Link congestion control load balancing routing method based on routing domain division

Abstract

The invention discloses a link congestion control load balancing routing method based on routing domain division. The method belongs to the technical field of low-orbit satellite routing switching application. The problem that a traditional routing algorithm is high in complexity and slow in response time under the condition of a large network scale is solved. A giant constellation is divided into non-overlapping routing domains, global routing is divided into intra-domain local routing calculation and cross-domain routing interaction, on-satellite computing power consumption is reduced, and the method is adaptive to a large-scale satellite network. Meanwhile, a traditional distributed routing algorithm relates to signaling flooding when the state of a satellite link is changed, the signaling overhead is huge in a large-scale satellite network, local response is carried out in the congestion process by means of a sub-domain architecture and inter-satellite link availability prediction, and the signaling overhead of the whole network is reduced, so that the availability of the routing algorithm is improved.

Description

Technical Field

[0001] This invention belongs to the field of low-Earth orbit satellite routing and switching application technology, specifically relating to a link congestion control load balancing routing method based on routing domain partitioning. Background Technology

[0002] Low Earth Orbit (LEO) satellite networks offer advantages such as flexible bandwidth allocation, low transmission latency, and wide coverage, making them a future trend in network development. However, uneven distribution of terrestrial services can lead to excessive load on some inter-satellite links, causing congestion and performance degradation. On one hand, traditional routing algorithms have limitations in handling dynamic topologies and non-uniform traffic, requiring optimization of load balancing capabilities. On the other hand, as satellite networks continue to expand, the complexity of routing algorithms is also gradually increasing. Traditional routing algorithms suffer from high complexity and slow response times in large-scale networks. Summary of the Invention

[0003] To address the issues of high complexity and slow response time in traditional routing algorithms on large-scale networks, this invention provides a link congestion control load balancing routing method based on routing domain partitioning.

[0004] Includes the following steps: S1. Constructing a Satellite Network Topology Model: Utilizing the regularity of low-Earth orbit (LEO) satellite networks, the LEO satellite network is characterized as a two-dimensional mesh network graph. The construction of the two-dimensional mesh graph follows these principles: Each satellite node... It can establish links with the two nearest satellites in the same orbit and the two nearest satellites in different orbits; S2. Satellite Network Routing Domain Generation: The satellite network topology model is segmented to obtain the satellite routing domain dimension vector. S3. Link congestion threshold calculation: A partial flooding mechanism within the routing domain is adopted to broadcast link state data packets within the routing domain. While ensuring no packet loss on inter-satellite links, link thresholds are defined within the satellite network. ; S4. Source Node Path Generation: The constellation network is divided into several non-overlapping routing domains. When a route is generated, the source node determines whether the destination node is within the current routing domain. If it is within the routing domain, an intra-domain routing algorithm is selected to generate the next-hop path direction. If it is not within the routing domain, an inter-domain forwarding algorithm is selected to generate the next-hop path direction. After the next hop is completed, the node in the next-hop path direction is taken as the source node, and the routing domain determination and next-hop path direction selection are performed again until the complete path is output.

[0005] Furthermore, the two-dimensional mesh network graph is modeled as a graph model. , among them Let E be the set of vertices of the graph, representing satellite nodes in the network, and let E be the set of links between nodes. Each satellite node... A two-dimensional vector tuple express, It is a satellite node Its orbital numbering It is a satellite node The corresponding satellite orbit number.

[0006] Furthermore, step S2 specifically includes: S21. Select the satellite closest to the equatorial plane as the first satellite, and assign a two-dimensional logical index to each satellite based on this. ; S22. Determine the routing cluster size: The number of satellite dimensions in the same orbit within the routing domain. The number of satellite dimensions in different orbits within the routing domain, depending on different , Size calculation network routing convergence time ,in In the formula , These are the propagation delays between satellites on the same orbital plane and between satellites on different orbital planes, respectively. The average update latency across routing clusters, This represents the maximum propagation delay within the routing cluster. S23. Adjust the routing domain size based on ISL connectivity: If a routing domain size contains more than or equal to... A broken ISL link will split the routing domain. ,in This refers to the number of ISLs that do not contain interruptions in the same orbital dimension. The remaining dimensions result in a final satellite routing domain dimension size vector of... .

