Time-sensitive traffic scheduling method for joint routing in satellite scene

Abstract

The invention relates to the technical field of satellite time-sensitive network communication, in particular to a time-sensitive traffic scheduling method for joint routing in a satellite scene. According to the method, the joint routing algorithm is introduced into the low-orbit satellite network, so that efficient scheduling and path optimization of time-sensitive traffic are realized. In the initial configuration stage, path calculation and time slot resource allocation are carried out by using a particle swarm optimization algorithm so as to maximize the transmission efficiency of the time-sensitive flow. And in the real-time correction stage, a time slot margin and priority adjustment mechanism is dynamically monitored to adapt to frequently changing satellite link conditions, and stable transmission of traffic is ensured. According to the method, the problems of time delay and flow conflict caused by instability of a satellite link are effectively solved, and the service quality and the system performance of a satellite network are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of satellite time-sensitive network communication, and in particular to a time-sensitive traffic scheduling method for joint routing in a satellite scenario. Background Art

[0002] With the development of communication technology and aerospace technology, the network service capabilities of low-orbit satellite (LEO) networking technology are increasing. However, due to the initial research stage, the existing LEO network still needs to be improved in terms of quality of service (QoS) assurance and differentiated service capabilities. In low-orbit satellite networks, how to provide time-sensitive services has always been an issue to be solved. However, satellite networks face many challenges in performing time-sensitive services. On the one hand, the satellite network has a large spatial scale and a wide coverage area, and the inter-satellite link delay of the satellite is usually tens of milliseconds; on the other hand, the inter-satellite link of the satellite network is easily affected by environmental factors such as changes in satellite communication distance, Doppler frequency shift, etc., resulting in unstable inter-satellite link channel quality. It is quite difficult to directly apply existing time-sensitive technology to satellite networks.

[0003] In the existing time-sensitive network (TSN) scheduling mechanism, the circular queue forwarding (CQF) technology transmits data streams through fixed time slots to achieve deterministic low-latency transmission. However, CQF faces problems such as clock synchronization difficulties, fixed time slots that cannot adapt to dynamic link conditions, and delay accumulation in satellite networks. The DetNet working group proposed an improved strategy based on CQF, namely circular specified queue forwarding (CSQF), in combination with segmented routing technology, which to a certain extent solved the clock synchronization and fixed time slot adaptability problems faced by CQF. Its multi-queue structure reduces the dependence on strict clock synchronization and enhances flexibility in the case of unstable links. However, CSQF's adaptability to link status changes is still insufficient, and there is a lack of effective processing mechanism for delayed data frames caused by link instability. Summary of the invention

[0004] Aiming at the lack of delayed data frames caused by unstable link status in the existing CSQF, the present invention provides a time-sensitive traffic scheduling method for low-orbit satellites. First, the centralized controller introduces a joint routing algorithm in the initial configuration stage to achieve efficient scheduling and path optimization of time-sensitive flows. Then, real-time adjustments are made at each forwarding node to address the time slot resource changes and flow conflicts caused by link delay changes, thereby achieving time-sensitive traffic scheduling for low-orbit satellites.

[0005] The time-sensitive traffic scheduling method for low-orbit satellites has the following specific steps:

[0006] Step 1: Build an integrated low-orbit satellite-ground network architecture and implement traffic control between low-orbit satellites and ground networks through a centralized controller;

[0007] The LEO satellite-ground network integrated architecture includes a data plane and a control plane. The data plane is composed of aircraft and LEO satellites, which are connected to each other through wired / wireless channels. The control plane is responsible for managing flow tables and network topology, and monitoring traffic in the ground and LEO networks. The centralized controller implements traffic registration, QoS coordination, and access control. Under this architecture, the network's data plane and control plane coordinate with each other to forward time-sensitive traffic according to established rules.

[0008] Step 2: Introduce the low-orbit satellite timing diagram model, and the centralized controller collects time-sensitive service requests based on the CSQF mechanism;

[0009] The low-orbit satellite timing diagram model divides the system cycle into several discrete sub-cycles {t1, t2, ..., t m}, the topological state within the system cycle is discretized into a set of static topological graphs {G1, G2, ..., G m}.

