Low earth orbit satellite constellation inter-satellite link flash prediction and pre-routing handover method and system

Abstract

The low-orbit satellite constellation inter-satellite link flash prediction and pre-routing switching method and system can break through the limitation of existing inter-satellite link passive operation and maintenance, change the link flash failure from "after remediation" to "prevention", and is beneficial to completely solve the business interruption problem caused by flash. The characteristic is that according to the satellite orbit data and the space environment parameters, the multi-dimensional link state perception model and the flash probability prediction algorithm are used to determine the flash risk level of a single inter-satellite link in advance; for high-risk links, the standby transmission route is calculated and locked in advance, the pre-switching of business data is completed before the link flash occurs, and the on-board processing unit and the ground constellation operation and control system are linked to realize the lightweight and autonomous link operation and maintenance and route scheduling, thereby avoiding the business interruption problem caused by after switching.

Description

Technical Field

[0001] This invention relates to the field of low-Earth orbit (LEO) satellite internet technology, and particularly to the fields of satellite constellation operation and maintenance management, inter-satellite link (ISL) status control and constellation routing scheduling. Specifically, it relates to a method and system for predicting inter-satellite link outages and pre-routing switching for LEO satellite constellations. Background Technology

[0002] Currently, low-Earth orbit satellite constellations commonly deploy inter-satellite links to achieve direct data transmission between satellites, building a space backbone network, reducing over-reliance on ground gateway stations, and realizing low-latency data transmission across the entire network. Existing satellite network routing and link operation and maintenance solutions mainly revolve around the following two modes:

[0003] The first type is the passive fault handling mode. The ground control system or on-board processing (OBP) only monitors basic indicators such as the connectivity, signal quality, and transmission delay of the inter-satellite links in real time. When an actual fault occurs, such as a momentary disconnection, packet loss, or interruption, the system triggers a route recalculation and link switching process to find a new available transmission path. This type of solution relies on emergency response after a fault occurs and is a post-fault remedial mechanism.

[0004] The second category is the conventional route maintenance mode, which performs fixed-period route pre-calculation and topology updates based on satellite ephemeris data and orbital trajectories, without making specific predictions for inter-satellite link intermittent interruptions; some solutions only perform static scheduling for link latency and bandwidth, without taking into account multi-dimensional factors such as the physical state of the link itself, space environment interference, and the relative motion rate of satellites to conduct dynamic risk assessment.

[0005] Existing industry-related technical solutions mostly focus on the establishment of inter-satellite links, bandwidth allocation, data transmission protocol optimization, or route reconstruction after a single satellite failure. There is a lack of mature and practical technologies for predicting, risk-classifying, and pre-routing switching schemes based on prediction results for inter-satellite link interruptions (sudden interruptions at the millisecond to second level, non-permanent failures) caused by the high-speed movement of low-orbit satellites.

[0006] Existing low-Earth orbit (LEO) satellite communication systems suffer from four major pain points in link management: First, passive handling leads to high latency and significant service interruption impact. Due to the high speed of LEO satellites (up to 7.5 km / s) and rapid changes in relative position, coupled with space debris interference and signal obstruction, inter-satellite links are prone to momentary interruptions. Current solutions can only initiate rerouting after a failure occurs, resulting in lengthy recalculation and switching times. This leads to a sharp increase in data packet loss and transmission latency, severely impacting real-time services such as voice, video, and industrial IoT, and even causing widespread network topology oscillations. Second, there is a lack of accurate prediction of the causes of interruptions and a high rate of false switching. Existing link monitoring only considers the current state and cannot identify link interruption risks in advance. It may mistake temporary fluctuations for permanent faults, triggering unnecessary routing switches. This not only increases onboard computing power and signaling overhead but also reduces overall network stability. Third, route pre-calculation lacks specificity and adaptability. Conventional route pre-planning only performs static calculations based on orbital positions, without incorporating dynamic information such as link health and intermittent failure probability for real-time adjustments. Pre-generated routes cannot effectively avoid links that are about to fail, and transmission interruptions still occur in actual operation, making it difficult to adapt to the highly dynamic operation and maintenance scenarios of giant low-Earth orbit constellations. Fourth, there is insufficient space-ground coordination and low operation and maintenance efficiency. The ground operation and control center centrally processes all link data, resulting in high computational pressure and high feedback latency. Meanwhile, the satellite has almost no autonomous prediction and local partial switching capabilities, making it unable to effectively cope with the real-time operation and maintenance needs of massive inter-satellite links. Overall operation and maintenance efficiency and system responsiveness are significantly limited. Summary of the Invention

[0007] This invention addresses the shortcomings or defects in existing technologies by providing a method and system for predicting and pre-routing inter-satellite link outages in low-Earth orbit satellite constellations. It overcomes the limitations of passive operation and maintenance of existing inter-satellite links by transforming link outage faults from "post-event remediation" to "pre-event prevention," which helps to completely solve the problem of service interruption caused by outages.