[0007] Furthermore, link threshold pass Obtain, among which, This is the maximum queue length for satellites. ,in, This represents the number of hops from the source node to the current forwarding node. This refers to the latency of data packets on the inter-satellite link. The rate at which water flows into the buffer. For ISL transmission capacity, Queue occupancy rate This represents the number of cached data packets for the satellite.

[0008] Furthermore, the intra-domain routing algorithm is as follows: S51. The satellite node extracts the destination satellite identifier from the data packet, including the orbit number. and satellite number , with its own identifier If the target satellite is found among the neighboring nodes, the data packet is forwarded directly to that neighboring node. If the target satellite is neither itself nor a neighboring node, the path selection process begins. S52, Compare target satellite identifiers With its own identity Get the main path direction set set of secondary path directions :like The main direction of the path is along the increasing track number, and vice versa; if The primary path direction is ascending along the satellite number, and conversely, descending along the satellite direction; the secondary path direction set... Exclude from the four ISL link directions The remaining directions; S53. After generating available directions, path evaluation is performed. For inter-satellite links within the routing domain, link status is advertised using limited flooding. The source and forwarding nodes evaluate the path based on the link status information in each direction. The evaluation vector is... ; The meanings of each dimension in a vector are as follows: Refers to the set of main path directions set of secondary path directions The corresponding neighboring satellite has not been traversed by the data packet; if the neighboring satellite has not been traversed, return 1; otherwise, return 0. This is based on locally maintained... Variables, variables The initial value of is L, and the set of main path directions is counted. set of secondary path directions The number of faulty ISLs in each direction; for each ISL link that is interrupted in a direction, The value is decremented by 1 to determine the total number of faults on the path of the candidate direction; Q is used to assess the link congestion level. If the condition is true, then Q=0; otherwise, Q=1. This indicates the queue length at each moment; The value is 1 if the path has the fewest hops, and 0 otherwise. By evaluating vectors The principle for evaluation is: for each path in each direction, the evaluation vector... The comparison is performed according to the following rules: starting from the lowest dimension, all vectors are compared sequentially in the same dimension. If different elements appear in a certain dimension, the path containing the vector with the higher value is selected as the next hop path direction. If all dimensions have the same value, the path with the same track number is selected as the next hop direction.

[0009] Furthermore, the inter-domain forwarding algorithm is as follows: S61. For each routing domain, four border satellites are selected preferentially, with satellites located at the four corners of the routing domain being preferred as border satellite nodes: , , , ; S62. For each boundary satellite node, determine the availability of its adjacent cross-domain ISL links, and evaluate the dimension vector as follows: The comparison rule is to start from the lowest dimension and compare the elements of all vectors in the same dimension in turn. If different elements appear in a certain dimension, the vector is selected directly based on the difference in that dimension. Smaller and For shorter paths, if all dimensions have the same value, the path with the same orbit number is selected as the next hop direction; satellite nodes report periodically, matching the link status update frequency within the routing domain; S63. The controller collects the link status information sent by the boundary nodes. If the link status is unavailable, the controller changes the boundary node. The new boundary node is changed according to the direction of increasing or decreasing along the satellite number in the orbit and the direction of increasing or decreasing along the orbit number. S64. After a cross-route domain route is generated, the source node sends a request packet to the control node. Upon receiving the packet, the control node plans the next route domain to be reached based on the segmented topology of the network, selects the destination node based on the status information of the route domain and the boundary nodes, and sends the relevant information to the source node. Attached Figure Description

