A large-scale constellation load balancing routing method, system, and medium based on dynamic region segmentation.

Abstract

This invention relates to a large-scale constellation load balancing routing method, system, and medium based on dynamic region segmentation. The method operates on dual time scales, including a long-timeslot-scale dynamic region segmentation phase and a short-timeslot-scale hybrid routing phase. The method includes: in the dynamic region segmentation phase, dynamically adjusting the cluster affiliation of satellite nodes based on their real-time load and historical traffic information using a dynamic region segmentation algorithm; in the hybrid routing phase, the head node of the current cluster selects the next-hop cluster for the data flow in a distributed manner based on the real-time geographical location and load status of neighboring clusters; and the head node of the current cluster constructs low-load routing paths within the cluster for the data flow in a centralized manner based on a residual topology method to maximize system throughput. This invention achieves load balancing routing for data transmission flows in large-scale satellite networks, improving the overall network load balancing effect and the reliability of transmission flows.

Description

Technical Field

[0001] This invention relates to the field of satellite routing technology, and in particular to a large-scale constellation load balancing routing method, system, and medium based on dynamic region segmentation for mega-constellation scenarios. Background Technology

[0002] In recent years, the continuous evolution and widespread application of 5G and 6G mobile communication technologies have significantly improved the transmission efficiency and service capabilities of communication systems, thereby driving the rapid growth of global mobile data traffic and the widespread adoption of diversified access services. To meet the increasing communication demands, terrestrial communication networks have made significant progress in coverage, transmission rates, and system capacity. However, constrained by geographical limitations and deployment costs, terrestrial networks cannot achieve seamless global coverage, indicating their inability to independently meet the future demands for ubiquitous, highly reliable, and multi-service converged communication. Against this backdrop, satellite communication networks, with their advantages of wide-area coverage, flexible networking, and lack of geographical limitations, have become an indispensable component of the 5G Non-Terrestrial Network (NTN) architecture. Especially in remote areas where terrestrial infrastructure is difficult to cover, along air routes, and in open ocean areas, satellite communication plays a crucial role and will assume a key position in the evolution and construction of future 6G mobile communication systems.

[0003] Currently, given the inherent characteristics of Low Earth Orbit (LEO) satellites, designing a high-performance routing algorithm for mega-LEO satellite networks faces a series of critical challenges. The primary challenge stems from the scalability issues arising from network expansion. Specifically, as satellite constellations continue to grow, the number of space nodes increases exponentially, evolving into mega-constellation networks; for example, the Starlink system's planned constellation size is already in the tens of thousands of satellites. This rapid increase in node size directly leads to a significant increase in the complexity of routing calculations and a substantial increase in control signaling overhead, making traditional, fully centralized routing management strategies extremely difficult to implement in practice. Early global routing algorithms designed for small-scale satellite constellations, when applied to such mega-networks, experience a sharp rise in computational complexity and a significant increase in overall network overhead. Furthermore, they struggle to effectively allocate distributed satellite resources within the constellation, resulting in decreased resource utilization.

[0004] Secondly, the highly dynamic nature of giant LEO satellite constellation networks poses a severe challenge to the transmission stability of inter-satellite links (ISL) and the reliable delivery of end-to-end data services. As the number of satellite nodes in the constellation continues to increase, the relative positions and ISL distances between these rapidly moving nodes are constantly and frequently changing. This dynamic change further leads to continuous and rapid shifts in the satellite network topology, exhibiting significant dynamic instability. This high degree of dynamism inherently requires routing algorithms to possess the ability to quickly perceive changes in network topology and to dynamically optimize data transmission paths in an adaptive manner, thereby maintaining the continuity of network services and the overall operational stability of the system.

[0005] Furthermore, due to the significant spatiotemporal differences in service traffic demands across different geographical regions globally, the traffic load in satellite networks exhibits a highly uneven distribution. This makes some critical nodes and core links in the network prone to congestion, ultimately increasing the probability of data loss and severely impacting the quality of user experience for various services. The LEO satellite network aims to provide access services to users worldwide; however, the inherent unevenness of global service distribution results in a similarly uneven spatial distribution of data traffic carried on satellite links. This uneven traffic distribution pattern easily triggers localized network congestion, significantly increasing packet queuing latency and the probability of data packet drop, thereby degrading the overall network performance.

[0006] Currently, routing algorithms for low-Earth orbit (LEO) satellites have received widespread attention in academia, but discussions on mega-constellations remain limited. Existing research on satellite network routing algorithms mostly focuses on small-scale satellite constellations, typically around 100 satellites. However, due to the enormous number of nodes and the significant overhead of link state collection in large-scale satellite networks, routing algorithms for small-scale constellations are difficult to directly apply to mega-constellations. Therefore, there is an urgent need to address the data transmission and load balancing issues in large-scale satellite networks. Summary of the Invention

[0007] This invention provides a large-scale constellation load balancing routing method, system, and medium based on dynamic region segmentation, aiming to achieve load balancing routing of data transmission flows in large-scale satellite networks, thereby improving the overall load balancing effect and the reliability of transmission flows.

[0008] This invention provides a large-scale constellation load balancing routing method based on dynamic region segmentation. The method operates on dual time scales, including a long-timeslot-scale dynamic region segmentation phase and a short-timeslot-scale hybrid routing phase. The method includes the following steps:

[0009] Step S10: In the dynamic region segmentation stage, based on the dynamic region segmentation algorithm, the cluster affiliation of satellite nodes is dynamically adjusted according to their own real-time load and historical traffic information.