[0010] The time-sensitive service request specifically includes the source node, destination node, service volume, priority, data transmission cycle, and worst end-to-end delay requirement of the time-sensitive service, which is expressed as:

[0011] f i ={stc i ,des i , size i , priority i , period i ,ddl i , path i ,sid i}

[0012] Among them, src i Indicates the source node of time-sensitive services, des i Indicates the destination node, size i Indicates the size of the business volume, priority i Indicates priority, period i Indicates the data sending cycle, ddl i Worst end-to-end delay requirement, path i Indicates the path, sid i Indicates a time slot.

[0013] Step 3: Based on the LEO satellite timing diagram model, the centralized controller allocates transmission paths and time slot resources for all time-sensitive flows, and establishes a path calculation and time slot allocation model with the optimization goal of maximizing the number of time-sensitive flows burdened by the LEO network;

[0014] The time-sensitive flow is executed repeatedly in each discrete sub-cycle. The lowest common multiple of all time-sensitive flow data sending cycles is set as the super-cycle. The time-sensitive flow scheduling in the system cycle is converted into super-cycle scheduling in each discrete sub-cycle.

[0015] Super period H = LCM {period1, period2, ..., period N}, which is represented by the least common multiple of the N time-sensitive stream data transmission periods.

[0016] The optimization objective is expressed as:

[0017]

[0018] The constraints include path acyclic constraint, slot length constraint, injection slot offset constraint, deadline constraint, and queue resource constraint, which are described as follows:

[0019] A. Path acyclic constraint

[0020] In the process of time-sensitive flow scheduling, loops in the path will occupy more network resources. To ensure that the path is loop-free, it is necessary to meet in Represents the flow f i Does the path pass through [v a , v b ].

[0021] B. Maximum Slot Length Constraint

[0022] The time slot is the minimum time granularity in the entire scheduling process. This means that all time periods in the scheduling process should be divisible by the time slot. That is, the maximum time slot length max(slot len ) = GCD{F.period}, which is the greatest common divisor of the set of time-sensitive flow periods.

[0023] C. Minimum slot length constraint

[0024] Link delay is used as a separate mapping. It is necessary to ensure that all packets in the queue are forwarded within one time slot. In addition, the time slot deviation caused by link delay changes must be controlled within the range of two time slots.

[0025]

[0026] Among them, Q_Len is the queue capacity, MTU is the maximum transmission unit, Bdw is the link bandwidth, and d proc The processing delay of the satellite transponder, Sync clock is the clock synchronization accuracy, d e_prop , d prop_max and prop_minare the mean, maximum and minimum values ​​of the intersatellite link propagation delay, respectively.

[0027] D. Injection Slot Offset Constraints

[0028] The injection slot offset constraint is a constraint on the injection slot selection of the source node. The maximum injection slot offset f of the flow i .slot_max=f i .ddl-(hops+1)×slot len .

[0029] E. Deadline Constraints

[0030] For each flow, all packets must arrive at the destination within a specific deadline.

[0031]

[0032] Among them, d prop (f i ,h) represents the link delay of a single hop, Δ(f i , h) represents the offset time slot of the forwarding node.

[0033] F. Queue resource constraints

[0034] In satellite transponders, queue resources are limited. The entire scheduling process must ensure that the queue resources occupied by time-sensitive flows do not exceed the available queue resources, otherwise, flow scheduling fails.

[0035]

[0036] in, At port p(v a , v b ) The queue resources occupied in time slot Ti, At port p(v a , v b ) in time slot T i The remaining queue resources in.

[0037] Step 4: Solve the path calculation and time slot allocation model based on the particle swarm algorithm to determine the optimal path and allocated time slot resources for each time-sensitive flow;

[0038] The specific process is as follows:

[0039] Step 401: Initialize the particle swarm algorithm parameters: number of particles, maximum number of iterations, inertia weight w and acceleration coefficients c1, C2.

[0040] Step 402: Generate an initial particle swarm, where each particle represents a possible SID and traffic routing solution, and initialize the speed and position of each particle.

[0041] Step 403: For each particle, calculate its fitness value. The fitness value is the number of successfully deployed time-sensitive flows. According to the current fitness value, record the local optimal solution of each particle and the global optimal solution of the entire group.

[0042] Step 404: Update the velocity and position of the particle.

