Low-orbit satellite congestion control method based on artificial immune algorithm

Abstract

The invention discloses a low-orbit satellite congestion control method based on an artificial immune algorithm, and the method comprises the steps: a satellite node periodically monitors the occupancy rate of an output queue in each direction, and when congestion is detected, a satellite generates a congestion early-warning packet as an antigen, and transmits the congestion early-warning packet to a direct adjacent node; the adjacent nodes receiving the congestion early warning packet search matching items of the affected target nodes in a historical memory cell bank according to the similarity between the current scene and historical records, and the matching items serve as routing strategies of the target nodes; when the matching is not inquired, determining an optimal detouring strategy based on the direction of the congested link and the position information of the satellite; aiming at the generated control strategy, comprehensively evaluating the satellite in a time window; when the affinity continuously does not reach a threshold value, the satellite continuously monitors the change trend of the queue occupancy rate, and when it is found that a plurality of continuous periods are in a rising trend, the shunting proportion is increased; according to the invention, the response speed, the network performance and the service quality of congestion control are improved.

Description

Technical Field

[0001] The invention relates to the technical field of satellite communication networks, and in particular to a low-orbit satellite congestion control method based on an artificial immune algorithm. Background Art

[0002] As global communication needs continue to grow, low-orbit satellite (LEO satellite) networks are becoming a key component of the next-generation communication system with their advantages of low latency and wide coverage. However, with the continuous expansion of satellite network scale and the sharp increase in business traffic, network congestion has become increasingly prominent, which has seriously affected the network's service quality and user experience.

[0003] Traditional ground network congestion control methods are no longer applicable to low-orbit satellite networks. This is mainly because low-orbit satellite networks have the following unique characteristics: first, the high-speed movement of satellite nodes leads to dynamic changes in network topology; second, the limited bandwidth and high transmission delay of inter-satellite links put forward higher requirements for congestion control; finally, the computing and storage resources of satellite nodes are limited, making it difficult to support complex control algorithms.

[0004] At present, most existing satellite network congestion control schemes use a simple routing detour strategy, which directly transfers traffic to an alternative path when link congestion is detected. Although this method is simple to implement, it has serious defects: on the one hand, due to the lack of a reasonable traffic scheduling mechanism, it is easy to cause congestion problems to spread in the network; on the other hand, the complete detour strategy often significantly increases transmission delays and reduces network resource utilization. More importantly, this type of method cannot learn from historical experience and is difficult to cope with dynamic changes in network status.

[0005] In addition, existing congestion control schemes often ignore the physical characteristics of low-orbit satellite networks, especially the link instability problem in special environments such as polar regions. Since the satellite orbits in the polar regions show significant convergence characteristics, the relative motion between adjacent satellites is more intense, coupled with the influence of harsh climate environment, it is difficult for traditional congestion control strategies to ensure the stable operation of the network.

[0006] Based on the above problems, there is an urgent need for an intelligent control strategy that can fully consider the characteristics of low-orbit satellite networks, effectively control network congestion, and maintain network stability. This strategy should have adaptive learning capabilities, be able to acquire knowledge from historical experience, and make timely adjustments based on the dynamic changes in network status. At the same time, it should also take into account the physical characteristics and resource constraints of the network to ensure the practicality and reliability of the control strategy. Summary of the invention

[0007] In order to solve the congestion control problem in low-orbit satellite networks, the present invention provides a low-orbit satellite congestion control method based on an artificial immune algorithm. By introducing the adaptive mechanism of the biological immune system into the congestion control process and combining it with a grid routing framework, a new solution is provided for the intelligent congestion control of low-orbit satellite networks. The solution can not only effectively alleviate network congestion, but also has learning and adaptability, thereby improving the service quality and resource utilization of the network, and laying a foundation for building a more efficient and reliable satellite communication network.