[0010] Figure 1 This is a flowchart of a link congestion control load balancing routing method based on routing domain partitioning in an embodiment of the present invention; Figure 2 This is a two-dimensional network diagram generated by the Walker-delta constellation in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the generation of cross-routing domain routing paths in an embodiment of the present invention; Figure 4 These are 2D and 3D simulation diagrams of the constellation STK in this embodiment of the invention; Figure 5 This is a comparison chart showing the increase in routing end-to-end latency as the average data packet size increases in an embodiment of the present invention; Figure 6This is a schematic diagram illustrating the increase in routing packet loss rate and average path queue length as the average data packet size increases in an embodiment of the present invention. Detailed Implementation

[0011] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0012] As satellite networks continue to expand, the complexity of routing algorithms is also increasing. Traditional routing algorithms suffer from high complexity and slow response times in large-scale networks. To reduce the complexity of routing algorithms, this embodiment divides the giant constellation into non-overlapping routing domains, decomposing global routing into intra-domain local routing calculations and cross-domain routing interactions, reducing on-board computing power consumption and adapting to large-scale satellite networks. Meanwhile, traditional distributed routing algorithms involve signaling flooding when satellite link states change, leading to huge signaling overhead in large-scale satellite networks. By relying on a domain-based architecture and predicting the availability of inter-satellite links, local responses are implemented during congestion, reducing overall network signaling overhead and improving the availability of the routing algorithm.

[0013] Example 1 This embodiment proposes a segmented routing algorithm for low-Earth orbit satellite constellation networks based on link congestion status, aiming to solve the inter-satellite link congestion problem in the Giants constellation network, achieve balanced distribution of network traffic, reduce packet loss rate, and improve throughput.

[0014] This embodiment targets low-Earth orbit (LEO) satellite networks, aiming to address the issues of uneven load distribution, congestion, and high algorithm complexity in large-scale LEO satellite networks. By reducing routing signaling interaction time and avoiding link interruptions through routing domain division, routing is implemented within each domain using direction-first and link state assessment for route selection. For cross-domain routes, the controller selects boundary nodes, and the shortest path algorithm chooses the next routing domain for forwarding. Combined with a link state update mechanism within each routing domain, this achieves low packet loss and low latency, adapting to large-scale constellation networks.

[0015] The flowchart of the method provided in this embodiment is as follows: Figure 1 As shown, the specific implementation steps are described below.

[0016] Step 1: Construct a satellite network topology model A low Earth orbit (LEO) constellation may consist of hundreds to thousands of satellites. These satellites are organized into several orbits. They can also be interconnected via inter-satellite links, forming LEO networks. Most current mega-satellite constellations still use the Walker constellation architecture. In a Walker-delta constellation, M represents the total number of satellites in the constellation, Z represents the number of orbits in the constellation, and A represents the phase offset parameter between adjacent orbits. The phase difference between adjacent orbits in an M / Z / A configuration constellation can be calculated as: The number of satellites in each orbit is The set of satellite nodes in a constellation can be represented as: any one of the satellites ( Its orbital number It can be determined as: (1) Leveraging the regularity of low-Earth orbit (LEO) satellite networks, these networks can be characterized as two-dimensional mesh graphs. The construction of these two-dimensional mesh graphs follows these principles: each satellite node can establish links with the two nearest satellites in the same orbit and the two nearest satellites across different orbits. The two-dimensional mesh graph generated by the Walker-delta constellation is shown below. Figure 2 As shown. The above two-dimensional satellite network diagram can be modeled as a graph model. , among them Let be the set of vertices of the graph, representing the satellite nodes in the network, and E be the set of links between the nodes. Each satellite node... It can be represented by a two-dimensional vector tuple express, It is a satellite node Its orbital numbering It is a satellite node The corresponding satellite orbit number.

[0017] Step 2: Satellite Network Routing Domain Generation The next step is to segment the large-scale satellite network, with each segment having a routing domain size of [size missing]. . The number of satellite dimensions in the same orbit within the routing domain. This represents the number of satellite dimensions in different orbits within the routing domain. An excessively large network size leads to excessive signaling overhead within the routing domain, while a small routing domain results in increased cross-domain routing and a greater number of controllers. Therefore, a balance needs to be struck between excessively large and small routing domains. The specific segmentation algorithm steps are as follows: (1) Input a weighted undirected connected graph Number of orbits Number of satellites in each orbital plane The first satellite index is determined by selecting the satellite closest to the equatorial plane. Each satellite is assigned a two-dimensional logical index. .