[0043]

[0044] Among them, r1, r2 are random numbers, is the local optimal position of particle i, g best is the global optimal position, represents the position of particle i at time t, that is, the SID configuration and traffic routing scheme at the current time, represents the velocity of particle i at time t.

[0045] Step 405: Calculate a new fitness value according to the new velocity and position of the particle, and update the local optimal solution and the global optimal solution.

[0046] Step 406: Determine the termination condition, that is, stop the iteration when the maximum number of iterations is reached or the change of the global optimal solution is less than a preset threshold, and output the optimal SID configuration and traffic routing solution.

[0047] Step 5: The centralized controller distributes the obtained path information and SID label to each time-sensitive flow and configures them to each node in the network.

[0048] Step 6: After the traffic reaches each forwarding node, each forwarding node uses multiple priority queues to process different types of traffic according to a predetermined queue management mechanism, and performs queue management based on the traffic type and the SID tag it carries.

[0049] The queue management model uses eight queues to meet different service requirements. Q0-Q3 use best-effort forwarding to handle regular data services. Q4 can handle regular data services and can also be used to buffer time-sensitive flows that conflict due to changes in time slot resources caused by changes in link delay. High-priority queues Q5-Q7 use the CSQF mechanism proposed by IETF DetNet to handle time-sensitive flows.

[0050] The specific process of team management is as follows:

[0051] Step 601: When traffic arrives at the forwarding node, determine its traffic type. If the traffic is regular data service, directly enter the queue. If the traffic is time-sensitive flow, execute step 602.

[0052] Step 602: Obtain the actual arrival time t of the time-sensitive flowa , and read its SID tag to obtain its expected arrival time t e .

[0053] Step 603: Compare t a With t e , determine the arrival status of the time-sensitive flow as on-schedule arrival, delayed arrival, or early arrival.

[0054] Step 604: Perform an enqueue operation according to the arrival status of the time-sensitive flow.

[0055] Assume that the queue delay tolerance boundary is t L , the look-ahead tolerance boundary t F , CSQF lead tolerance boundary is t C (t C <t F ), the specific rules include:

[0056] ① The flow arrives on schedule, or the lead time is less than t C , the traffic enters Q5~Q7 according to the SID label. ② The flow arrives late and the delay time is less than t L , or arrives ahead of time and leads by time t + Satisfy t C <t + <t F , the flow enters Q4. ③ The flow arrives ahead of time and the lead time is greater than t F , or the flow arrives late and the delay time is greater than t L , the traffic is dropped.

[0057] Step 7: After the traffic enters the queue, each forwarding node monitors the current time slot margin in real time, dynamically schedules various types of traffic to exit the queue based on the current time slot margin and traffic priority, and completes the time-sensitive flow scheduling of the current time slot.

[0058] The dequeue process mainly includes the following steps:

[0059] Step 701: At the start of a time slot, the forwarding node calculates the current time slot margin.

[0060] Step 702: Count all time-sensitive flows to be sent and sort them according to priority and flow urgency U.

[0061] The calculation method of traffic urgency U is as follows: where f i .ddl is the deadline of the time-sensitive flow, Δt is the time difference between the start time of traffic transmission and the current time, and hop r Indicates the number of remaining hops for traffic.

[0062] Step 703: According to the sorting result of step 702, the time-sensitive flow with the highest sorting is dequeued, the time slot margin is recalculated, and it is determined whether the time slot margin is less than the minimum value min{f} of all the time-sensitive flow traffic to be sent. i .size}, if yes, execute step 704; otherwise, return to execute step 701.

[0063] Step 704: traverse all regular data service flows to be sent, and dequeue the regular data service flows that meet the time slot margin until the time slot margin is exhausted.

[0064] Step 8, update the waiting times of the traffic that failed to leave the queue in the current time slot, and determine whether the waiting times of each traffic is greater than the threshold. If so, adjust the priority of the traffic forward; otherwise, the priority remains unchanged; then the traffic that failed to leave the queue enters Q4, and repeats steps 7 and 8 in the next time slot until all traffic scheduling is completed.

[0065] The advantages of the present invention are:

[0066] (1) The present invention targets the dynamic topology characteristics of low-orbit satellites and, through time slot margin monitoring and priority adjustment mechanisms, can adapt to the frequently changing satellite link conditions and ensure stable transmission of time-sensitive traffic.