[0008] The present invention discloses a low-orbit satellite congestion control method based on an artificial immune algorithm, comprising:

[0009] Step 1: Map the satellites of the low-orbit constellation to virtual grid coordinates through gridding;

[0010] Step 2: Based on the constructed grid framework, the satellite node periodically monitors the occupancy rate of the output queues in each direction; when congestion is detected, the satellite node generates a congestion warning packet and sends it to the directly adjacent nodes;

[0011] Step 3: The neighboring node that receives the congestion warning packet analyzes the affected destination node set according to its own forwarding needs, and searches for matching items of the affected destination nodes in the historical memory cell library by the similarity between the current scene and the historical record; when a memory cell with a similarity exceeding the threshold is found, the alternative path and diversion ratio stored in it are directly applied as the routing strategy of the affected node; otherwise, the antibody generation process of step 4 is triggered;

[0012] Step 4: Based on the congestion information obtained in step 3, the satellite first analyzes the directional characteristics of the congested link and determines the optimal detour strategy in combination with the location information in step 1. After determining the alternative path, the satellite sets the initial diversion threshold and generates a random number between [0, 1] for each data packet to be forwarded to make a diversion decision, thereby achieving balanced distribution of traffic.

[0013] Step 5: For the control strategy generated in step 4, the satellite performs a comprehensive evaluation in the time window [t0, t0+Δt]. When the affinity continues to exceed the threshold, the strategy information is passed to step 7 for storage; otherwise, it goes to step 6 for optimization.

[0014] Step 6: For the strategy with insufficient affinity in step 5, the satellite continuously monitors the queue occupancy rate change trend. When it is found that there is an upward trend for multiple consecutive cycles, the diversion ratio is increased; the optimized strategy re-enters step 5 for evaluation until the expected performance requirements are met;

[0015] Step 7: The satellite stores the high-performance strategy from step 5 as effective antibodies in the form of quads into the memory cell library.

[0016] As a further improvement of the present invention, the step 1 specifically comprises:

[0017] Each satellite node is mapped to a virtual grid coordinate (x, y) according to its orbital plane number i and satellite number j through the following mapping relationship; where:

[0018] x=i(i∈[0,N-1])

[0019] y=(j+i×δ)mod M(j∈[0,M-1])

[0020] Where δ is the phase offset between adjacent orbital planes. The low-orbit constellation consists of N orbital planes, and M satellites are evenly distributed on each orbital plane.

[0021] As a further improvement of the present invention, in step 2, three levels of congestion states are set: idle state: occupancy rate of output queue ρ<50%, mild congestion: 50%≤ρ<75% and severe congestion: ρ≥75%.

[0022] As a further improvement of the present invention, in step 3, the similarity between the current scene and the historical record is calculated by the Jaccard coefficient:

[0023]

[0024] Where D is the set of affected nodes stored in the historical memory cell, and D′ is the affected destination node;

[0025] When a memory cell with a similarity exceeding the threshold θ is found, its stored alternative pathways and diversion ratios are directly applied; otherwise, the antibody generation process in step 4 is triggered.

[0026] As a further improvement of the present invention, in step 4, the method for determining the optimal detour strategy includes:

[0027] For the east-west congested link, the satellite compares the row coordinates of the current node and the target node: when the target node is located above, it chooses to bypass to the north, and when it is located below, it chooses to bypass to the south;

[0028] For north-south congested links, the route is chosen to go east or west based on the column coordinate difference.

[0029] As a further improvement of the present invention, in step 5, the affinity calculation formula is:

[0030] Affinity(Ab)=w1×ΔρR+w2×ηT+w3×(1-ρmax)

[0031] In the formula, ΔρR=(ρ(t0)-ρ(t0+Δt)) / ρ(t0) represents the relative reduction degree of queue occupancy, ηT=T(t0+Δt) / T(t0) represents the throughput maintenance rate, ρmax represents the maximum queue occupancy in the window, and the weight coefficient satisfies w1+w2+w3=1.

[0032] As a further improvement of the present invention, in step 6, when it is found that there is an upward trend for three consecutive cycles, the diversion ratio p is adjusted to p_new=p_current+5%.

[0033] As a further improvement of the present invention, in step 6, in order to ensure network stability, the satellite sets upper and lower limit thresholds of the diversion ratio: the lower limit of 20% ensures the basic utilization of the main path, and the upper limit of 80% prevents the alternative path from being overloaded; when the diversion ratio reaches the threshold, the satellite locks the current configuration for at least 100 time units to avoid network shocks caused by frequent adjustments.

