A low-maintenance overhead and reliable low-orbit satellite internet routing method

Abstract

The application discloses a kind of low maintenance overhead and reliable low-orbit satellite internet routing method, belong to low-orbit satellite network routing technical field.The application includes: initialization network topology utilizes satellite quantity, the satellite quantity on each orbit, orbit inclination and orbit height, initializes network topology;Using improved Dijkstra method, calculate and save a satellite network period routing table;In real traffic flow, if congestion occurs, then using FRR-BP based on alternate route fast rerouting mechanism to rebuild the latest path, realize intelligent flow scheduling;In addition, when satellite node failure causes path disconnection, also using FRR-BP based on alternate route fast rerouting mechanism to rebuild path, guarantee the reliability of routing.The application reduces maintenance overhead from two levels of period and inter-period, and uses fast rerouting and intelligent flow scheduling mechanism to enhance the reliability of routing method.

Description

Technical Field

[0001] This invention relates to the field of low-Earth orbit satellite network routing technology, and in particular to a low-maintenance and reliable low-Earth orbit satellite internet routing method. Background Technology

[0002] Low Earth Orbit (LEO) satellite networks consist of ground stations, satellite communication terminals, and LEO satellite constellations with inter-satellite links distributed around the world. Unlike traditional internet service providers, LEO satellite internet service is not geographically limited, meaning users can easily access the internet even in remote areas or places without reliable network coverage. While LEO satellite internet connection speeds and latency may not be as high as traditional internet service providers, its performance is continuously improving with technological advancements.

[0003] Among existing routing algorithms, the BDSR routing algorithm proposed by Zhang et al. considers multiple factors and aims to improve the performance of satellite communication networks by simultaneously optimizing latency and bandwidth. This method relies on each satellite node independently calculating its routing table, which may have limitations due to a lack of global information. Furthermore, this algorithm needs to run continuously during satellite network operation, resulting in high computational and maintenance costs. In addition, the algorithm does not consider path stability; frequent path switching caused by satellite motion will lead to significant routing maintenance costs. Summary of the Invention

[0004] In view of this, this invention proposes a low-maintenance and reliable routing method for low-Earth orbit satellite internet, based on the traditional shortest path algorithm and combined with the periodicity and predictability of low-Earth orbit satellite networks. This method reduces maintenance overhead at both the intra-cycle and inter-cycle levels, and enhances the reliability of the routing method using fast rerouting and intelligent traffic scheduling mechanisms.

[0005] The technical solution adopted in this invention is as follows:

[0006] A low-maintenance and reliable low-Earth orbit satellite internet routing method includes the following steps:

[0007] (1) Based on the characteristics that the satellites in the Walker constellation have the same orbital altitude and orbital inclination and are evenly distributed with the Earth as the center, the network topology is initialized using the number of satellites, the number of satellites in each orbit, the orbital inclination and the orbital altitude.

[0008] (2) The period of the satellite orbit is determined according to Kepler's third law. The rotation period of the ground station is equal to the Earth's rotation period. The period of the entire satellite network is determined to be the least common multiple of the period of the satellite orbit and the rotation period of the ground station.

[0009] (3) Using the improved Dijkstra method, calculate and save the routing table for one satellite network cycle; based on the periodicity of the satellite network, route calculation will not be performed in subsequent cycles, and the path selection will be performed directly by referring to the saved routing table.

[0010] The improvement in the Dijkstra method refers to determining the weights of inter-satellite and satellite-to-ground connections using the following method:

[0011] Define a constant λ, 0 < λ < 1. For a candidate neighbor of the current node, if the candidate neighbor is still within the communication range after N time slices, then set the weight of the corresponding edge to the normal distance; if the candidate neighbor is not within the communication range after N time slices, then the candidate neighbor is considered unstable, and the weight of the corresponding edge is increased, so that the weight of the corresponding edge is: distance / λ.

[0012] (4) In real business flow, nodes where multiple paths intersect are defined as "hot spots". The total traffic at the "hot spots" is used to determine whether network congestion has occurred. If congestion occurs, the latest path is rebuilt using the FRR-BP fast rerouting mechanism based on the backup path to achieve intelligent traffic scheduling. In addition, when a satellite node fails and causes a path disconnection, the path is also rebuilt using the FRR-BP fast rerouting mechanism based on the backup path to ensure the reliability of the route.