[0008] The technical solution of the present invention is as follows:

[0009] The method for predicting and pre-routing inter-satellite link outages in low-Earth orbit satellite constellations is characterized by using a multi-dimensional link status awareness model and an outage probability prediction algorithm based on satellite orbit data and space environment parameters to pre-determine the outage risk level of a single inter-satellite link; for high-risk links, backup transmission routes are calculated and locked in advance, and pre-switching of service data is completed before the link outage occurs. At the same time, the on-board processing unit and the ground constellation operation and control system are linked to achieve lightweight and autonomous link operation and maintenance and route scheduling, thereby avoiding service interruption problems caused by post-event switching.

[0010] Includes the following steps:

[0011] Step 1: Collect multi-dimensional real-time data, including link quality indicators, satellite operation parameters, and space environment parameters;

[0012] Step 2: Preprocess the multi-dimensional real-time data, including using sliding window filtering to remove abnormal noise data, normalizing the data to the [0,1] interval, and uploading it to the ground operation and control center in fixed cycles;

[0013] Step 3: The ground control center uses a flash failure probability prediction algorithm to calculate the flash failure probability, including inputting the multi-dimensional real-time data into a pre-trained time series prediction model, calculating the flash failure probability within a preset time period, and classifying the link flash failure risk into three levels: low risk, medium risk, and high risk according to a preset threshold.

[0014] Step 4: Pre-generate backup routes. After eliminating high-risk candidate links and calculating multi-target optimized paths for high-risk links in use, at least one backup route is generated in advance before the predicted time of the outage.

[0015] Step 5: Distribute the backup route to the relevant satellite nodes to complete the routing table pre-update;

[0016] Step 6: Within the preset switching time window before the interruption occurs, automatically perform a pre-switching of service traffic from the high-risk link in use to the backup route.

[0017] Step 7: Perform pre-switch status monitoring on the original link in order to perform a back-switch operation or mark it as a permanent fault pending maintenance. The back-switch operation includes switching service traffic back to the original link and releasing backup routing resources.

[0018] Step 7 includes: if the original link has no risk of intermittent failure or does not exceed the low risk level within 10 consecutive seconds after switching, it will automatically switch back to the original link; if the original link experiences an actual intermittent failure and does not recover within 30 seconds, it will be determined as a permanent fault and marked for maintenance.

[0019] The pre-switching trigger time mentioned in step 6 is 1 second before the predicted flashover time. The switching adopts a seamless fast switching mechanism, and the switching time is less than 10 milliseconds.

[0020] Step 5 includes marking the backup route as "standby" in the pre-updated routing table, which does not take effect immediately.

[0021] In step 4, the optimization objectives in the multi-objective path optimization calculation include path hop count, end-to-end delay, and maximum link load of the path. The comprehensive cost is calculated by weighted summation, and one or two paths with the lowest comprehensive cost are selected as backup routes. The weight of the path hop count is 0.3, the weight of the end-to-end delay is 0.5, and the weight of the maximum link load of the path is 0.2.

[0022] The time series prediction model described in step 3 is a two-layer long short-term memory network LSTM. The model parameters are obtained by training with historical ground data before deployment and are incrementally learned periodically using newly generated flash records during operation.

[0023] Step 1 involves collecting multi-dimensional real-time data, including implementing a differentiated reporting strategy using the inter-satellite link transceiver module: low-risk links report summary data every 10 seconds, medium-risk links report complete data every 2 seconds, and high-risk links report high-precision data every 0.5 seconds. The space environment parameters include space debris warning level, solar activity intensity, and geomagnetic index.

[0024] The inter-satellite link interruption prediction and pre-routing switching system for low-Earth orbit satellite constellations is characterized by comprising a combination of the following modules to execute the aforementioned inter-satellite link interruption prediction and pre-routing switching method for low-Earth orbit satellite constellations: an inter-satellite link status perception module, an interruption risk prediction and classification module, a pre-route generation and scheduling module, a pre-switching execution and reversal module, and a satellite-ground collaborative management and control module.