[0034] As a further improvement of the present invention, in step 7, the satellite stores the high-performance strategy from step 5 in the memory cell library in the form of a four-tuple MC=(L, A, S, D), wherein L is the congested link identifier, A is the set of alternative paths determined in step 4, S is the final optimized diversion ratio, and D is the set of affected destination nodes obtained by the analysis in step 3.

[0035] As a further improvement of the present invention, in step 7, the satellite performs a cleanup every preset time unit, evaluates the usage frequency of memory cells, and removes memory cells with zero usage times in the most recent evaluation cycle; when a new congestion scenario occurs, the satellite preferentially calculates the Jaccard similarity with the stored memory cells, and when a memory cell with a similarity exceeding a threshold is found, the control strategy stored therein is directly reused to achieve a rapid response to repetitive congestion.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. Based on the grid routing framework, the present invention adopts a virtual grid mapping strategy to simplify the complex physical topology into a regular structure, significantly reducing the complexity of routing calculations. In particular, in dealing with the problem of unstable links in polar regions, efficient and reliable path selection is achieved through a location awareness mechanism.

[0038] 2. The present invention is similar to the working mechanism of biological immune system, but different from the traditional immune algorithm, the present invention adopts a lightweight control mechanism, which significantly reduces the control overhead through local control and directional propagation. At the same time, the optimized progressive adjustment strategy avoids drastic network fluctuations, making the solution more suitable for deployment on satellite nodes with limited resources.

[0039] 3. The present invention adds adaptive learning capabilities to the satellite network, but the process is transparent to other parts of the network, does not require changes to the existing network architecture, and has good deployability. Through the memory cell mechanism, the system can quickly learn from historical experience, significantly improving the response speed to repetitive congestion.

[0040] 4. Existing satellite network congestion control strategies mostly adopt complete bypass solutions, which are simple to implement but often lead to serious waste of network resources. In contrast, the present invention dynamically adjusts traffic distribution through an intelligent diversion mechanism, which not only ensures the rational use of congested links, but also avoids network oscillation. Especially when dealing with burst traffic, the progressive adjustment strategy can better maintain network stability.

[0041] In summary, the present invention has obvious advantages over the prior art in terms of routing efficiency, resource utilization, network stability, control overhead, etc., and provides an innovative solution for the intelligent management of large-scale satellite networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The present invention is a flowchart of a low-orbit satellite congestion control method based on an artificial immune algorithm.

[0043] Figure 2 Schematic diagram of satellite congestion link in the present invention.

[0044] Figure 3 The figure is a schematic diagram of the diversion process of the congested link by the artificial immune algorithm in the present invention.

[0045] Figure 4 The figure is a flowchart of the antibody production in the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0048] The present invention provides a low-orbit satellite congestion control method based on an artificial immune algorithm, which can effectively solve the performance bottleneck problem faced by traditional satellite network congestion control by establishing a grid routing framework and combining it with a biological immune mechanism, thereby improving the congestion control response speed and enhancing the service quality and resource utilization of the network.

[0049] Existing satellite network congestion control methods have many limitations: first, traditional methods mostly adopt passive response strategies and lack the ability to predict congestion, resulting in delayed response; second, the current complete traffic switching solution is too simple and crude, and suddenly transferring all traffic to the backup path can easily cause network shocks and chain reactions; third, existing methods often do not fully consider the special regional effects in satellite networks, such as the orbital convergence characteristics of polar regions; finally, these methods lack the necessary learning ability and cannot summarize experience from historical congestion events. Even when faced with similar congestion scenarios, the complete decision-making process needs to be re-executed.

[0050] In the solution of the present invention, an efficient grid mapping mechanism is first designed based on the physical characteristics of the satellite constellation. Since the LEO satellite constellation is usually deployed in polar orbits, it consists of multiple orbital planes, and a number of satellites are evenly distributed on each orbital plane, forming a regular network structure. By mapping this physical topology into a virtual grid structure, each satellite node can quickly determine its position in the grid based on its orbital plane number and satellite number, thereby significantly simplifying the routing calculation process.