[0018] (2) Determine the routing cluster size: The number of satellite dimensions in the same orbit within the routing domain. This represents the number of satellite dimensions in different orbits within the routing domain. (Depending on different...) , Size calculation network routing convergence time ,in In the formula , These represent the propagation delay between satellites on the same orbital plane and the propagation delay between satellites on different orbital planes, respectively. The average update latency across routing clusters, This represents the maximum propagation delay within the routing cluster (the propagation delay along the diagonal).

[0019] (3) Adjusting the routing domain size based on ISL connectivity: Given the characteristics of time-varying networks, inter-satellite links on the same orbit are relatively stable, while inter-satellite links on different orbits experience periodic connectivity issues. Link connectivity can be predicted; if a routing domain size contains links greater than or equal to... A broken ISL link will split the routing domain. ,in This refers to the number of ISLs that do not contain interruptions in the same orbital dimension. The remaining dimensions result in a final satellite routing domain dimension size vector of... .

[0020] Step 3: Calculate the link congestion state threshold To avoid excessive network signaling overhead caused by large-scale network status updates, this embodiment adopts a partial flooding mechanism within the routing domain. Each satellite node will periodically send data packets to planned adjacent nodes in the network, mainly focusing on the connectivity status of the links. The nodes themselves will also collect their own queue lengths and send them to other satellite nodes.

[0021] Specifically, each satellite maintains an independent queue for each ISL link it connects to (denoted as...). ), corresponding to satellite To neighboring satellites The ISL link. The queue is used to buffer data packets to be transmitted through this ISL link, avoiding momentary congestion caused by the limited transmission capacity of the ISL. The current number of data packets in the queue is defined as... The number of cached data packets for the satellite is Then the queue occupancy rate is: (2) set up For FSPL (Free Space Path Loss), The carrier frequency is [value]. All antennas have fixed transmit power and gain in the main lobe direction; the satellite antenna transmit power is [value]. transmit gain Indicates the receive gain. This indicates. Therefore, from Received at the location The signal strength at that location is: (3) In the formula At the speed of light, For satellite To satellite The physical distance. Assume all satellites have sufficiently narrow antenna beams and precise beam alignment capabilities. In an interference-free environment, the gain of all established ISL links is... At any time from arrive The signal-to-noise ratio (SNR) of continuous transmission can be expressed as: (4) in Boltzmann's constant, Thermal noise, measured in Kelvin. This represents the channel bandwidth, measured in Hertz (Hz). Satellite bandwidth can be calculated using Shannon's formula. and Link capacity between for: (5) use Indicates the average length of the data packet. This refers to the ISL transmission capacity. The waiting time for a data packet in the queue can be calculated as follows: (6) Each satellite periodically monitors its maximum queue length. and the queue length at each time step The average waiting time of all queues, and the link connectivity status (when the link is disconnected). (If the value is 0, the average waiting time is infinite). To prevent excessive signaling overhead between networks, a partial packet flooding strategy is adopted, broadcasting link-state packets within the routing domain. The packet delay on the inter-satellite link can be expressed as the sum of the transmission delay and the propagation delay: (7) To ensure that data packets are not lost on inter-satellite links, the following requirements must be met. , Define the number of hops from the source node to the current forwarding node. The rate at which water flows into the buffer. (Definition) The link threshold in a satellite network can then be defined as: (8) when At this point, it can be assumed that the probability of data packet loss on the current link is high, and the link is in a congested state.