[0013] Furthermore, the specific method of step (1) is as follows:

[0014] (101) Input the number of satellites N respectively s The number of satellites N in each orbit s Given perP, track inclination angle i, and track height h, initialize the network topology;

[0015] (102) Number the satellites and ground stations:

[0016] S ID =(P num *N s perP)+offset num

[0017] Among them, S ID Indicates the satellite's designation, P num Indicates the orbital number, N s perP represents the number of satellites in each orbit, offset num Indicates the offset;

[0018] At simulation time t=0, the first satellite starting from the ascending node has an offset number "0"; satellites are numbered sequentially according to their orbital motion direction, up to N. sperP-1, the first satellite was located in orbit number 1, and the orbit numbers increased eastward.

[0019] Furthermore, the specific method of step (2) is as follows:

[0020] (201) Calculate the satellite's orbital period using Kepler's third law:

[0021]

[0022] Where T represents the satellite orbital period, a represents the semi-major axis, and if the orbit of the Walker constellation satellites is considered as a circle, then the semi-major axis is approximately the orbital altitude, G represents the gravitational constant, and M represents the mass of the Earth.

[0023] (202) Calculate the least common multiple of the period of the ground station as the Earth rotates and the period of the satellite orbit, and use it as the period of the entire satellite network.

[0024] Furthermore, the specific method of step (4) is as follows:

[0025] (401) Determine if a "hot spot" exists; In the case of multiple services running in parallel, if multiple paths pass through the same node, then this node is called a "hot spot";

[0026] (402) Determine whether the total traffic of the "hot spot" exceeds the traffic threshold. If it exceeds the threshold, network congestion is considered to have occurred.

[0027] (403) If network congestion occurs, the "hot spot" is removed from the topology graph and all connections associated with it are deleted. Then the improved Dijkstra method is rerun to calculate the alternative path and obtain a new path that avoids the "hot spot" to ensure normal network operation.

[0028] (404) When a satellite node fails and causes a path to be disconnected, the faulty node is removed from the topology graph and all connections associated with it are deleted. Then, the improved Dijkstra method is rerun to calculate an alternative path and obtain a new path that avoids the faulty node, thus ensuring the normal operation of the network.

[0029] The beneficial effects of this invention are as follows:

[0030] 1. This invention takes into account both path stability and path delay, and prioritizes selecting neighbors with longer continuous connection times as the next-hop node. Although the path obtained by this method may not be the shortest path, it reduces path switching caused by the movement of low-Earth orbit satellite constellations and reduces path maintenance overhead at the cycle level.

[0031] 2. This invention utilizes the periodicity of low-Earth orbit satellite networks, namely, the topology and routing of low-Earth orbit satellite constellations are exactly the same at the same time in different periods. Therefore, it is only necessary to save the routing information within a complete period, and there is no need to calculate and maintain the routing table in subsequent periods, thereby reducing the path maintenance overhead at the periodic level.

[0032] 3. This invention proposes a fast rerouting mechanism for satellite node failures, which searches for alternative paths when a node fails, ensuring normal communication between ground stations and increasing the fault tolerance of the network system. In addition, an intelligent traffic scheduling mechanism is proposed for situations where multiple paths pass through the same node (hotspot), switching to alternative paths to avoid hotspot nodes and alleviating network congestion. The fast rerouting and intelligent traffic scheduling mechanisms enhance the reliability and fault tolerance of the routing algorithm. Attached Figure Description

[0033] Figure 1 This is a schematic diagram illustrating the routing scenario using the traditional Dijkstra algorithm.

[0034] Figure 2 This is a schematic diagram illustrating the routing situation using the method of the present invention.

[0035] Figure 3 A diagram illustrating the fast rerouting mechanism for switching alternative paths.

[0036] Figure 4 This is a diagram illustrating network congestion.

[0037] Figure 5 A diagram illustrating the switching of alternative paths for the intelligent traffic scheduling mechanism.