[0025] The technical effects of this invention are as follows: The method and system for predicting and pre-routing inter-satellite link outages in low-Earth orbit satellite constellations can overcome the limitations of existing passive operation and maintenance of inter-satellite links. It is the first to create a proactive operation and maintenance solution with a full process of "advance prediction - risk classification - pre-route generation - pre-switching", which transforms "post-event remediation" into "pre-event prevention", which is conducive to completely solving the problem of service interruption caused by outages. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the inter-satellite link topology of the low-Earth orbit satellite constellation involved in the inter-satellite link inter-disconnection prediction and pre-routing switching method of the low-Earth orbit satellite constellation of the present invention. Figure 1 Satellite nodes are represented by circles and labeled A, B, C, D, E, and F respectively. Solid lines represent normal links. Dashed lines represent high-risk links (about to be interrupted). Thick dotted lines represent pre-generated backup routes (C–A–D). Figure 1 The legend is located in the lower right corner and is used to distinguish different line types and nodes.

[0027] Figure 2 This is a flowchart of the implementation of the method for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations according to the present invention. Figure 2 It contains three lifelines, from left to right: on-board processing unit, ground control center, and inter-satellite link; key time points are marked: T0 (data acquisition), T1 (predictive calculation), T2 (backup route generation and distribution), T3 (pre-switch trigger, 1 second before the flashover), and T4 (predicted flashover point); solid arrows indicate message interaction, and dashed lines indicate time reference lines or the activity itself; Figure 2The legend, located in the upper left corner, illustrates how message interactions and time reference lines are represented.

[0028] Figure 3 This is the core module interaction architecture diagram for implementing the low-Earth orbit satellite constellation inter-satellite link flashover prediction and pre-routing switching system of the present invention. Figure 3 The system is divided into two main areas: the onboard processing unit (left) and the ground operations and control center (right). The onboard processing unit includes an inter-satellite link status awareness module, a pre-switching execution and back-switching module, and a satellite-ground collaborative management and control module (onboard part). The ground operations and control center includes a flashover risk prediction and classification module, a pre-route generation and scheduling module, and a satellite-ground collaborative management and control module (ground part). Solid arrows represent data flow, and dashed arrows represent control flow, clearly showing the interaction between modules. Detailed Implementation

[0029] The following is in conjunction with the attached diagram ( Figures 1-3 The present invention will be described in conjunction with the examples.

[0030] Figure 1 This is a schematic diagram of the inter-satellite link topology of the low-Earth orbit satellite constellation involved in the inter-satellite link inter-disconnection prediction and pre-routing switching method of the low-Earth orbit satellite constellation of the present invention. Figure 2 This is a flowchart of the implementation of the method for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations according to the present invention. Figure 3 This is a diagram illustrating the core module interaction architecture of the low-Earth orbit satellite constellation inter-satellite link intermittent interruption prediction and pre-routing switching system of this invention. (Reference) Figures 1 to 3 As shown, the inter-satellite link flashover prediction and pre-routing switching method for low-Earth orbit satellite constellations uses a multi-dimensional link status awareness model and flashover probability prediction algorithm based on satellite orbit data and space environment parameters to determine the flashover risk level of a single inter-satellite link in advance. For high-risk links, backup transmission routes are calculated and locked in advance, and pre-switching of service data is completed before the link flashover occurs. At the same time, the on-board processing unit and the ground constellation operation and control system are linked to achieve lightweight and autonomous link operation and maintenance and route scheduling, thereby avoiding service interruption problems caused by post-event switching.

[0031] A method for predicting and pre-routing inter-satellite link outages in low-Earth orbit satellite constellations includes the following steps: Step 1, collecting multi-dimensional real-time data, including link quality indicators, satellite operating parameters, and space environment parameters; Step 2, preprocessing the multi-dimensional real-time data, including using sliding window filtering to remove abnormal noise data, normalizing the data to the [0,1] interval, and uploading it to the ground control center in fixed-period packages; Step 3, the ground control center uses an outage probability prediction algorithm to calculate the outage probability, including inputting the multi-dimensional real-time data into a pre-trained time-series prediction model, calculating the outage probability within a preset time period, and classifying the link outage risk according to a preset threshold. The risk levels are categorized into low, medium, and high. Step 4 involves pre-generating backup routes. After eliminating high-risk candidate links and performing multi-target optimized path calculations on the high-risk links in use, at least one backup route is generated in advance before the predicted time of the outage. Step 5 involves distributing the backup routes to relevant satellite nodes to complete the routing table pre-update. Step 6 involves automatically performing a pre-switch of service traffic from the original high-risk links to the backup routes within a preset switching time window before the outage occurs. Step 7 involves monitoring the status of the original links after the pre-switch to enable back-switch operations or marking them as permanently faulty and requiring maintenance. The back-switch operation includes switching service traffic back to the original links and releasing backup route resources.