[0022] Step 4: Intra-domain routing algorithm To achieve efficient traffic forwarding in large-scale networks, the large-scale constellation network is divided into multiple disjoint routing domains. When a route is generated, the source node first checks whether the destination node is in the current routing domain. If the destination node is in the current routing domain, the intra-domain routing algorithm is triggered; otherwise, the inter-domain routing algorithm is triggered. In the intra-domain routing algorithm, this embodiment adopts a limited flooding state notification mechanism and supports the generation of multiple paths to ensure the success rate of routing.

[0023] If the current satellite and the target satellite are in the same orbit (same orbit number), the primary path is a single direction along the orbit (determined by the difference in satellite numbers), and the secondary path is a cross-orbit direction orthogonal to that direction. If the current satellite and the target satellite are not in the same orbit, the XY path hop counts corresponding to the two cross-orbit directions (X component) are calculated first, and the direction with the shorter hop count is selected as the primary path direction. If both the X component and the in-orbit direction (Y component) of the primary path are non-zero, there are two primary paths: "X first then Y" and "Y first then X". Each satellite independently completes the primary and secondary path calculations for all target satellites. The calculations only depend on the satellite orbit number and serial number, resulting in lower computational complexity and making it more suitable for the scalability requirements of giant constellations.

[0024] After the source node generates an intra-domain routing entry, the specific execution steps are as follows: (1) The satellite node extracts the destination satellite identifier (orbit number) from the data packet. Satellite number ), and its own identifier If the target satellite is found among the neighboring nodes (by matching a pre-stored list of neighbors), the data packet is forwarded directly to that neighboring node. If the target satellite is neither itself nor a direct neighbor, the process proceeds to the path selection stage.

[0025] (2) Compare the target satellite identifier With its own identity Compare and obtain the main path direction set. set of secondary path directions ;like If the main direction is increasing along the track number, then the main direction is decreasing along the track number; if The primary path direction is the direction that increases along the satellite number, and vice versa. Secondary path directions are also included. Exclude from the four ISL link directions The remaining directions.

[0026] (3) After generating available directions, path evaluation is performed. For inter-satellite links within the routing domain, link status is announced using a limited flooding method. The source node and forwarding node evaluate the path based on the link status information in the direction. The evaluation vector is... The meanings of each dimension in the vector are as follows: The definition refers to the set of main path directions. set of secondary path directions The value is 1 if the corresponding neighboring satellite has not been traversed by the data packet, and 0 if it has. This is based on locally maintained... Variables, variables The initial value of is L, and the set of main path directions is counted. set of secondary path directions The number of faulty ISLs in each direction; for each ISL link that is interrupted in a direction, The value is decremented by 1 to determine the total number of faults on the path of the candidate direction. This is used to assess the link congestion level; if the number of faults on that link direction is [not specified], [the text abruptly ends here]. If the condition is true, then Q=0; otherwise, Q=1. It measures whether the path has the minimum number of hops. If the path has the minimum number of hops, it is 1; otherwise, it is 0.

[0027] By evaluating vectors The principle for evaluation is: for each path in each direction, the evaluation vector... The comparison is performed according to the following rules: starting from the lowest dimension, all vectors are compared sequentially in the same dimension. If different elements appear in a certain dimension, the path containing the vector with the higher value is selected as the next hop path direction. If all dimensions have the same value, the path with the same track number is selected as the next hop direction.

[0028] The above methods can be summarized as follows: Algorithm 1: Next-hop path direction generation algorithm within the routing domain Step (1): Receive data packets and parse the packet header; Step (2): Assume the current node is The destination node is ; Step (3): If : according to as well as Calculate the principal direction ; Else according to , calculate as well as ; Step (4): For all nodes in all directions (up, down, left, right); Step (5): Calculate the number of interrupted links in this direction. ; calculate ; Step (6): If the current link queue length is... ; ; Else ; Step (7): If the current direction belongs to ; ; Else ; Step (8): Generate the decision vector for this direction. ; Step (9): For each direction The comparison is performed according to the following rules: starting from the lowest dimension, all vectors are compared in turn in the same dimension. If different elements appear in a certain dimension, the vector with the higher value is selected. Step 10: Forward the data packet to the next node, take the node in the next hop path direction as the source node, and perform routing domain judgment and next hop path direction selection again until the complete path is output.