[0038] Figure 6 This is a comparison chart showing the performance of the LMCRA algorithm of this invention with the traditional Dijkstra's shortest path algorithm.

[0039] Figure 7 A bar chart showing the impact of different networking schemes on path stability and latency.

[0040] Figure 8 This is a data graph showing the impact of time span on path stability.

[0041] Figure 9 (a), (b), and (c) in the figure are bar charts showing the impact of time span on path delay under the Ideal, +Grid, and Sparse networking schemes, respectively.

[0042] Figure 10 This is a 50x50 constellation topology diagram.

[0043] Figure 11 This is a bar chart showing the impact of constellation density on path stability and delay. Detailed Implementation

[0044] A low-maintenance and reliable low-Earth orbit satellite internet routing method includes the following steps:

[0045] (1) Based on the characteristics that the satellites in the Walker constellation have the same orbital altitude and orbital inclination and are evenly distributed with the Earth as the center, the network topology is initialized using the number of satellites, the number of satellites in each orbit, the orbital inclination, and the orbital altitude; the specific method is as follows:

[0046] (101) Input the number of satellites N respectively s The number of satellites N in each orbit s Given perP, track inclination angle i, and track height h, initialize the network topology;

[0047] (102) Number the satellites and ground stations:

[0048] S ID =(P num *N s perP)+offset num

[0049] Among them, S ID Indicates the satellite's designation, P num Indicates the orbital number, N s perP represents the number of satellites in each orbit, offset num Indicates the offset;

[0050] At simulation time t=0, the first satellite starting from the ascending node has an offset number "0"; satellites are numbered sequentially according to their orbital motion direction, up to N. s perP-1, the first satellite's orbital number is 1, and the orbital numbers increase eastward;

[0051] (2) The period of the satellite orbit is determined according to Kepler's third law. The rotation period of the ground station is equal to the Earth's rotation period. The period of the entire satellite network is determined as the least common multiple of the period of the satellite orbit and the rotation period of the ground station. The specific method is as follows:

[0052] (201) Calculate the satellite's orbital period using Kepler's third law:

[0053]

[0054] Where T represents the satellite orbital period, a represents the semi-major axis, and if the orbit of the Walker constellation satellites is considered as a circle, then the semi-major axis is approximately the orbital altitude, G represents the gravitational constant, and M represents the mass of the Earth.

[0055] (202) Calculate the least common multiple of the period of the ground station as the Earth rotates and the period of the satellite orbit, and use it as the period of the entire satellite network;

[0056] (3) Using the improved Dijkstra method, calculate and save the routing table for one satellite network cycle; based on the periodicity of the satellite network, route calculation will not be performed in subsequent cycles, and the path selection will be performed directly by referring to the saved routing table.

[0057] The improvement in the Dijkstra method refers to determining the weights of inter-satellite and satellite-to-ground connections using the following method:

[0058] Define a constant λ, 0 < λ < 1. For a candidate neighbor of the current node, if the candidate neighbor is still within the communication range after N time slices, then set the weight of the corresponding edge to the normal distance; if the candidate neighbor is not within the communication range after N time slices, then the candidate neighbor is considered unstable, and the weight of the corresponding edge is increased, so that the weight of the corresponding edge is: distance / λ.

[0059] (4) In real business flows, nodes where multiple paths intersect are defined as "hot spots." The total traffic at these "hot spots" is used to determine if network congestion has occurred. If congestion occurs, the latest path is reconstructed using the FRR-BP fast rerouting mechanism based on backup paths, achieving intelligent traffic scheduling. Furthermore, when a satellite node fails, causing a path disconnection, the FRR-BP fast rerouting mechanism based on backup paths is also used to reconstruct the path, ensuring routing reliability. Specifically:

[0060] (401) Determine if a "hot spot" exists; In the case of multiple services running in parallel, if multiple paths pass through the same node, then this node is called a "hot spot";

[0061] (402) Determine whether the total traffic of the "hot spot" exceeds the traffic threshold. If it exceeds the threshold, network congestion is considered to have occurred.