[0067] (2) The present invention uses a joint routing and time slot allocation model in the initial configuration phase of the controller, combined with a particle swarm algorithm, to optimize resource allocation to maximize the transmission efficiency of time-sensitive flows and significantly improve the overall performance of the system.

[0068] (3) In the present invention, each forwarding node monitors the time slot margin and the traffic conflict in the network in real time, and timely adjusts the queue order and traffic priority, thereby effectively reducing traffic conflicts and ensuring the delay requirements of time-sensitive flows. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is a schematic diagram of the low-orbit satellite-ground network integrated architecture adopted in the implementation of the present invention;

[0070] Figure 2 It is a schematic diagram of an embodiment of the queue management scheduling method of the present invention;

[0071] Figure 3 is an algorithm flow chart of solving the time slot configuration model of joint routing in the initial configuration phase in an embodiment of the present invention;

[0072] Figure 4 is a flow chart of the enqueueing of a service flow in the real-time adjustment phase in an embodiment of the present invention;

[0073] Figure 5 It is a dequeue flow chart of the service flow in the real-time adjustment stage in an embodiment of the present invention.

[0074] Specific implementation instructions

[0075] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below.

[0076] like Figure 1 As shown, the low-orbit satellite-ground network integrated architecture used in the implementation of the present invention includes a data plane and a control plane. The data plane is composed of an aircraft and a low-orbit satellite, which are interconnected through wired / wireless channels. The control plane is responsible for managing the flow table and network topology, and monitoring the traffic in the ground and LEO networks. The centralized controller implements traffic registration, QoS coordination and access control. Under this architecture, the data plane and control plane of the network coordinate with each other to forward time-sensitive traffic according to established rules.

[0077] like Figure 2 As shown, the queue management model used in the implementation of the present invention uses eight queues to meet different service requirements. Q0-Q3 use best-effort forwarding to process regular data services. Q4 can process regular data services and can also be used to buffer time-sensitive flows that conflict due to changes in time slot resources caused by changes in link delays. High-priority queues Q5-Q7 use the CSQF mechanism proposed by IETF DetNet to process time-sensitive flows.

[0078] The present invention proposes a time-sensitive traffic scheduling method for joint routing in a satellite scenario. The entire scheme is mainly divided into two stages: an initial configuration stage and a real-time correction stage.

[0079] The initial configuration phase is mainly carried out in the centralized controller. The purpose of this phase is to maximize the number of time-sensitive flows in the network and ensure the latency and bandwidth requirements of each time-sensitive flow. At this stage, routing and time slot scheduling schemes are generated for time-sensitive flows based on the CSQF mechanism, and Q4 only handles regular data services. The LEO network can be modeled as a network undirected graph G = (V, E), where V represents the satellite node and E represents the inter-satellite link (ISL). Considering the predictability and periodicity of the satellite orbit, the timing graph model (TGM) is introduced to divide the system cycle into several discrete sub-periods {t1, t2, ..., t m}, the topological state within the system cycle is discretized into a set of static topological graphs {G1, G2, ..., G m}, the process mainly includes the following steps.

[0080] (1) A centralized controller collects time-sensitive service requests;

[0081] The centralized controller is responsible for collecting all time-sensitive flow requests from each node in the network. The time-sensitive service requests specifically include the source node, destination node, service volume, priority, data transmission cycle, and worst end-to-end delay requirements of the time-sensitive service. At the same time, the CSQF embeds the SID information into the content of the data frame, thereby mapping the planning results of the controller to the forwarding plane. That is, the time-sensitive flow in the network can be defined as f i ={stc i ,des i , size i , priority i , period i ,ddl i , path i ,sid i}.

[0082] (2) After collecting time-sensitive service requests, the controller will allocate transmission paths and time slot resources for all time-sensitive flows. Since the LEO network topology within the system cycle has been discretized into several static topology graphs, and the time-sensitive flow is a periodic flow, it will be executed repeatedly in each discrete sub-cycle. Therefore, the lowest common multiple of all time-sensitive flow data transmission periods is set as the super-period, and the time-sensitive flow scheduling within the system cycle is converted into super-period scheduling within each discrete sub-period. Super period H = LCM{period1, period2, ..., period N}, expressed as the least common multiple of the N time-sensitive flow periods.