[0029] Step 5: Cross-domain routing algorithm For routing entries that cross routing domains, data packets need to be transmitted to the corresponding routing domain. This embodiment employs a cross-routing domain forwarding mode that forwards data packets within the current routing domain to the controller node. Each controller node within a routing domain holds information about the satellite network's routing domain partitions. When a route enters another routing domain, the routing entry is split into routes within the two routing domains and routes between the two controllers across the routing domains. Routes between routing domains are abstracted into a fixed routing topology. The Dijisitra shortest path algorithm is run on the controller node to find the set of routing directions through each routing domain. Routing packets execute the routing algorithm mechanism within each routing domain.

[0030] like Figure 3 As shown, each router controller is aware of the network's routing topology and obtains segmentation information of routes in the satellite network through signaling interaction. The diagram contains three routing domains: Within each routing domain, network information is shared through limited signaling interactions. The topology across routing domains is perceived by the controller, which generates a graph model within each controller to facilitate path calculation. The specific steps of the path generation algorithm for cross-domain routes are as follows: (1) Four boundary satellites are selected for each routing domain, with satellites located at the four corners of the routing domain being selected as boundary satellite nodes: , , , ; (2) For each boundary satellite node, determine the availability of its adjacent cross-domain ISL links, and evaluate the dimension vector as follows: , The number of interrupted links in the inter-satellite link direction at this boundary is determined by comparing elements of all vectors in the same dimension, starting from the lowest dimension. If different elements appear in a certain dimension, the vector is selected based on the difference in that dimension. Smaller and For smaller links, if all dimensions have the same value, the path with the same orbit number is selected as the next hop direction; boundary satellite nodes report periodically (e.g., once every 30 seconds) to match the link status update frequency within the routing domain.

[0031] (3) The controller collects the link status information sent by the boundary nodes. If the link status is unavailable, the controller changes the boundary node. The new boundary node is changed according to the direction of increasing or decreasing satellite number in the orbit and the direction of increasing or decreasing orbit number.

[0032] (4) After a cross-route domain route is generated, the source node sends a request packet to the control node. After receiving the packet, the control node plans the next route domain to be reached based on the segmented topology of the network, selects the destination node based on the status information of the route domain and the boundary nodes, and sends the relevant information to the source node.

[0033] The above methods can be summarized as follows: Algorithm 2: Cross-routing domain path generation algorithm Step (1): Generate segmented routing domain topology ; Step (2): Edge nodes generate cross-domain ISL evaluation vectors ; Step (3): The controller replaces edge nodes based on the evaluation vector; Step (4): The source node sends a routing request to the controller node, and the controller... The Dijistra algorithm is used to find the next routing domain to be reached. ; Step (5): The controller selects the edge node and returns it to the source node; Step (6): The source node runs Algorithm 1 to complete the routing domain. Intra-routing forwarding; Step (7): If the destination node is not present Repeat the above process.

[0034] Example 2 This embodiment employs a segmented routing control method, dividing the Giantsat satellite network into appropriately sized routing domains to facilitate flooding and efficient signaling interaction. Distributed routing and forwarding mechanisms are used in both inter-domain and intra-domain path selection processes, avoiding completely centralized routing decisions. Simultaneously, an XY-direction routing decision mechanism is used within each routing domain, comprehensively considering connectivity, latency, and link congestion to achieve network congestion control with low computational complexity.

[0035] This embodiment uses the Walker-Delta constellation with 3600 stars as the simulation object, such as Figure 4 The 2D and 3D simulation diagrams of the constellation used in this embodiment in STK are shown respectively.

[0036] The permanent inter-satellite link within the constellation orbit is 2198.8 km long, with a propagation delay of approximately 7 ms. The intermittent links across orbits range from approximately 1846.7 km to 2527.6 km, with an average length of 2186 km. Specific parameters are shown in Table 1.