[0032] Step 7 includes: if the original link has no risk of intermittent failure or does not exceed the low risk level within 10 consecutive seconds after switching, it will automatically switch back to the original link; if the original link experiences an actual intermittent failure and does not recover within 30 seconds, it will be determined as a permanent fault and marked for maintenance.

[0033] The pre-switching trigger time mentioned in step 6 is 1 second before the predicted flashover time. The switching adopts a seamless fast switching mechanism, and the switching time is less than 10 milliseconds.

[0034] Step 5 includes marking the backup route as "standby" in the pre-updated routing table, which does not take effect immediately.

[0035] In step 4, the optimization objectives in the multi-objective path optimization calculation include path hop count, end-to-end delay, and maximum link load of the path. The comprehensive cost is calculated by weighted summation, and one or two paths with the lowest comprehensive cost are selected as backup routes. The weight of the path hop count is 0.3, the weight of the end-to-end delay is 0.5, and the weight of the maximum link load of the path is 0.2.

[0036] The time series prediction model described in step 3 is a two-layer long short-term memory network LSTM. The model parameters are obtained by training with historical ground data before deployment and are incrementally learned periodically using newly generated flash records during operation.

[0037] Step 1 involves collecting multi-dimensional real-time data, including implementing a differentiated reporting strategy using the inter-satellite link transceiver module: low-risk links report summary data every 10 seconds, medium-risk links report complete data every 2 seconds, and high-risk links report high-precision data every 0.5 seconds. The space environment parameters include space debris warning level, solar activity intensity, and geomagnetic index.

[0038] The inter-satellite link inter-satellite disconnection prediction and pre-routing switching system for low-Earth orbit satellite constellations includes a combination of the following modules to execute the aforementioned inter-satellite link inter-satellite disconnection prediction and pre-routing switching method for low-Earth orbit satellite constellations: inter-satellite link status awareness module, inter-satellite disconnection risk prediction and classification module, pre-routing generation and scheduling module, pre-switching execution and reversal module, and satellite-ground collaborative management and control module.

[0039] Addressing the pain points of existing low-Earth orbit (LEO) satellite constellation inter-satellite link operation and maintenance, this invention aims to solve the following core technical problems: achieving accurate early prediction of inter-satellite link intermittent failure risks, distinguishing between momentary interruptions and permanent faults, and reducing the false alarm rate; generating backup routes in advance based on prediction results, completing pre-switching before the actual link interruption, and achieving uninterrupted service transmission; adapting to the high dynamic and high concurrency characteristics of large LEO constellations, achieving on-board autonomous prediction and satellite-ground collaborative scheduling in a lightweight manner, reducing computing power and signaling overhead; and avoiding network oscillations caused by inter-satellite link intermittent failures, ensuring the stability and low latency of satellite internet service transmission.

[0040] This invention proposes a method and system for predicting and pre-routing inter-satellite link outages in low-Earth orbit (LEO) satellite constellations. By constructing a multi-dimensional link status awareness model and an outage probability prediction algorithm, combined with satellite orbit data and space environment parameters, the outage risk level of individual inter-satellite links is determined in advance. For high-risk links, backup transmission routes are calculated and locked in advance, and pre-switching of service data is completed before the link outage occurs. Simultaneously, the onboard processing unit and the ground constellation operation and control system are linked to achieve lightweight and autonomous link operation and maintenance and route scheduling, completely solving the service interruption problem caused by post-outage switching. The components are as follows:

[0041] 1. Inter-satellite link status awareness module: Deployed on the on-board processing unit, it collects core link indicators in real time, including signal transmission and reception power, signal-to-noise ratio (SNR), transmission delay, bit error rate, satellite relative position deviation, and space debris early warning information. It also collects auxiliary parameters such as satellite attitude, battery power, and antenna pointing at the same time, and periodically packages and transmits the data back to the ground operation and control prediction platform.