[0051] When the system detects that the queue occupancy rate of a link exceeds the preset threshold, it will be considered to be congested. At this time, the congested node will generate a congestion warning package (CWP) containing the congested link identification and status information, and send it to the neighboring nodes as an antigen to be identified. The neighboring node that receives the CWP will first analyze the affected set of destination nodes, and then look for similar historical congestion scenarios in the memory cell library. Through the similarity calculation based on the Jaccard coefficient, if a memory cell with a similarity exceeding the threshold is found, the node can directly reuse the stored control strategy to achieve a rapid response to congestion.

[0052] For newly emerged congestion patterns, the system triggers the antibody generation process. Based on the grid structure and the location of the target node, the system first determines the appropriate detour direction and generates a set of alternative paths. Then, the data packets are diverted through a random decision-making mechanism, which not only ensures the priority use of the main path, but also achieves balanced distribution of traffic. During the control process, the system will continuously evaluate the effectiveness of the strategy. When the affinity does not meet expectations, the control strategy will be optimized through progressive parameter adjustments. This closed-loop immune response mechanism not only ensures the real-time and effectiveness of congestion control, but also improves the system's adaptability through continuous learning and optimization.

[0053] The main contents of the present invention include:

[0054] 1. Grid routing and geographic location awareness mechanism: An efficient mapping scheme based on physical network to virtual grid is constructed, which maps the orbit number of satellite nodes to two-dimensional grid coordinates, and significantly simplifies routing calculations. Through the dynamic location mapping algorithm, the system can calculate the geographic location of satellite nodes in real time and accurately identify special areas such as polar convergence areas. Based on this, an adaptive path selection strategy that takes geographic location into consideration is designed. By giving priority to links pointing to the equator, the transmission problems caused by unstable links in polar regions are effectively avoided.

[0055] 2. Congestion control strategy based on artificial immune algorithm: The biological immune system mechanism is introduced to design a complete congestion control framework. Through the multi-level congestion state identification mechanism based on queue occupancy, the system can accurately quantify the congestion level of the link and notify relevant nodes in a timely manner. The congestion state is used as the antigen and the control strategy is used as the antibody. The random decision-making mechanism is used to dynamically adjust the traffic distribution ratio, avoiding the network oscillation caused by the complete detour in the traditional solution. At the same time, a progressive optimization mechanism based on the queue occupancy change trend is designed to realize the dynamic adjustment of the strategy.

[0056] 3. Immune memory and learning mechanism: A complete immune memory mechanism is designed to store key information such as congested link identification, alternative paths, diversion ratio and impact range through memory cell structure. The similarity calculation method based on Jaccard coefficient can quickly identify similar congestion scenarios and realize efficient reuse of historical experience. The control strategy is continuously evaluated and optimized through the dynamic update mechanism, and the storage management strategy based on frequency of use and success rate is combined to ensure the efficient use of memory resources and significantly improve the system response speed.

[0057] The low-orbit satellite network congestion control method based on artificial immune algorithm proposed in the present invention has significant advantages in the field of modern satellite communications, can effectively improve the service quality and resource utilization efficiency of the network, and is suitable for multiple application scenarios such as the new generation of low-orbit satellite constellation systems, space backbone networks, and ground-ground integrated networks. The application of this method is expected to achieve wide application and commercial value in the field of on-board networks.

[0058] Specific:

[0059] like Figure 1 As shown, the present invention provides a low-orbit satellite congestion control method based on an artificial immune algorithm, comprising:

[0060] Step 1: Grid routing mapping:

[0061] Based on the physical characteristics of the LEO satellite constellation, a virtual grid structure is constructed. Since the LEO constellation is usually deployed in polar orbits, it consists of N orbital planes, and M satellites are evenly distributed on each orbital plane, forming a regular network structure. Each satellite node is mapped to the virtual grid coordinate (x, y) according to its orbital plane number i and satellite number j through the following mapping relationship; where,

[0062] x=i(i∈[0,N-1])

[0063] y=(j+i×δ)mod M(j∈[0,M-1])

[0064] Where δ is the phase offset between adjacent orbital planes.