[0062] (403) If network congestion occurs, the "hot spot" is removed from the topology graph and all connections associated with it are deleted. Then the improved Dijkstra method is rerun to calculate the alternative path and obtain a new path that avoids the "hot spot" to ensure normal network operation.

[0063] (404) When a satellite node fails and causes a path to be disconnected, the faulty node is removed from the topology graph and all connections associated with it are deleted. Then, the improved Dijkstra method is rerun to calculate an alternative path and obtain a new path that avoids the faulty node, thus ensuring the normal operation of the network.

[0064] This method ensures low maintenance overhead for routing by considering multiple aspects. First, from a periodic perspective, this method recognizes the periodicity and predictability of the LEO satellite network. Based on these characteristics, this method proposes a scheme that only needs to store a routing table for one period. In this case, subsequent periods can directly refer to the pre-stored routing table, avoiding redundant calculations and thus greatly reducing the computation and maintenance costs of the routing table. From an intra-period perspective, since the satellite network is dynamic and path switching caused by satellite movement is unavoidable, this method proposes a routing method that comprehensively considers path stability and delay (referred to as the LMCRA algorithm) by combining the predictability of satellite positions with the traditional shortest path algorithm. This algorithm prioritizes the neighbor with a longer duration as the next hop based on whether the two nodes can still establish a connection after N steps, based on the predictability of satellite positions, thus improving path stability, reducing routing table maintenance overhead, and keeping the delay within a small range.

[0065] Finally, to ensure the fault tolerance and reliability of the routing algorithm, this method employs a fast rerouting mechanism and an intelligent traffic scheduling mechanism. The fast rerouting mechanism allows the affected path to be switched to an alternative path when a satellite node fails, ensuring normal network communication. Based on the predictability and periodicity of the satellite network, this mechanism can quickly detect node failures and rapidly switch to an alternative path, thus ensuring network fault tolerance. The intelligent traffic scheduling mechanism allows for intelligent path switching when the satellite network experiences congestion, ensuring smooth network communication and guaranteeing network reliability.

[0066] The principles of this invention will be explained in more detail below:

[0067] Because satellites are constantly moving, the connections between them are constantly changing over time. After a period of time, two satellites may exceed their maximum communication distance. Therefore, path switching caused by the movement of the satellite constellation is unavoidable, which can lead to severe latency jitter. For example, if there is a link between satellite A and satellite B at time t, and the link distance between satellite A and satellite B exceeds the maximum communication distance at time t+1, then the link will also cease to exist at time t+1. Due to this characteristic, although the shortest path found using Dijkstra's algorithm has the shortest distance, its path stability often fails to meet requirements; that is, the path will frequently switch. Utilizing the periodicity of satellite networks, the position of any satellite at any time can be predicted. Based on the current state of the satellite network, the positions of all satellites after N steps can be predicted. Therefore, selecting candidate neighbor nodes that are still within communication range after N steps as the next hop ensures that the path remains stable for at least N steps without switching, thus improving path stability.

[0068] This method is based on Dijkstra's shortest path algorithm and, considering path stability, proposes the LMCRA algorithm. The main idea of ​​the LMCRA algorithm is: when a node selects its next hop, among the candidate neighbors, it prioritizes selecting a neighbor that remains within communication range after N steps as the final next hop node. This ensures a longer connection duration between the two satellite nodes. If there are multiple candidate neighbors satisfying the above condition, based on Dijkstra's algorithm, the nearest neighbor is selected as the final next hop. This achieves a balance between path stability and path length. In practice, a constant λ (0 < λ < 1) can be defined. For a candidate neighbor of the current node, if the candidate neighbor remains within communication range after N steps, the weight of the corresponding edge is set to the normal distance; if the candidate neighbor is no longer within communication range after N steps, the candidate neighbor is considered unstable, and the weight of the corresponding edge is increased, which can be set to distance / λ. In this way, the weight is amplified. Since Dijkstra's algorithm prioritizes edges with smaller weights, i.e., closer nodes, this unstable edge is less likely to be selected. If there are multiple stable candidate neighbors, the algorithm will select the one with the smallest weight as the next hop. Therefore, this is an algorithm that comprehensively considers path stability and path delay.