[0042] 2. Intermittent Disconnection Risk Prediction and Classification Module: Deployed at the ground constellation operation and control center, it integrates on-board backhaul data, satellite ephemeris data, and historical intermittent disconnection cases. Through a lightweight machine learning algorithm, it calculates the probability of link intermittent disconnection within a preset time period (e.g., 5s, 10s) and classifies the risk into three levels: low risk (normal operation), medium risk (fluctuation warning, continuous monitoring), and high risk (imminent intermittent disconnection, pre-switch initiation).

[0043] 3. Pre-route generation and scheduling module: For high-risk links, based on the current constellation topology, remaining available link bandwidth, and satellite load, quickly generate 1-2 optimal backup routes, mark route priorities, and synchronously send backup route information to the corresponding satellite nodes to complete the pre-update of the routing table.

[0044] 4. Pre-switching execution and back-off module: The on-board processing unit receives the pre-switching instruction and, at a preset time point before the actual link failure, smoothly switches the service data to the backup route; if the predicted risk is eliminated, it automatically switches back to the original link; if the link has indeed failed, it maintains the transmission of the backup route and simultaneously marks the faulty link for maintenance.

[0045] 5. Space-Ground Collaborative Management Module: This module enables collaboration between centralized ground-based prediction and autonomous onboard execution. In high-risk scenarios, the onboard system can autonomously perform local switching, while the ground system is responsible for global topology updates, risk assessments, and optimization of prediction model parameters. For example... Figure 3 As shown in the architecture diagram, the on-board processing unit includes a state awareness, pre-switching execution and coordination module (on-board part), while the ground operations control center includes a prediction and classification, route generation and coordination module (ground part). The two interact bidirectionally through data flow and control flow.

[0046] The specific implementation steps are as follows:

[0047] Step 1: Data Acquisition and Preprocessing

[0048] Each satellite node continuously collects real-time link operation indicators and satellite operation parameters via the inter-satellite link transceiver module, at a frequency of once per second. The collected raw data includes:

[0049] 1. Link quality indicators: received signal power (dBm), signal-to-noise ratio (SNR, in dB), bit error rate (BER), and transmission delay (ms).

[0050] 2. Satellite motion parameters: relative three-dimensional position deviation between the local satellite and neighboring satellites (Δx, Δy, Δz, in meters), relative velocity (m / s), and antenna pointing angle deviation (degrees);

[0051] 3. Space environment parameters: Space debris warning level (0-3), solar activity intensity (SFU), geomagnetic index (Kp).

[0052] Preprocessing steps: Sliding window filtering is used to remove abnormal noise points (data points exceeding 3 times the standard deviation), and the data is normalized to the [0,1] interval. The data is then packaged and uploaded to the ground operation and control center at fixed intervals.

[0053] Step 2: Calculation of Intermittent Disconnection Probability (Specific Algorithm Implementation Example)

[0054] The ground prediction module uses an improved two-layer LSTM (Long Short-Term Memory) model to predict the probability of flashover, as implemented below:

[0055] Input feature dimensions: The above 10 core indicators (signal power, SNR, BER, delay, Δx, Δy, Δz, relative velocity, space debris level, geomagnetic index) are selected as the input vector for each time step, and the time window length is 5 seconds (i.e., 5 historical time steps).

[0056] Network structure:

[0057] 1. First LSTM layer: 64 hidden units, output sequence;

[0058] 2. Second LSTM layer: 32 hidden units, outputs the hidden state at the last time step;

[0059] 3. Fully connected layer: 32→16→1, activation function is Sigmoid, output flicker probability (0~1).

[0060] Training data: Historical inter-satellite link flashover records (at least 100,000 samples) were used, with a ratio of positive samples (data segments within 5 seconds before the flashover) to negative samples (normal link data segments) of 1:3.

[0061] Predicted output: The model outputs the probability P_flash of the link flashing out within the next 5 seconds.

[0062] Risk classification:

[0063] 1. P_flash ≥ 0.8 → High risk (imminent flash failure, immediately initiate pre-switch);

[0064] 2. 0.5 ≤ P_flash < 0.8 → Medium risk (fluctuation warning, continuous monitoring, no switching for now);

[0065] 3. P_flash < 0.5 → Low risk (normal operation).

[0066] Computing power consumption assessment: A single prediction (5-second window) takes about 8ms of CPU time (200MHz ARM core) and about 1.2MB of memory (including model parameters and intermediate cache) in the on-board simulation environment, which is far below the upper limit of available computing power of the on-board processing unit (usually 50MHz·s and 4MB of memory are reserved), which meets the "lightweight" requirements.