[0083] The optimization goal is set as: maximizing the number of time-sensitive flows that the LEO satellite network can bear.

[0084]

[0085] Input variables: satellite network spatiotemporal graph TGM, time-sensitive flow set F.

[0086] Decision variables: f for all flows i .path,f i .sid.

[0087] Constraints: mainly include path acyclic constraints, time slot length constraints, injection time slot offset constraints, deadline constraints, and queue resource constraints, which are described in detail as follows.

[0088] A. Path acyclic constraint

[0089] In the process of time-sensitive flow scheduling, loops in the path will occupy more network resources. To ensure that the path is loop-free, it is necessary to meet in Represents the flow f i Does the path pass through [v a , v b ].

[0090] B. Maximum Slot Length Constraint

[0091] The time slot is the minimum time granularity in the entire scheduling process. This means that all time periods in the scheduling process should be divisible by the time slot. That is, the maximum time slot length max(slot len ) = GCD{F.period}, which is the greatest common divisor of the set of time-sensitive flow periods.

[0092] C. Minimum slot length constraint

[0093] Link delay is used as a separate mapping. It is necessary to ensure that all packets in the queue are forwarded within one time slot. In addition, the time slot deviation caused by link delay changes must be controlled within the range of two time slots.

[0094]

[0095] Among them, Q_Len is the queue capacity, MTU is the maximum transmission unit, Bdw is the link bandwidth, and d proc The processing delay of the satellite transponder, Sync clock is the clock synchronization accuracy, d e_prop , d prop_max and prop_min are the mean, maximum and minimum values ​​of the intersatellite link propagation delay, respectively.

[0096] D. Injection Slot Offset Constraints

[0097] The injection slot offset constraint is a constraint on the injection slot selection of the source node. The maximum injection slot offset f of the flow i .slot_max=f i .ddl-(hops+1)×slot len .

[0098] E. Deadline Constraints

[0099] For each flow, all packets must arrive at the destination within a specific deadline.

[0100]

[0101] Among them, d prop (f i ,h) represents the link delay of a single hop, Δ(f i , h) represents the offset time slot of the forwarding node.

[0102] F. Queue resource constraints

[0103] In satellite transponders, queue resources are limited. The entire scheduling process must ensure that the queue resources occupied by time-sensitive flows do not exceed the available queue resources, otherwise, flow scheduling fails.

[0104]

[0105] in, At port p(v a , v b ) The queue resources occupied in time slot Ti, At port p(v a , v b ) in time slot T i The remaining queue resources in.

[0106] (3) The example of the present invention uses a particle swarm algorithm to solve the time slot allocation model of the joint routing. Figure 3 The solution of f in the example of the present invention is given i .path and f i .sid algorithm flow chart, the solution process mainly includes:

[0107] Initial conditions: satellite network spatiotemporal graph TGM, time-sensitive flow set F.

[0108] Step 1: Initialize the algorithm parameters, number of particles, maximum number of iterations, inertia weight w and acceleration coefficients c1, c2.

[0109] Step 2: Generate an initial particle swarm, where each particle represents a possible SID and traffic routing solution. Initialize the speed and position of each particle.

[0110] Step 3: For each particle, calculate its fitness value. The fitness value is the number of time-sensitive flows that are successfully deployed. According to the current fitness value, record the local optimal solution of each particle and the global optimal solution of the entire group.

[0111] Step 4: Update the particle's velocity and position.

[0112]

[0113] Among them, r1, r2 are random numbers, is the local optimal position of particle i, g best is the global optimal position, It represents the position of particle i at time t, that is, the SID configuration and traffic routing scheme at the current time.

[0114] Step 5: Calculate the new fitness value according to the new velocity and position of the particle, and update the local optimal solution and the global optimal solution.

[0115] Step 6: Determine the termination condition, that is, stop the iteration when the maximum number of iterations is reached or the change of the global optimal solution is less than the preset threshold, and output the optimal SID configuration and traffic routing solution.

[0116] (4) Finally, the controller assigns the obtained path information and SID label to each time-sensitive flow and configures them to each node in the network.