[0069] For a simple example, in Figure 1 In the topology diagram shown, each circle represents a network node (satellite), with the numbers inside representing node numbers and the numbers on the edges representing weights. Assuming the starting and ending points are nodes 1 and 6 respectively, there are two paths to choose from: ['1','2','6'] and ['1','4','5''6'], with weights of 40 and 70 respectively. Using the traditional Dijkstra's algorithm, the algorithm would choose the shortest path, ['1','2','6']. If a new routing algorithm that comprehensively considers stability and delay is used, and based on the periodicity of the satellites, it is predicted that the connection between nodes 1 and 2 cannot be maintained for more than N steps, a constant λ = 0.01 is taken. Then, at that moment, the topology diagram is updated as follows: Figure 2 In this case, the weight of the edge between node 1 and node 2 is updated to 10 / λ, which is 100. The algorithm then selects ['1', '4', '5', '6']. Compared to the traditional Dijkstra's algorithm, this path has better stability, but at the cost of a slightly larger distance, leading to a slightly increased delay. If both paths are relatively stable and can be maintained for N steps, the weights remain unchanged. In this case, the algorithm is the same as the traditional Dijkstra's algorithm, but the new algorithm selects the path with the smallest weight ['1', '2', '6'] while ensuring path stability.

[0070] To meet QoS requirements, after the routing algorithm solves for the path, the path is filtered, and only those paths that meet the QoS requirements are retained and displayed. Since there is a maximum communication distance between satellite nodes, communication between satellites is impossible beyond this distance, so there may be cases where no path exists. After filtering, the paths are divided into three categories: paths exist and meet QoS requirements, paths exist but do not meet QoS requirements, and paths do not exist. Different prompts are output according to the different categories of paths.

[0071] Leveraging the periodicity of satellite networks can reduce the computational overhead of routing tables. Specifically, the routing information for a given period is completely saved by dividing it into time slices (meaning the constellation topology remains unchanged within each time slice). Routing information for subsequent periods can then be directly referenced from this saved information. This method trades storage overhead for computational and maintenance overhead. The aforementioned new algorithm ensures strong stability of routes generated within a satellite period, thus guaranteeing low maintenance overhead at the periodic level.

[0072] Fast Reroute (FRR) is a network fault recovery technique used to quickly switch data traffic from the failed path to a backup path when a network failure occurs, thus avoiding the impact of network failure on service quality and reliability. In the FRR mechanism, the calculation of the backup path is crucial. This method employs a Fast Reroute Based on Backup Path (FRR-BP) mechanism to ensure the normal operation of the network.

[0073] When a node in the satellite network fails, the Fast Rerouting Based on Backup Path (FRR-BP) mechanism can be activated immediately to recalculate the backup path from the origin to the destination and avoid the failed node, thereby minimizing the impact of network failure on service quality and reliability.

[0074] by Figure 2 Taking the topology graph as an example, if node 5 fails at a certain moment, this will affect the path ['1','4','5','6'] from node 1 to node 6. At this time, the system immediately starts the fast rerouting mechanism, deleting node 5 and its related connections from the topology graph. The updated topology is as follows: Figure 3 As shown, at this point, the original path ['1','4','5','6'] has become invalid, and the fast rerouting mechanism searches for alternative paths. Figure 3 The alternative paths shown are ['1','2','6'].

[0075] Intelligent traffic scheduling is a technology for addressing network congestion. It can automatically select alternative paths and avoid "hotspot" nodes, thereby maintaining network stability and performance. However, implementing intelligent traffic scheduling requires the support of multiple services running in parallel. To this end, we adopted a parallel routing algorithm approach to find the optimal path for different origin-destination pairs.

[0076] During path calculation, multiple paths are allowed to exist simultaneously. If multiple paths pass through the same node, that node becomes a "hotspot." The total traffic at the "hotspot" is checked; if the total traffic exceeds a threshold, the "hotspot" becomes a congested node. For handling congested nodes, the newly generated paths passing through the "hotspot" are rerouted using the FRR-BP mechanism to avoid the "hotspot" and alleviate network congestion.