[0067] Step 3: Pre-generate backup routes

[0068] When a link is determined to be high-risk, the system automatically performs the following operations:

[0069] Candidate link exclusion: High-risk links are marked as unavailable, while links that are faulty or have a load exceeding 80% are excluded;

[0070] Multi-objective optimized path calculation: Based on the current constellation topology snapshot, an improved Dijkstra algorithm is adopted, which comprehensively considers three cost functions: path hop count (weight 0.3), end-to-end delay (weight 0.5), and maximum link load of the path (weight 0.2), to generate 1-2 backup routes with the minimum comprehensive cost;

[0071] Route pre-deployment: Backup route information (including source satellite, relay satellite list, and target satellite label switching path) is deployed to relevant satellite nodes via satellite-to-ground link. Each node updates its local routing table in advance (marked as "standby" and not effective immediately).

[0072] Step 4: Pre-switch trigger execution

[0073] Based on the expected flashover time output by the flashover prediction model (the model can also output the remaining seconds before the flashover occurs, obtained through time distribution regression), a pre-switching is triggered 1 second before the predicted flashover point:

[0074] 1. The onboard processing unit sends a "switching command" to the source satellite, switching the next-hop outgoing interface of the service traffic from the original high-risk link to the first hop of the backup route;

[0075] 2. Seamless fast switching mechanism: First, a forwarding entry for the backup path is established, and then the primary table entry is updated atomically. The switching time is less than 10ms, and the business is unaware of it.

[0076] 3. After a successful switchover, the source satellite sends a confirmation message to the ground control center while continuously monitoring the status of the original link.

[0077] like Figure 2 In the timing sequence shown, time T3 corresponds to the pre-switching triggered 1 second before the flashover, time T4 is the predicted flashover point, and after the switch is completed, the satellite transmits confirmation back to the ground.

[0078] Step 5: Condition Monitoring and Switchback / Fault Marking

[0079] 1. If the probability of intermittent disconnection of the original link drops below 0.3 within 10 consecutive seconds after the switchover, and there is no actual packet loss, the switchback will be automatically triggered, and the service traffic will be switched back to the original link, releasing the backup routing resources.

[0080] 2. If the original link actually experiences a brief interruption after prediction (loss of 3 consecutive heartbeat packets) and does not recover within 30 seconds, it is determined to be a permanent fault, the link is added to the maintenance list, and the ground operation and control center arranges manual or automated troubleshooting.

[0081] 3. If the original link is briefly interrupted and then restored (<500ms), the backup route will remain in place. Once the original link is restored, it will be marked as "to be observed" and will not be immediately switched back to avoid ping-pong switching.

[0082] To reduce the bandwidth consumption of the satellite-to-ground link by high-frequency data reporting, a differentiated reporting strategy is adopted:

[0083] 1. Low-risk link: Report summary data (key indicators only) every 10 seconds;

[0084] 2. Medium-risk link: Report complete data every 2 seconds;

[0085] 3. High-risk link: High-precision data is reported every 0.5 seconds, and the satellite autonomously triggers pre-switching, while the ground only receives status feedback.

[0086] Simulation calculations show that the average uplink bandwidth of a single satellite is less than 8kbps, and the total uplink bandwidth of 10,000 satellites is less than 80Mbps, which is far below the typical satellite-to-ground link capacity (usually >100Mbps) and will not cause congestion.

[0087] Technical Results: Based on a simulation environment of a typical low-Earth orbit constellation (orbital altitude 550km, 4 inter-satellite links / satellite, totaling 1584 satellites), the traditional passive routing scheme (OSPF + fault-triggered rerouting) and the scheme of this invention are compared:

[0088]

[0089] 1. A method for predicting inter-satellite link outages and pre-routing handover in a low-Earth orbit satellite constellation, comprising the following steps:

[0090] Collect multi-dimensional real-time data of inter-satellite links, including link quality indicators, satellite motion parameters, and space environment parameters;

[0091] The multi-dimensional real-time data is input into a pre-trained time series prediction model to calculate the probability of a flashover within a preset time period in the future, and the link risk is divided into three levels: low risk, medium risk, and high risk according to a preset threshold.

[0092] For high-risk links, at least one backup route is generated in advance before the predicted time of the outage and distributed to the relevant satellite nodes to complete the pre-update of the routing table;

[0093] Within a preset switching time window before the interruption occurs, the pre-switching of service traffic from the original link to the backup route is automatically performed.