[0065] At the same time, the satellite node calculates its geographic location (longitude σs, latitude φs) in real time through a dynamic location mapping algorithm, and determines whether it is in a high-latitude area (|φ|>65°) based on the location information. When located in a high-latitude area, the satellite will give priority to the link pointing to the equator for data forwarding to avoid the instability of the link in the polar region.

[0066] Step 2: Antigen recognition and response:

[0067] Based on the grid routing framework constructed in step 1, the satellite node periodically monitors the occupancy rate ρi(t) of the output queues in each direction and sets three levels of congestion states: idle state (ρ<50%), mild congestion (50%≤ρ<75%) and severe congestion (ρ≥75%). When congestion is detected (reaching mild congestion or severe congestion), the satellite node generates a congestion warning packet (CWP), which contains a link identifier (Link_ID) and a link state (Link_State); and is only sent to directly adjacent nodes.

[0068] Step 3: Memory cell matching:

[0069] The neighboring node that receives the CWP first analyzes the affected destination node set D' according to its own forwarding requirements, then searches for matching items in the memory cell library, and calculates the similarity between the current scenario and the historical record using the Jaccard coefficient:

[0070]

[0071] Where D is the set of affected nodes stored in the historical memory cells. When a memory cell with a similarity exceeding a threshold value θ (such as 0.8) is found, the stored alternative paths and diversion ratios are directly applied; otherwise, the antibody generation process in step 4 is triggered.

[0072] Step 4: Antibody Generation

[0073] Based on the congestion information obtained in step 3, the satellite first analyzes the directional characteristics of the congested link, and combines the location information in step 1 to determine the optimal bypass strategy; for the horizontal (east-west) congested link, the satellite compares the row coordinates of the current node and the target node: when the target node is located above, it chooses to bypass north, and when it is located below, it chooses to bypass south; for the vertical (north-south) congested link, it chooses to bypass east or west according to the difference in column coordinates; after determining the alternative path, the satellite sets the initial diversion threshold of 0.7, that is, 70% of the traffic remains on the main path, and 30% of the traffic is forwarded through the alternative path; for each data packet to be forwarded, a random number between [0,1] is generated to make a diversion decision to achieve balanced distribution of traffic.

[0074] Step 5: Affinity calculation and evaluation:

[0075] For the control strategy generated in step 4, the satellite performs a comprehensive evaluation in the time window [t0, t0+Δt]. When the affinity continues to exceed the threshold (such as 0.75), the strategy information is passed to step 7 for storage; otherwise, it goes to step 6 for optimization. The affinity calculation formula is:

[0076] Affinity(Ab)=w1×ΔρR+w2×ηT+w3×(1-ρmax)

[0077] In the formula, ΔρR=(ρ(t0)-ρ(t0+Δt)) / ρ(t0) represents the relative reduction degree of queue occupancy, ηT=T(t0+Δt) / T(t0) represents the throughput maintenance rate, ρmax represents the maximum queue occupancy in the window, and the weight coefficient satisfies w1+w2+w3=1.

[0078] Step 6. Antibody optimization:

[0079] For the strategy with insufficient affinity in step 5, the satellite adopts a progressive optimization scheme; by continuously monitoring the trend of queue occupancy, when it is found that there is an upward trend for three consecutive cycles, the diversion ratio p is adjusted to p_new = p_current + 5%; to ensure network stability, the satellite sets upper and lower thresholds for the diversion ratio: the lower limit of 20% ensures the basic utilization of the main path, and the upper limit of 80% prevents the alternative path from being overloaded. When the diversion ratio reaches the threshold, the satellite locks the current configuration for at least 100 time units to avoid network shock caused by frequent adjustments. The optimized strategy re-enters step 5 for evaluation until the expected performance requirements are met.