[0077] by Figure 4 Taking the topology diagram as an example, the traffic threshold is set to 25. There are currently two paths: one is ['1','2','3','6'], and the other is ['4','5','3','7'], marked with dashed lines in the diagram. Both paths pass through node 3, making node 3 a "hotspot." The numbers on the connections in the diagram represent service traffic. As shown in the diagram, the total traffic at node 3 is 30, exceeding the traffic threshold. Therefore, network congestion occurs at node 3. The second path is rerouted to bypass node 3, and the new path is ['4','5','7']. The updated topology is as follows: Figure 5 As shown.

[0078] The LMCRA algorithm is tested below:

[0079] In a satellite constellation of 10x10 (representing 10 orbital planes, 10 satellites per plane), with an inclination of 60 degrees and an altitude of 1200 km, a new algorithm that comprehensively considers path stability and distance is compared with the bidirectional Dijkstra algorithm, using a step size of 10 seconds and N=1. A segment of the results is shown below. Figure 6 As shown, the left side represents the results of the new algorithm, and the right side represents the results of the bidirectional Dijkstra algorithm. At 1400s, satellite node 39 was still within the communication range of the starting point. However, after one more step, node 39 was no longer within the starting point's communication range. Therefore, the new routing algorithm predicted that selecting node 39 would result in a path switch after one step. Node 29, on the other hand, remained within the starting point's communication range in both the current and next step steps, thus preventing a path switch. Therefore, the more "stable" node 29 was prioritized.

[0080] To investigate the impact of different networking schemes on algorithm performance, a 10x10 satellite constellation with an inclination of 60 degrees and an orbital altitude of 1200km was constructed. Ground stations included locations around the world. The step size and time span were both set to 10 seconds, meaning that during the next-hop selection process, it was determined whether the next long-term candidate neighbor was still within communication range. To ensure the reliability of the results, the total runtime was set to 1*10^6 seconds. 5 s, i.e., 1*10 4 The stability of a path is compared based on the average duration of each path, with the path delay being "path length / speed of light".

[0081] The following compares the differences between the Beijing-Los Angeles link using the new routing algorithm and the traditional Dijkstra algorithm under different networking schemes:

[0082] The three networking schemes are as follows: In the Ideal scenario, a satellite can establish connections with all other satellites within its visible range. In the +Grid scenario, each satellite can establish four links: one between two satellites in the same orbit and the closest satellite in an adjacent orbit. Sparse is a special case of +Grid, where each satellite can only establish two connections.

[0083] The results of the path stability and delay comparison are as follows Figure 7 As shown in the figure, the cost of achieving higher stability often leads to an increase in path length. However, the increase in length caused by the new algorithm is not significant and is within an acceptable range. Compared to the other two schemes, the Sparse scheme shows a more significant improvement in path stability, but the resulting latency cost is also greater.

[0084] As mentioned above, the new algorithm prioritizes candidate neighbors that are still within communication range after N steps, aiming to extend the duration of inter-satellite connections. N steps are defined as the time span. To investigate the impact of the time span on algorithm performance, N = 1, 2, 3, 4, and 5 were used. The performance of the new routing algorithm and the traditional Dijkstra algorithm was compared under different time spans and different network topologies. The results of the stability comparison are as follows: Figure 8 As shown, when using the traditional algorithm, the +Grid scheme exhibits significantly better stability than the other two schemes. When using the new routing algorithm, the +Grid scheme shows a peak in path stability when N=3, while the Sparse scheme generally shows a decreasing trend in path stability as the time span increases. In the Ideal case, the stability remains relatively stable with changes in N, showing little variation. The latency comparison results are as follows... Figure 9As shown, under the Ideal scheme, the new routing algorithm exhibits minimal latency increases across all time spans. Under the +Grid scheme, the new routing algorithm shows slightly larger latency increases at time spans N=4 and N=5, but minimal increases in other cases. Under the Sparse scheme, the new routing algorithm shows very small latency increases at time spans N=1 and N=2, and slightly larger latency increases at N=3, 4, and 5.