[0117] The real-time correction stage is mainly carried out at each forwarding node. This stage aims to deal with the time slot resource changes and flow conflicts caused by link delay changes during network operation. After the traffic arrives at each forwarding node, it is first managed in the queue according to the traffic type and the SID tag it carries. However, in the actual operation process, due to the unstable inter-satellite link delay, the time when the time-sensitive flow arrives at the forwarding node may be advanced or delayed within a certain range. In order to effectively ensure the priority and delay requirements of the traffic, Q4, in addition to processing conventional data services, also absorbs and buffers time-sensitive flows that conflict due to time slot resource changes caused by link delay changes. After the queue is completed, various types of traffic are dynamically scheduled to exit the queue based on time slot margin monitoring and priority adjustment. The overall process is described as follows.

[0118] (1) Business Inflow Team: Figure 4 The enqueuing flow chart of various types of traffic after arriving at the forwarding node in the real-time correction phase is given. The enqueuing process mainly includes the following steps.

[0119] Step 1: When traffic arrives at the forwarding node, first determine its traffic type. If the traffic is regular data service, directly enter the queue. If the traffic is time-sensitive flow, execute step 2.

[0120] Step 2: Get the actual arrival time t of the time-sensitive flow a , and read its SID tag to obtain its expected arrival time t e .

[0121] Step 3: Compare t a With t e , determine the arrival status of the flow, such as on-time arrival, delayed arrival, or early arrival.

[0122] Step 3: According to the arrival status of the flow, perform the queue operation. Assume that the queue delay tolerance boundary is t, and the advance tolerance boundary is t F , CSQF lead tolerance boundary is t C (t C <t F ), the specific rules include: ① The flow arrives on time, or the lead time is less than t C , the traffic enters Q5~Q7 according to the SID label. ② The flow arrives late and the delay time is less than t L , or arrives ahead of time and leads by time t+ Satisfy t C <t + <t F , the flow enters Q4. ③ The flow arrives ahead of time and the lead time is greater than t F , or the flow arrives late and the delay time is greater than t,, the traffic is discarded.

[0123] (2) Business outflow team: Figure 5 A flowchart of dynamically scheduling various types of traffic out of the queue during the real-time correction phase is given. The present invention ensures that high-priority time-sensitive flows have the opportunity to be transmitted first by dynamically adjusting the scheduling order of the traffic, and will not affect the overall time slot resource utilization of the network. The dequeue process mainly includes the following steps.

[0124] Step 1: At the beginning of a time slot, the forwarding node calculates the current time slot margin.

[0125] Step 2: Count all time-sensitive flows to be sent and sort them according to priority and traffic urgency U.

[0126] The calculation method of traffic urgency U is: where f i .ddl is the deadline of the time-sensitive flow, Δt is the time difference from the start time of traffic sending to the current time, and hopr indicates the remaining hops of the traffic.

[0127] Step 3: According to the sorting result of step 2, the time-sensitive flow with the highest sorting is dequeued, and the time slot margin is recalculated to determine whether the time slot margin is less than the minimum value min{f i .size}, if yes, go to step 4; otherwise, go back to step 1.

[0128] Step 4: Traverse all regular data service flows to be sent, and dequeue the regular data service flows that meet the time slot margin until the time slot margin is exhausted.

[0129] Step 5: After the current time slot ends, update the waiting times wait_count of the traffic that failed to leave the queue in the current time slot, and determine whether the waiting times of each traffic is greater than the threshold Max_wait. If so, adjust the priority of the traffic forward; otherwise, the priority remains unchanged; then the traffic that has not left the queue enters Q4, and repeats the above dequeue process in the next time slot until all traffic scheduling is completed.