[0080] Step 7: Memory cell update:

[0081] The satellite stores the high-performance strategy from step 5 in the memory cell library in the form of a four-tuple MC = (L, A, S, D), where L is the congested link identifier, A is the set of alternative paths determined in step 4, S is the final optimized diversion ratio, and D is the set of affected destination nodes analyzed in step 3; the satellite cleans up every 1000 time units, evaluates the usage frequency of memory cells, and removes memory cells with zero usage in the most recent evaluation cycle. When a new congestion scenario occurs, the satellite first calculates the Jaccard similarity with the stored memory cells. When a memory cell with a similarity greater than 0.8 is found, the control strategy stored in it is directly reused to achieve a rapid response to repetitive congestion.

[0082] Example:

[0083] The present invention provides a low-orbit satellite congestion control method based on an artificial immune algorithm, comprising:

[0084] S1. Grid routing mapping:

[0085] Based on the physical characteristics of the LEO satellite constellation, a virtual grid structure is constructed. Since the LEO constellation is usually deployed in polar orbits, it consists of N orbital planes, and M satellites are evenly distributed on each orbital plane, forming a regular network structure. Each satellite node is mapped to the virtual grid coordinate (x, y) according to its orbital plane number i and satellite number j through the following mapping relationship; where,

[0086] x=i(i∈[0,N-1])

[0087] y=(j+i×δ)mod M(j∈[0,M-1])

[0088] Where δ is the phase offset between adjacent orbital planes.

[0089] At the same time, the satellite node calculates its geographic location (longitude σs, latitude φs) in real time through a dynamic location mapping algorithm, and determines whether it is in a high-latitude area (|φ|>65°) based on the location information. When located in a high-latitude area, the satellite will give priority to the link pointing to the equator for data forwarding to avoid the instability of the link in the polar region.

[0090] S2. Antigen recognition and response:

[0091] like Figure 2As shown in the figure, when satellite node S(4,5) finds through periodic monitoring that the occupancy rate of its westbound output queue continues to rise and reaches 80% at time t1, exceeding the severe congestion threshold (75%), the congestion response mechanism is immediately triggered. At this time, S(4,5) generates a congestion warning packet (CWP) containing the congested link identifier "S(4,5)-S(4,4)" and the link state "CONGESTED", and sends the CWP to its three direct neighboring nodes: northbound neighbor S(3,5), eastbound neighbor S(4,6) and southbound neighbor S(5,5). Based on the directional propagation mechanism, these neighboring nodes will not continue to spread the information to other nodes after receiving the CWP, effectively avoiding unnecessary control overhead.

[0092] S3. Memory cell matching:

[0093] The neighboring nodes that receive the CWP immediately begin to process the congestion. Taking the S(3,5) node as an example, the node first checks all the current data packets that need to be forwarded through the S(4,5)-S(4,4) link, and determines the affected destination node set D'={S(4,3),S(4,2),S(3,3),S(3,2)} through analysis. Subsequently, S(3,5) searches for similar historical congestion scenarios in its memory cell library and finds a memory cell instance that contains a congested link identifier, an alternative path set {S(3,4)}, a main path traffic ratio of 0.7, and a historical affected node set {S(4,3),S(4,2),S(3,3)}. By calculating the Jaccard similarity, it is 0.85, which exceeds the preset threshold of 0.8. Therefore, if Figure 3 As shown, S(3,5) directly reuses the control strategy stored in the memory cell, sets S(3,4) as the alternative next hop, and diverts traffic at a ratio of 70% / 30%. The dotted arrow in the figure indicates the diversion direction.

[0094] S4. Antibody production:

[0095] Taking another neighbor node S(4,6) as an example, since the satellite failed to find historical records with a similarity exceeding the threshold in the memory cell library, a new control strategy needs to be generated. First, the satellite analyzed the characteristics of the congested link S(4,5)-S(4,4), confirmed that it was a westbound link, and identified that the affected destination node set {S(4,2), S(4,1)} were all located at the same level as the current node. Based on the grid routing rule, the satellite selected an alternative path that detoured north: S(4,6)-S(3,6)-S(3,5)-S(3,4). At the same time, an initial diversion threshold of 0.8 was set, and a random decision-making mechanism was used for traffic distribution: when the random number r≤0.8, the data packet is forwarded through the main path (via S(4,5)); when r>0.8, it is forwarded through the alternative path (via S(3,6)). The entire antibody generation process is as follows. Figure 4 shown.