[0094] 2. The time-series prediction model is a two-layer long short-term memory network (LSTM). The input features include 10 dimensions of data, such as signal received power, signal-to-noise ratio, bit error rate, transmission delay, satellite relative three-dimensional position deviation, relative velocity, space debris warning level, and geomagnetic index. The time window length is 5 seconds, and the output is the probability of a flashover occurring within the next 5 seconds.

[0095] 3. The thresholds for risk classification are as follows: a disconnection probability ≥ 0.8 is considered high risk, a disconnection probability ≤ 0.5 < 0.8 is considered medium risk, and a disconnection probability < 0.5 is considered low risk.

[0096] 4. The backup route is generated using a multi-objective optimization algorithm. The optimization objectives include path hop count, end-to-end delay, and maximum link load of the path. The comprehensive cost is calculated by weighted summation, and 1-2 paths with the minimum comprehensive cost are selected as backup routes.

[0097] 5. The pre-switching trigger timing is 1 second before the predicted flashover time. The switching adopts a seamless fast switching mechanism, and the switching time is less than 10 milliseconds.

[0098] 6. It also includes status monitoring and back-switch steps after pre-switch: If the probability of a brief interruption of the original link drops below 0.3 within 10 consecutive seconds after the switch, it will automatically switch back to the original link; if the original link experiences an actual brief interruption and does not recover within 30 seconds, it will be judged as a permanent fault and marked for maintenance.

[0099] 7. The data collection adopts a differentiated reporting strategy: low-risk links report summary data every 10 seconds, medium-risk links report complete data every 2 seconds, and high-risk links report high-precision data every 0.5 seconds.

[0100] 8. A system for predicting and pre-routing inter-satellite link outages in low-Earth orbit satellite constellations, comprising:

[0101] The inter-satellite link status awareness module, deployed in the on-board processing unit, is used to collect multi-dimensional real-time data.

[0102] The intermittent fault risk prediction and classification module is deployed in the ground constellation operation and control center and is used to perform the intermittent fault probability calculation and risk classification as described in any one of claims 1-3;

[0103] The pre-route generation and scheduling module is used to perform the backup route generation and distribution as described in claim 1 or 4;

[0104] The pre-switching execution and back-switch module is deployed on the on-board processing unit and is used to execute the pre-switching triggering and back-switch operations as described in claims 1, 5, and 6.

[0105] The satellite-ground collaborative management module is used to achieve coordination between ground prediction and on-board execution, and to optimize reporting strategies.

[0106] 9. The fault risk prediction and classification module adopts an improved two-layer LSTM model. The model parameters are obtained by training with historical ground data before deployment, and incremental learning is performed periodically using newly generated fault records during operation.

[0107] 10. The satellite-ground collaborative management and control module supports differentiated reporting strategies. The onboard processing unit can autonomously adjust the reporting frequency according to the current link risk level, and the ground operation and control center can only perform confirmatory interactions when there is a high risk.

[0108] This invention has the following characteristics:

[0109] 1. Breaking through the limitations of existing passive operation and maintenance of inter-satellite links, we have pioneered a proactive operation and maintenance solution that integrates "advance prediction, risk classification, pre-route generation, and pre-switching" throughout the entire process, transforming "post-event remediation" into "pre-event prevention" and completely solving the problem of business interruption caused by intermittent outages.

[0110] 2. Multi-dimensional data fusion prediction takes into account link transmission status, satellite motion characteristics and space environment. The prediction accuracy is much higher than that of single indicator monitoring, effectively reducing the probability of misjudgment and false switching.

[0111] 3. Lightweight space-ground collaborative architecture, which does not require major modifications to existing satellite payloads and ground operation and control systems, is compatible with existing low-Earth orbit constellation deployments, and has low deployment costs and strong compatibility.

[0112] 4. The pre-routing switching process is smooth and seamless, with no packet loss or sudden delays in business transmission, perfectly adapting to high-requirement business scenarios such as direct mobile satellite connection, real-time communication, and industrial IoT.

[0113] 5. Adapts to the large-scale operation and maintenance needs of giant low-Earth orbit constellations, supports batch link synchronization prediction and scheduling, and improves operation and maintenance efficiency by more than 80% compared with the traditional mode.

[0114] 6. Computing power and bandwidth friendly: Through a lightweight LSTM model and differentiated reporting strategy, the CPU overhead for single-satellite prediction is <10ms (200MHz) and the uplink bandwidth is <8kbps. It can be deployed without loss to existing low-Earth orbit constellations without the need for hardware upgrades.

