized to improve the successful scheduling quantity of the time-triggered flow under the premise of meeting the latest deadline of the time-triggered flow transmission, so as to improve the throughput of the optical network time-triggered flow; Step 3, after the time-sensitive scheduling based on the time-sensitive network in step 2, the priority of the node switch buffer queue is converted in the next cycle; In the enhanced cyclic queuing and forwarding model, there are three queues for transmitting time-sensitive flow, and in each time slot, two open queues are used to receive data packets; one of the two open queues is a strict time-triggered flow alternative queue, which can only receive data packets and cannot send any data packets; the other queue can both receive and send data packets; the third queue does not receive data packets, but it can ensure that the data packets in its queue are completely forwarded; at the end of the current cycle, the third queue is empty because it only forwards data packets, so its priority is the lowest, and it only receives data packets in the next cycle and becomes an alternative queue; the alternative queue that only receives data packets in the last cycle will be crowded, and its priority will be high, so it will both forward and receive data packets in the cycle; the queue that both receives and forwards data packets in the last cycle will have the highest priority, and it will forward the data in the queue in the current cycle; every three cycles, a cycle for priority conversion is used to convert the priority of the queue; through the conversion of the priority of the queue, the time-triggered flow is guaranteed to be forwarded within two time slots, the upper bound of the time delay of the time-triggered flow is guaranteed, and the delay performance of the optical network system is improved.

2. The method of claim 1, wherein: The implementation method of step 1 is, Step 1.1, an alternative queue is set for each node switch for time-triggered flow scheduling; a simple topology of four nodes is defined, two time-triggered flows f1 and f2 arrive at switch SW2 at the same time, the queue of the switch can only buffer two data packets in a cycle, while f1 and f2 will send two data packets in each cycle, due to the limited buffer capacity, some data packets of the two time-triggered flows f1 and f2 will be lost, causing packet loss; by adding an alternative queue, the data packets that cannot be buffered in the main queue in the current cycle are flushed into the alternative queue, which will be forwarded in the next cycle, reducing the packet loss rate of the optical network system and improving the throughput of the time-triggered flow of the optical network system; Step 1.2, the SDN controller counts the time-triggered flow set F transmitted by the whole network, calculates the routing path set of each time-triggered flow and the available bandwidth of the path, so as to coordinate the time-triggered flow transmission by the SDN (Software Defined Network) controller.

3. The end-to-end cooperative latency-redundant time-sensitive flow scheduling method of claim 2, wherein: The implementation method of step 2 is, Step 2.1, the source node of the time-triggered stream, determines a maximum latency redundancy that the time-triggered stream can tolerate based on the latest arrival time of the time-triggered stream and the transmission path of the time-triggered stream where f i * .ddl represents the latest arrival time of flow f i * while |f i * .path| represents the number of hops of flow f i * . Step 2.2, for flow f i * At each hop on the path, the SDN controller computes the total bandwidth requirement for all time-triggered flows according to formula (2): wherein represents the total bandwidth of all time-triggered flows in the network, f i * .bond represents the transmission bandwidth of the time-triggered flow f i * . Step 2.3, for the hop, the bandwidth of the time-triggered flow in the backup queue is shown as formula (2): wherein denotes the required bandwidth of the time-triggered flow in the alternative queue, f i * . slot(i) represents the slot of the flow f i * whether the node buffers to the alternative queue, taking the value 0 or 1; Step 2.

4. For each hop of the time triggered flow transmission path [v i , v j ] define a weight factor to determine whether to flush the surplus time triggered flow into an alternative queue at this hop using time slot redundancy. According to formula (4), a set of weight factor sets W on the transmission path is determined; after obtaining W, the SDN controller determines the time triggered flow f i * The maximum time delay redundancy f i * .ext determines the number of delay transmission nodes, according to The size of f i * .slot(i), the requirement that the node with large load and buffer congestion injects the time triggered flow into the alternative queue is issued to each node; Step 2.5, according to steps 2.1 to 2.4, the time-sensitive scheduling of end-to-end cooperative synchronization based on the time-sensitive network is used to comprehensively utilize the delay redundancy allowed by the time-triggered flow, to improve the number of successfully scheduled time-triggered flows by sacrificing part of the time-triggered flow delay on the premise of meeting the latest deadline of the time-triggered flow transmission, so as to improve the throughput of the optical network time-triggered flow.

4. The end-to-end cooperative latency-redundant time-sensitive flow scheduling method of claim 3, wherein: It also includes step 4, based on steps 2 and 3, the delay redundancy time-sensitive flow scheduling of end-to-end cooperation based on the time-sensitive network in the optical network scenario is implemented, a special backup queue is set for the time-triggered flow by improving the traditional cyclic queuing and forwarding model, a protected transmission window is provided for the time-triggered flow, the number of successfully scheduled time-triggered flows is maximized on the premise of guaranteeing the transmission delay of the time-triggered flow, so as to improve the time-triggered flow throughput of the whole optical network.

Images