[0096] S5. Affinity evaluation:

[0097] After determining the control strategy, S(4,6) comprehensively evaluates the effect of the strategy within the time window [t0, t0+100]. The satellite recorded detailed changes in performance indicators: the queue occupancy rate dropped from the initial 80% to 65%, but a peak of 82% appeared at t0+10; the throughput dropped from the initial 100Mbps to 85Mbps. Substituting these data into the affinity calculation formula Affinity(Ab)=w1×ΔρR+w2×ηT+w3×(1-ρmax), where the weight coefficients are set to w1=0.5 and w2=w3=0.25, the calculated affinity is 0.35. Since this value is lower than the preset threshold of 0.75, the satellite immediately enters the antibody optimization stage.

[0098] S6. Antibody Optimization:

[0099] In response to the lower affinity, the satellite optimized the control strategy for the S(4,6) node. Through analysis, it was found that the queue occupancy rate showed an upward trend of 78%, 79%, and 80% in three consecutive detection cycles. The satellite adopted a progressive optimization strategy to reduce the diversion ratio of the main path from 80% to 75%. In the new round of evaluation cycle [t1, t1+100], all performance indicators have been significantly improved: the queue occupancy rate dropped to 58%, the throughput was maintained at 90Mbps, and the maximum queue occupancy rate was only 75%. The recalculated affinity increased to 0.81, exceeding the preset threshold, indicating that the optimized strategy has achieved the expected effect.

[0100] S7, memory cell update:

[0101] Finally, the S(4,6) node stores the optimized control strategy in the database in the form of a memory cell, which fully records the congested link identifier "S(4,5)-S(4,4)", the alternative path set {S(3,6)-S(3,5)-S(3,4)}, the optimized diversion ratio 0.75, and the affected destination node set {S(4,2), S(4,1)}. At the same time, the statistical information of the usage frequency (1 time) is initialized. This new memory cell will participate in the evaluation within the next 1000 time units. If it is not used during this period, it will be removed from the memory bank during the next cleanup, thereby ensuring the efficient use of memory resources.

[0102] The present invention is a congestion control strategy for low-orbit satellite networks based on an artificial immune algorithm. In this strategy, the physical topology of satellite nodes is mapped into a virtual grid structure. When congestion is detected, the system transmits the congestion information as an antigen to neighboring nodes. Through the memory cell matching mechanism, the system can quickly reuse historical successful experiences; for new congestion scenarios, the diversion strategy is dynamically adjusted through antibody generation and optimization processes. This adaptive control scheme based on the immune mechanism not only ensures a rapid response to congestion, but also improves network performance through continuous learning and optimization, effectively solving the congestion control problem in satellite networks.

[0103] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A low-orbit satellite congestion control method based on artificial immune algorithm, characterized in that: include: Step 1: Map the satellites of the low-orbit constellation to virtual grid coordinates through gridding; Step 2: Based on the constructed grid framework, the satellite node periodically monitors the occupancy rate of the output queues in each direction; when congestion is detected, the satellite node generates a congestion warning packet and sends it to the directly adjacent nodes; Step 3: The neighboring node that receives the congestion warning packet analyzes the affected destination node set according to its own forwarding requirements, and searches for matching items of the affected destination nodes in the historical memory cell library based on the similarity between the current scenario and the historical records; when a memory cell with a similarity exceeding the threshold is found, the alternative path and diversion ratio stored in it are directly applied as the route of the affected node; Otherwise, the antibody production process in step 4 is triggered; Step 4: Based on the congestion information obtained in step 3, the satellite first analyzes the directional characteristics of the congested link and determines the optimal detour strategy in combination with the position information in step 1; After determining the alternative path, the satellite sets the initial diversion threshold and generates a random number between [0,1] for each data packet to be forwarded to make a diversion decision, thus achieving balanced distribution of traffic. This routing strategy is an antibody against the antigen. Step 5: For the control strategy generated in step 4, the satellite performs a comprehensive evaluation in the time window [t0, t0+Δt]; the affinity is calculated based on the queue occupancy rate. When the affinity continues to exceed the threshold, the strategy information is passed to step 7 for storage; otherwise, it goes to step 6 for optimization; Step 6: For the strategy with insufficient affinity in step 5, the satellite continuously monitors the queue occupancy rate change trend. When it is found that there is an upward trend for multiple consecutive cycles, the diversion ratio is increased; the optimized strategy re-enters step 5 for evaluation until the expected performance requirements are met; Step 7: The satellite stores the high-performance strategy from step 5 as effective antibodies in the form of quads into the memory cell library.