[0115] Contents not described in detail in this specification are existing technologies known to those skilled in the art. It is hereby indicated that the above description is intended to help those skilled in the art understand the present invention, but does not limit the scope of protection of the present invention. Any equivalent substitutions, modifications, improvements, and / or simplifications of the above descriptions that do not depart from the essence of the present invention fall within the scope of protection of the present invention.

Claims

1. A method for predicting and pre-routing inter-satellite link outages in low-Earth orbit satellite constellations, characterized in that: Based on satellite orbit data and space environment parameters, a multi-dimensional link status perception model and a flashover probability prediction algorithm are used to determine the flashover risk level of a single inter-satellite link in advance. For high-risk links, backup transmission routes are calculated and locked in advance, and pre-switching of service data is completed before the link fails. At the same time, the on-board processing unit and the ground constellation operation and control system are linked to achieve lightweight and autonomous link operation and maintenance and route scheduling, thereby avoiding service interruption problems caused by post-event switching.

2. The method for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations according to claim 1, characterized in that, Includes the following steps: Step 1: Collect multi-dimensional real-time data, including link quality indicators, satellite operation parameters, and space environment parameters; Step 2: Preprocess the multi-dimensional real-time data, including using sliding window filtering to remove abnormal noise data, normalizing the data to the [0,1] interval, and uploading it to the ground operation and control center in fixed cycles; Step 3: The ground control center uses a flash failure probability prediction algorithm to calculate the flash failure probability, including inputting the multi-dimensional real-time data into a pre-trained time series prediction model, calculating the flash failure probability within a preset time period, and classifying the link flash failure risk into three levels: low risk, medium risk, and high risk according to a preset threshold. Step 4: Pre-generate backup routes. After eliminating high-risk candidate links and calculating multi-target optimized paths for high-risk links in use, at least one backup route is generated in advance before the predicted time of the outage. Step 5: Distribute the backup route to the relevant satellite nodes to complete the routing table pre-update; Step 6: Within the preset switching time window before the interruption occurs, automatically perform a pre-switching of service traffic from the high-risk link in use to the backup route. Step 7: Perform pre-switch status monitoring on the original link in order to perform a back-switch operation or mark it as a permanent fault pending maintenance. The back-switch operation includes switching service traffic back to the original link and releasing backup routing resources.

3. The method for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations according to claim 2, characterized in that, Step 7 includes: if the original link has no risk of intermittent failure or does not exceed the low risk level within 10 consecutive seconds after switching, it will automatically switch back to the original link; if the original link experiences an actual intermittent failure and does not recover within 30 seconds, it will be determined as a permanent fault and marked for maintenance.

4. The method for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations according to claim 2, characterized in that, The pre-switching trigger time mentioned in step 6 is 1 second before the predicted flashover time. The switching adopts a seamless fast switching mechanism, and the switching time is less than 10 milliseconds.

5. The method for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations according to claim 2, characterized in that, Step 5 includes marking the backup route as "standby" in the pre-updated routing table, which does not take effect immediately.

6. The method for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations according to claim 2, characterized in that, In step 4, the optimization objectives in the multi-objective path optimization calculation include path hop count, end-to-end delay, and maximum link load of the path. The comprehensive cost is calculated by weighted summation, and one or two paths with the lowest comprehensive cost are selected as backup routes. The weight of the path hop count is 0.3, the weight of the end-to-end delay is 0.5, and the weight of the maximum link load of the path is 0.

2.

7. The method for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations according to claim 2, characterized in that, The time series prediction model described in step 3 is a two-layer long short-term memory network LSTM. The model parameters are obtained by training with historical ground data before deployment and are incrementally learned periodically using newly generated flash records during operation.

8. The method for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations according to claim 2, characterized in that, Step 1 involves collecting multi-dimensional real-time data, including implementing a differentiated reporting strategy using the inter-satellite link transceiver module: low-risk links report summary data every 10 seconds, medium-risk links report complete data every 2 seconds, and high-risk links report high-precision data every 0.5 seconds. The space environment parameters include space debris warning level, solar activity intensity, and geomagnetic index.

9. A system for predicting and pre-routing inter-satellite link interruptions in low-Earth orbit satellite constellations, characterized in that: The method comprises a combination of the following modules to perform the low-Earth orbit satellite constellation inter-satellite link intermittent interruption prediction and pre-routing switching method according to any one of claims 1-8: inter-satellite link status awareness module, intermittent interruption risk prediction and classification module, pre-routing generation and scheduling module, pre-switching execution and reversal module, and satellite-ground collaborative management and control module.

Images