[0085] Constellation density has a crucial impact on satellite networks. To investigate the effect of constellation density on algorithm performance, we used satellite constellations of 10x10, 20x20, 30x30, 40x40, and 50x50 for our study. Figure 10 The experiment showcases constellation topologies of 10x10, 30x30, and 50x50. The network topology in this experiment uses the +Grid method. The start and end points of the paths are still set to Beijing and Los Angeles, with a step size of 10 seconds. The time span of the new algorithm is set to one step, resulting in a total of 1*10 iterations. 5 The stability and latency of the paths obtained by the traditional Dijkstra algorithm and the new routing algorithm are compared, and the results are as follows: Figure 11 As shown, in terms of path stability, the overall stability tends to decrease with increasing constellation density, but a local maximum occurs at a constellation density of 30x30. At this density, the new routing algorithm also significantly improves path stability. Another significant improvement is seen at a constellation density of 50x50, but path stability is poor at this density, and the constellation itself incurs high costs. Regarding path latency, the overall latency also tends to decrease with increasing constellation density, indicating that the new routing algorithm improves stability while maintaining a relatively low latency cost.

[0086] To ensure paths meet QoS requirements, filtering is necessary after path generation. This can be achieved by setting latency-limited paths. Since latency is calculated by dividing path length by the speed of light, latency-limited paths are equivalent to distance-limited paths. The maximum distance can be set to 1.8 * 10^6. 7For paths meeting the requirements, output the path and its length; for those not meeting the requirements, output the message "Path does not meet latency requirements". During the test, at 6670s, a path could not be formed because a neighboring satellite in a certain hop was out of communication range, so the message "Path does not exist" was output. At 6760s, a path ['-12','41','42','43','52','62','-71'] appeared with a length of 18,247,328m, which exceeded the maximum distance limit, so the message "Path does not meet requirements" was output, and the path was not marked on the topology map. At 6810s, a path ['-12','31','32','33','-71'] appeared with a length of 12,809,763m. This path met the maximum distance limit, so the path and its length were output and marked on the topology map.

[0087] To reduce the computational and maintenance overhead of the routing table, the routing algorithm can be run completely once within a cycle, and then run again at the beginning of each step to update the path. This method saves the routing table for a complete cycle. Starting from the next cycle, it only needs to refer to the saved routing table to look up the path at the corresponding time. This method trades storage overhead for computational and maintenance overhead, reducing computational and maintenance costs at the cycle level.

[0088] This embodiment tested the system's fast rerouting function. At 2650 seconds into the simulation, the path for the first service was path1: ['-12','37','47','57','67','-71']. At this point, satellite node 47 failed. The system immediately activated the fast rerouting mechanism, re-finding the path from the origin to the destination and completing the path switch. The new backup path was ['-12','37','38','39','-120','67','-71']. This backup path avoided the failed node, ensuring network reliability. The paths displayed on the topology diagram were also updated accordingly.

[0089] This embodiment tested the system's intelligent traffic scheduling function. After 8000 seconds of simulation, the traffic for path 1 of the first service (['11','1','91','-29','38','28']) was 20, and the traffic for path 2 of the second service (['15','25','35','36','37','38','39']) was also 20. Both paths passed through satellite node 38, making node 38 a "hotspot." Calculations showed that the total traffic at node 38 was 40, exceeding the threshold of 30, thus indicating network congestion. At this point, the system immediately activates the intelligent traffic scheduling mechanism. While path1 remains unchanged, path2 is switched to an alternative path that avoids node 38. Path2 changes from the original ['15','25','35','36','37','38','39'] to ['15','25','26','27','28','29','39'], thus avoiding the congested node. The paths and related information on the topology graph are also updated accordingly.

[0090] In summary, this invention addresses the challenges of finding more stable paths, balancing path stability with path length, and reducing the overhead of routing table computation. It also considers paths that meet certain QoS requirements (e.g., delay-limited paths with latency no greater than a certain value). This invention employs a fast rerouting mechanism; when a path experiences a disconnection (i.e., a node on the path fails), it can rebuild the route, find alternative paths, and establish a new connection. Furthermore, it utilizes an intelligent traffic scheduling mechanism; when multiple paths have duplicate nodes, these nodes become "hotspots," and this invention can perform path switching at these "hotspots" to prevent congestion.