2. The low-orbit satellite congestion control method according to claim 1, characterized in that: The step 1 specifically includes: Each satellite node is mapped to a virtual grid coordinate (x, y) according to its orbital plane number i and satellite number j through the following mapping relationship; where: x=i(i∈[0,N-1]) y=(j+i×δ)mod M(j∈[0,M-1]) Where δ is the phase offset between adjacent orbital planes. The low-orbit constellation consists of N orbital planes, and M satellites are evenly distributed on each orbital plane.

3. The low-orbit satellite congestion control method according to claim 1, characterized in that: In step 2, three levels of congestion states are set: idle state: occupancy rate of output queue ρ<50%, mild congestion: 50%≤ρ<75% and severe congestion: ρ≥75%.

4. The low-orbit satellite congestion control method according to claim 1, characterized in that: In step 3, the similarity between the current scene and the historical record is calculated using the Jaccard coefficient: Where D is the set of affected nodes stored in the historical memory cell, and D′ is the affected destination node; When a memory cell with a similarity exceeding the threshold θ is found, its stored alternative paths and diversion ratios are directly applied; Otherwise, the antibody production process in step 4 is triggered.

5. The low-orbit satellite congestion control method according to claim 1, characterized in that: In step 4, the method for determining the optimal detour strategy includes: For the east-west congested link, the satellite compares the row coordinates of the current node and the target node: when the target node is located above, it chooses to bypass to the north, and when it is located below, it chooses to bypass to the south; For north-south congested links, the route is chosen to go east or west based on the column coordinate difference.

6. The low-orbit satellite congestion control method according to claim 1, characterized in that: In step 5, the affinity calculation formula is: Affinity(Ab)=w1×ΔρR+w2×ηT+w3×(1-ρmax) In the formula, ΔρR=(ρ(t0)-ρ(t0+Δt)) / ρ(t0) represents the relative reduction degree of queue occupancy, ηT=T(t0+Δt) / T(t0) represents the throughput maintenance rate, ρmax represents the maximum queue occupancy in the window, and the weight coefficient satisfies w1+w2+w3=1.

7. The low-orbit satellite congestion control method according to claim 1, characterized in that: In step 6, when it is found that there is an upward trend for three consecutive cycles, the diversion ratio p is adjusted to p_new=p_current+5%.

8. The low-orbit satellite congestion control method according to claim 7, characterized in that: In step 6, to ensure network stability, the satellite sets upper and lower limit thresholds for the diversion ratio: the lower limit of 20% ensures the basic utilization of the main path, and the upper limit of 80% prevents the alternative path from being overloaded; when the diversion ratio reaches the threshold, the satellite locks the current configuration for at least 100 time units to avoid network shocks caused by frequent adjustments.

9. The low-orbit satellite congestion control method according to claim 1, characterized in that: In step 7, the satellite stores the high-performance strategy from step 5 as an effective antibody in the memory cell library in the form of a four-tuple MC=(L, A, S, D), where L is the congested link identifier, A is the set of alternative paths determined in step 4, S is the final optimized diversion ratio, and D is the set of affected destination nodes obtained by analysis in step 3.

10. The low-orbit satellite congestion control method according to claim 1, characterized in that: In step 7, the satellite performs a cleanup every preset time unit, evaluates the usage frequency of memory cells, and removes memory cells that have been used zero times in the most recent evaluation cycle; when a new congestion scenario occurs, the satellite preferentially calculates the Jaccard similarity with the stored memory cells, and when a memory cell with a similarity exceeding a threshold is found, the control strategy stored therein is directly reused to achieve a rapid response to repetitive congestion.

Images