Claims

1. A time-sensitive traffic scheduling method for joint routing in a satellite scenario, characterized in that: The specific steps are as follows: Step 1: Build an integrated low-orbit satellite-ground network architecture and implement traffic control between low-orbit satellites and ground networks through a centralized controller; The low-orbit satellite-ground network integrated architecture includes a data plane and a control plane. Under this architecture, the network's data plane and control plane coordinate with each other to forward time-sensitive traffic according to established rules. Step 2: Introduce the low-orbit satellite timing diagram model, and the centralized controller collects time-sensitive service requests based on the CSQF mechanism; The time-sensitive service request specifically includes the source node, destination node, service volume, priority, data transmission cycle, and worst end-to-end delay requirement of the time-sensitive service, which is expressed as: f i ={src i ,des i ,size i ,priority i ,period i ,ddl i ,path i ,sid i } Among them, src i Indicates the source node of time-sensitive services, des i Indicates the destination node, size i Indicates the size of the business volume, priority i Indicates priority, period i Indicates the data sending cycle, ddl i Worst end-to-end delay requirement, path i Indicates the path, sid i Indicates time slot; Step 3: Based on the LEO satellite timing diagram model, the centralized controller allocates transmission paths and time slot resources for all time-sensitive flows, and establishes a path calculation and time slot allocation model with the optimization goal of maximizing the number of time-sensitive flows burdened by the LEO network; The time-sensitive flow is executed repeatedly in each discrete sub-cycle. The lowest common multiple of all time-sensitive flow data transmission cycles is set as the super-cycle. The time-sensitive flow scheduling in the system cycle is converted to super-cycle scheduling in each discrete sub-cycle. The optimization objective is expressed as: Step 4: Solve the path calculation and time slot allocation model based on the particle swarm algorithm to determine the optimal path and allocated time slot resources for each time-sensitive flow; Step 5: The centralized controller assigns the obtained path information and SID label to each time-sensitive flow and configures them to each node in the network; Step 6: After the traffic reaches each forwarding node, each forwarding node uses multiple priority queues to process different types of traffic according to a predetermined queue management mechanism, and performs queue management based on the traffic type and the SID tag it carries. The specific process of team management is as follows: Step 601: When traffic arrives at the forwarding node, determine its traffic type. If the traffic is a regular data service, directly enter the queue. If the traffic is a time-sensitive flow, execute step 602. Step 602: Obtain the actual arrival time t of the time-sensitive flow a , and read its SID tag to obtain its expected arrival time t e ; Step 603: Compare t a With t e , determine the arrival status of the time-sensitive flow as on-schedule arrival, delayed arrival, or early arrival; Step 604: perform an enqueue operation according to the arrival status of the time-sensitive flow; Assume that the queue delay tolerance boundary is t L , the look-ahead tolerance boundary t F , CSQF lead tolerance boundary is t C (t C <t F ), the specific rules include: ① The flow arrives on schedule, or the lead time is less than t C , the traffic enters Q5~Q7 according to the SID label; ② The flow arrives delayed and the delay time is less than t L , or arrives ahead of time and ahead of time t + Satisfy t C <t + <t F , the flow enters Q4; ③ The flow arrives ahead of time and the lead time is greater than t F , or the flow arrives late and the delay time is greater than t L , the traffic is discarded; Step 7: After the traffic enters the queue, each forwarding node monitors the current time slot margin in real time, dynamically schedules various types of traffic to exit the queue based on the current time slot margin and traffic priority, and completes the time-sensitive flow scheduling of the current time slot; The dequeue process mainly includes the following steps: Step 701: At the start of a time slot, the forwarding node calculates the current time slot margin; Step 702: Count all time-sensitive flows to be sent and sort them according to priority and flow urgency U; The calculation method of traffic urgency U is as follows: where f i .ddl is the deadline of the time-sensitive flow, Δt is the time difference between the start time of traffic transmission and the current time, and hopr is the number of remaining hops of the traffic; Step 703: According to the sorting result of step 702, the time-sensitive flow with the highest sorting is dequeued, the time slot margin is recalculated, and it is determined whether the time slot margin is less than the minimum value min{f} of all the time-sensitive flow traffic to be sent. i .size}, if yes, execute step 704; otherwise, return to execute step 701; Step 704: traverse all regular data service flows to be sent, and dequeue the regular data service flows that meet the time slot margin until the time slot margin is exhausted; Step 8, update the waiting times of the traffic that failed to leave the queue in the current time slot, and determine whether the waiting times of each traffic is greater than the threshold. If so, adjust the priority of the traffic forward; otherwise, the priority remains unchanged; then the traffic that failed to leave the queue enters Q4, and repeats steps 7 and 8 in the next time slot until all traffic scheduling is completed.

2. According to claim 1, a time-sensitive traffic scheduling method for joint routing in a satellite scenario is characterized in that: The data plane is composed of aircraft and low-orbit satellites, which are interconnected through wired / wireless channels; the control plane is responsible for managing flow tables and network topology, and monitoring traffic in ground and LEO networks. The centralized controller implements traffic registration, QoS coordination and access control.