[0037] Table 1 Constellation Simulation Parameter Settings:

[0038] The specific simulation method involves randomly generating 1000 routing entries in the network, randomly generating source and destination nodes in the network, performing routing forwarding simultaneously, and calculating the average latency of different routing algorithms.

[0039] like Figure 5 The difference between Figures (a) and (b) lies in the different simulated satellite ISL link establishment conditions. When the link communication rate cannot meet the requirements, the ISL will actively close. In the simulation of this patent, the ISL crossing the orbital intersection will be closed. In Figure (a), the communication path interruption probability is 8%, and in Figure (b), the network link interruption probability is 16%. Proposd Algorithm refers to the method of this invention.

[0040] Inter-orbit links in satellite networks are intermittent links. Changing the link establishment conditions of intermittent links alters the probability of link interruption. Testing the adaptability of routing algorithms under different network conditions shows that as network conditions worsen, the routing algorithm proposed in this embodiment is less affected by link interruptions. Compared to shortest path routing, this algorithm can effectively avoid congested links in the network and achieve low-latency routing planning. Compared to the shortest path algorithm, the method in this invention reduces latency by at least 20.2%.

[0041] like Figure 6 As shown, with the increase of routing traffic, the packet loss rate of different routing algorithms increases. When the data volume is less than 20KB, the congestion on the link is not significant. In this embodiment, the average queue length of the method (marked as the patented algorithm in the figure) is consistent with that of the shortest path algorithm based on link information (marked as Dijkstra in the figure). As the data volume increases from 5KB to 50KB, the average queue length of the shortest path increases by 120.9%, while the average queue length of the method in this embodiment only increases by 55%. At the same time, with the average load of the satellite link being 50KB, the method in this embodiment still achieves a performance improvement of 27.3%.

Claims

1. A link congestion control load balancing routing method based on routing domain division, characterized in that, The method comprises the following steps: S1, constructing a satellite network topology model: using the regularity of low earth orbit satellite network, the low earth orbit satellite network is described as a two-dimensional mesh network graph, the construction of two-dimensional mesh graph follows the following principles: each satellite node can be linked with the nearest two satellites in the same orbit and the nearest two satellites across the orbit; S2, a satellite network routing domain is generated: the satellite network topology model is segmented to obtain a satellite routing domain dimension size vector; S3, link congestion state threshold calculation: take the mechanism of part flooding in the routing domain, broadcast link state packet in the routing domain, define the link threshold in the satellite network on the premise of ensuring no packet loss on inter-satellite link ; S4, a source node path is generated: the constellation network is divided into several disjoint routing domains, when a group of routes is generated, it is judged at the source node whether the destination node is in the current routing domain, an intra-domain routing algorithm is selected to generate a next hop path direction in the routing domain, an inter-domain forwarding algorithm is selected to generate a next hop path direction when the destination node is not in the routing domain; After the next hop is completed, the node of the next hop path direction is taken as a source node, routing domain judgment and next hop path direction selection are performed again until a complete path is output.

2. The method of claim 1, wherein, Two-dimensional mesh network graphs are modeled as graph models , among them Let be the set of vertices of the graph, representing the satellite nodes in the network. It is a set of links between nodes, each satellite node A two-dimensional vector tuple express, It is a satellite node Its orbital numbering It is a satellite node The corresponding satellite orbit number.

3. The method of claim 2, wherein, Step S2 is specifically: S21, select the satellite closest to the equatorial plane as the first satellite, and on this basis assign a two-dimensional logical index to each satellite ; S22, determining the routing cluster size: the number of dimensions for satellites in the same orbit in the routing domain, the number of dimensions for satellites in different orbits in the routing domain, according to different , the routing convergence time of the network wherein , wherein , are the propagation delay between the same orbital planes and the propagation delay between different orbital planes in the satellite network respectively, wherein is the average update delay across the routing cluster, is the maximum propagation delay within the routing cluster; S23, adjusting the size of the routing domain according to the ISL interruption: if there are more than or equal to interrupted ISL links in one routing domain size, split the routing domain , wherein is the number of non-interrupted ISLs in the same orbit dimension, is the number of remaining dimensions, and the finally generated satellite routing domain dimension size vector is .