Claims

1. A low-maintenance and reliable low-Earth orbit satellite internet routing method, characterized in that, Includes the following steps: (1) Based on the characteristics that the satellites in the Walker constellation have the same orbital altitude and orbital inclination and are evenly distributed with the Earth as the center, the network topology is initialized using the number of satellites, the number of satellites in each orbit, the orbital inclination and the orbital altitude. (2) The period of the satellite orbit is determined according to Kepler's third law. The rotation period of the ground station is equal to the Earth's rotation period. The period of the entire satellite network is determined to be the least common multiple of the period of the satellite orbit and the rotation period of the ground station. (3) Using the improved Dijkstra method, calculate and save the routing table for one satellite network cycle; based on the periodicity of the satellite network, route calculation will not be performed in subsequent cycles, and the path selection will be performed directly by referring to the saved routing table. The improvement in the Dijkstra method refers to determining the weights of inter-satellite and satellite-to-ground connections using the following method: Define a constant λ, 0 < λ < 1. For a candidate neighbor of the current node, if the candidate neighbor is still within the communication range after N time slices, then set the weight of the corresponding edge to the normal distance; if the candidate neighbor is not within the communication range after N time slices, then the candidate neighbor is considered unstable, and the weight of the corresponding edge is increased, so that the weight of the corresponding edge is: distance / λ. (4) In real business flow, the nodes where multiple paths intersect are defined as "hot spots". The total traffic at the "hot spots" is used to determine whether network congestion has occurred. In the event of congestion, the latest path is reconstructed using the FRR-BP fast rerouting mechanism based on the backup path, thus achieving intelligent traffic scheduling. In addition, when a satellite node fails and causes a path disconnection, the FRR-BP fast rerouting mechanism based on the backup path is also used to reconstruct the path, ensuring the reliability of the route.

2. The low-maintenance and reliable low-Earth orbit satellite internet routing method according to claim 1, characterized in that, The specific method for step (1) is as follows: (101) Input the number of satellites N respectively s The number of satellites N in each orbit s Given perP, track inclination angle i, and track height h, initialize the network topology; (102) Number the satellites and ground stations: S ID =(P num *N s perP)+offset num Among them, S ID Indicates the satellite's designation, P num Indicates the orbital number, N s perP represents the number of satellites in each orbit, offset num Indicates the offset; At simulation time t=0, the first satellite starting from the ascending node has an offset number "0"; satellites are numbered sequentially according to their orbital motion direction, up to N. s perP-1, the first satellite was located in orbit number 1, and the orbit numbers increased eastward.

3. The low-maintenance and reliable low-Earth orbit satellite internet routing method according to claim 1, characterized in that, The specific method for step (2) is as follows: (201) Calculate the satellite's orbital period using Kepler's third law: Where T represents the satellite orbital period, a represents the semi-major axis, and if the orbit of the Walker constellation satellites is considered as a circle, then the semi-major axis is approximately the orbital altitude, G represents the gravitational constant, and M represents the mass of the Earth. (202) Calculate the least common multiple of the period of the ground station as the Earth rotates and the period of the satellite orbit, and use it as the period of the entire satellite network.

4. The low-maintenance and reliable low-Earth orbit satellite internet routing method according to claim 1, characterized in that, The specific method for step (4) is as follows: (401) Determine if a "hot spot" exists; In the case of multiple services running in parallel, if multiple paths pass through the same node, then this node is called a "hot spot"; (402) Determine whether the total traffic of the "hot spot" exceeds the traffic threshold. If it exceeds the threshold, network congestion is considered to have occurred. (403) If network congestion occurs, the "hot spot" is removed from the topology graph and all connections associated with it are deleted. Then the improved Dijkstra method is rerun to calculate the alternative path and obtain a new path that avoids the "hot spot" to ensure normal network operation. (404) When a satellite node fails and causes a path to be disconnected, the faulty node is removed from the topology graph and all connections associated with it are deleted. Then, the improved Dijkstra method is rerun to calculate an alternative path and obtain a new path that avoids the faulty node, thus ensuring the normal operation of the network.

Images