3. The time-sensitive traffic scheduling method for joint routing in a satellite scenario according to claim 1 is characterized in that: The low-orbit satellite timing diagram model divides the system cycle into several discrete sub-cycles {t1, t2, ..., t m }, the topological state within the system cycle is discretized into a set of static topological graphs {G1, G2, ..., G m }.

4. The time-sensitive traffic scheduling method for joint routing in a satellite scenario according to claim 1 is characterized in that: The super period H = LCM {period1, period2, ..., period N }, which is represented by the least common multiple of the N time-sensitive stream data transmission periods.

5. The time-sensitive traffic scheduling method for joint routing in a satellite scenario according to claim 1 is characterized in that: The constraints of the optimization objective include path acyclic constraint, time slot length constraint, injection time slot offset constraint, deadline constraint and queue resource constraint, which are specifically described as follows: A. Path acyclic constraint In the process of scheduling time-sensitive flows, to ensure that the path is loop-free, in Represents the flow f i Does the path pass through [v a , v b ]; B. Maximum Slot Length Constraint The time slot is the minimum time granularity in the entire scheduling process. The maximum time slot length is max(slot len ) = GCD{F.period}, which is the greatest common divisor of the set of time-sensitive flow periods; C. Minimum slot length constraint Link delay is used as a separate mapping. It is necessary to ensure that all packets in the queue are forwarded within one time slot. In addition, the time slot deviation caused by link delay changes must be controlled within the range of two time slots. Among them, Q_Len is the queue capacity, MTU is the maximum transmission unit, Bdw is the link bandwidth, and d proc The processing delay of the satellite transponder, Sync clock is the clock synchronization accuracy, d e-prop , d prop_max and prop_min are the mean, maximum and minimum values ​​of intersatellite link propagation delay respectively; D. Injection Slot Offset Constraints The injection slot offset constraint is a constraint on the injection slot selection of the source node. The maximum injection slot offset f of the flow is i .slot_max=f i .ddl-(hops+1)×slot len ; E. Deadline Constraints For each flow, all packets must arrive at the destination within a specific deadline; Among them, d prop (f i ,h) represents the link delay of a single hop, Δ(f i , h) represents the offset time slot of the forwarding node; F. Queue resource constraints In the satellite transponder, the entire scheduling process must ensure that the queue resources occupied by time-sensitive flows do not exceed the available queue resources, otherwise, the flow scheduling fails; in, At port p(v a , v b ) The queue resources occupied in time slot Ti, At port p(v a , v b ) in time slot T i The remaining queue resources in.

6. The time-sensitive traffic scheduling method for joint routing in a satellite scenario according to claim 1 is characterized in that: The specific process of solving the path calculation and time slot allocation model based on the particle swarm algorithm is as follows: Step 401: Initialize the particle swarm algorithm parameters: number of particles, maximum number of iterations, inertia weight w and acceleration coefficients c1, c2; Step 402: Generate an initial particle swarm, where each particle represents a possible SID and traffic routing solution; initialize the speed and position of each particle; Step 403: For each particle, calculate its fitness value, which is the number of successfully deployed time-sensitive flows, and record the local optimal solution of each particle and the global optimal solution of the entire group according to the current fitness value; Step 404: Update the velocity and position of the particle; Among them, r1, r2 are random numbers, is the local optimal position of particle i, g best is the global optimal position, represents the position of particle i at time t, that is, the SID configuration and traffic routing scheme at the current time, represents the velocity of particle i at time t; Step 405: Calculate a new fitness value according to the new speed and position of the particle, and update the local optimal solution and the global optimal solution; Step 406: Determine the termination condition, that is, stop the iteration when the maximum number of iterations is reached or the change of the global optimal solution is less than a preset threshold, and output the optimal SID configuration and traffic routing solution.

7. The time-sensitive traffic scheduling method for joint routing in a satellite scenario according to claim 1 is characterized in that: The queue management model uses eight queues to meet different service requirements. Q0-Q3 use best-effort forwarding to process regular data services. Q4 can process regular data services and can also be used to buffer time-sensitive flows that conflict due to changes in time slot resources caused by changes in link delays. High-priority queues Q5-Q7 use the CSQF mechanism proposed by IETF DetNet to process time-sensitive flows.

Images