4. The method of claim 3, wherein, link threshold by obtained, wherein, is the maximum queue length of the satellite, wherein, is the hop number from the source node to the current forwarding node, is the delay of the data packet in the inter-satellite link, is the rate of the inflow buffer, is the ISL transmission capacity, is the queue occupancy rate, is the number of buffered data packets of the satellite.

5. The method of claim 4, wherein, The intra-domain routing algorithm is specifically: S51, the satellite node extracts the destination satellite identifier in the data packet, including the orbit number and the satellite number , and compares with the self-identifier . If the destination satellite exists in the neighbor node, the data packet is directly forwarded to the neighbor node. If the destination satellite is neither the self nor the neighbor node, the path selection link is entered. S52, compare the destination satellite identifier with the self identifier , obtain the primary path direction set and the secondary path direction set : if , the primary path direction is along the orbit number increasing, otherwise, the primary path direction is along the orbit number decreasing; if , the primary path direction is along the satellite number increasing, otherwise, the primary path direction is along the satellite number decreasing; the secondary path direction set is the remaining direction excluding the four ISL link directions S53, after generating the available directions, path evaluation is performed, the link state is announced in a limited flooding manner for the inter-satellite link within the routing domain, and the source node and the forwarding node evaluate the path according to the link state information in the direction, and the evaluation vector is ; The meanings of each dimension in a vector are as follows: Refers to the set of main path directions set of secondary path directions The corresponding neighboring satellite has not been traversed by the data packet; if the neighboring satellite has not been traversed, return 1; otherwise, return 0. This is based on locally maintained... Variables, variables The initial value of is L, and the set of main path directions is counted. set of secondary path directions The number of faulty ISLs in each direction; for each ISL link that is interrupted in a direction, The value is decremented by 1 to determine the total number of faults on the path of the candidate direction; Q is used to assess the link congestion level. If the condition is true, then Q=0; otherwise, Q=1. This indicates the queue length at each moment; The value is 1 if the path has the fewest hops, and 0 otherwise. By evaluating the vector The principle of evaluation is: for each direction path, the evaluation vector Comparison, comparison rules: from the lowest dimension, compare the same dimension elements of all vectors in turn, if there are different elements in a certain dimension, select the path where the higher value vector is located as the next hop path direction, if all dimensions have the same value, select the path with the same track number as the next hop direction.

6. The method of claim 5, wherein, The inter-domain forwarding algorithm is specifically: S61, each routing domain preferentially selects 4 border satellites, and preferentially selects the satellites located at the four corners of the routing domain as the border satellite nodes: , , , ; S62, judging the availability of the cross-domain ISL link adjacent to each boundary satellite node, and evaluating the dimension vector as The comparison rule is to compare the same dimension elements of all vectors from the lowest dimension, and if different elements appear in a dimension, directly select the path with smaller and smaller according to the dimension difference, and if all dimensions have the same value, select the path with the same track number as the next hop direction; The satellite nodes periodically report and the link state update frequency is matched with the routing domain; S63, the controller collects the link state information sent by the boundary node, if the link state is not available, the boundary node is changed, and the new boundary node is changed according to the increasing or decreasing direction of the satellite number along the orbit and the increasing or decreasing direction of the number of orbits along the orbit; S64, after the cross-routing domain route is generated, the source node sends a request packet to the control node, the control node receives the request packet, plans the next routing domain to be reached according to the segmented topology structure of the network, selects the destination node according to the state information of the routing domain and the boundary node, and sends the related information to the source node. 7.An electronic device comprising a memory and a processor, the memory storing a computer program, wherein, The processor executes the computer program to realize the steps of the method in any one of claims 1-6.

8. A computer readable storage medium for storing computer instructions, characterized in that, The computer instructions are executed by the processor to realize the steps of the method in any one of claims 1-6.

Images