[0010] Step S20: In the hybrid routing phase, the head node of the current cluster selects the next-hop cluster for the data flow in a distributed manner based on the real-time geographical location and load status of the neighboring clusters.

[0011] Step S30: The head node of the current cluster constructs low-load routing paths within the cluster for the data flow in a centralized manner based on the residual topology method to maximize system throughput.

[0012] A further technical solution of the present invention is that, prior to step S10, the following is included:

[0013] Step S00 involves constructing a LEO mega-constellation network model, a traffic and load model, a satellite communication model, and an optimization problem for low-load routing paths. The steps in step S00 for constructing the LEO mega-constellation network model include:

[0014] Step S01, using Indicates the LEO satellite constellation. This indicates the total number of satellites in the constellation. This indicates the number of orbits in a constellation, with all orbits having the same altitude. and orbital inclination and The satellites are evenly distributed in On each track, The orbital phase factor is used to determine the phase difference between corresponding satellites in adjacent orbital planes. The phase deviation between satellites in adjacent orbits is denoted as . In a constellation, the set of satellites is denoted as... Each satellite can establish ISLs with four neighboring satellites, including two inter-plane ISLs and two intra-plane ISLs;

[0015] A dual-time-scale method is used to divide the time slots. First, the entire operating cycle is divided into... Discrete time slots This allows for long-term changes in satellite and cluster topology. This represents the set of time slots, and then each time slot... Further subdivided into An evenly spaced hourly gap, denoted as This allows for the relay and forwarding of traffic on a short timescale; therefore, satellite constellation topology uses... It means that, among them, For time slots Internally, the topology of the satellite constellation, For a set of satellite nodes, For time slots Internal satellite link set, and in time slots Inside, satellite and satellite The link established between them is used express;

[0016] Step S02, model the satellite constellation: the satellite constellation set is denoted as... , For time slots Number of internal clusters; clusters In the time slot The number of internal satellite nodes is used Representation; Definition of binary variables This indicates the mapping relationship between satellites and the constellation. Indicates in time slot Inside, satellite Located in the cluster Inside, otherwise Cluster Topology It means that, among them, Indicates time slot Inner Satellite Cluster The topology, Indicates in time slot Internal cluster The set of satellite nodes Indicates in time slot Internal cluster A collection of satellite links;

[0017] Step S03, Model the ground area: Divide the ground to be covered by the satellite into... Each region is defined by latitude and longitude, forming a set of ground regions, denoted as . A ground station is set up at the center of each ground area. The ground station selects the satellite with the closest geographical distance as its access point, and all traffic in that area is uploaded to the satellite network through that satellite, or the traffic is downloaded. It is assumed that when a satellite is over a certain ground area, it can communicate with the ground station in that area. Define binary variables. This indicates the relationship between the ground area and the satellite. Indicates in time slot Inside, ground area Select Satellite As an access satellite, otherwise .

[0018] A further technical solution of the present invention is that, as the basic unit of transmission, the data stream has specific statistical characteristics in its generation and transmission process; the data stream is entirely generated by the ground station and ultimately transmitted to another ground station; the step of constructing the traffic and load model in step S00 includes:

[0019] In each hour gap Inside, ground area The generated traffic Obtain the parameter as Poisson distribution:

[0020] (1)

[0021] in, Represents the Poisson distribution function. It is a natural constant. Represents ground area The traffic arrival rate; taking historical traffic data into consideration as a reference factor for cluster adaptive changes, the following traffic indicators are defined in the mathematical model: First, defined in the current time slot Each hour gap Inside, satellite The incoming traffic is:

[0022] (2)

[0023] in, Represents ground area In the hour gap The internal flow, Represents ground area With satellite Whether connected; in time slot satellite The forwarded traffic is: Therefore, in time slots The total flow through each satellite is:

[0024] (3)

[0025] in, and They represent satellites In the hour gap Traffic accessed internally and traffic forwarded; defined in time slots. The total traffic successfully relayed by each satellite was The result is obtained by formula (4) in the time slot. Inner satellite load level :

[0026] (4)

[0027] in, This represents the sum of the total bandwidth of all satellite payloads within the same time slot. Indicates in time slot Inland flow through satellite Total flow;

[0028] Introducing discount factors Appropriate discounting is applied to the satellite payload in historical time slots to calculate the payload in the current time slot. Within each satellite and cluster, the discount weights reduce the impact of satellite loads farther from the current time slot on cluster changes, while satellite loads closer to the current time slot have a greater impact on cluster changes; within the time slot... Inside, satellite Discount weight for:

[0029] (5)

[0030] in Indicates historical time slots, Indicates historical time slots The weighting coefficient corresponding to the load, Indicates historical time slots Inside, satellite Normalized load level; cluster Discount weight This is obtained by summing the discounted weights of all its internal satellites:

[0031] (6)

[0032] in, Indicates in time slot Inner satellite Discount weighting.

[0033] A further technical solution of the present invention is that the step of constructing the satellite communication model in step S00 includes:

[0034] The satellite's position is described using a spherical coordinate system. The coordinates are represented as ,in Earth's altitude The altitude of the satellite orbit. and Satellites The longitude and latitude; based on these coordinates, the satellite and satellite In the time slot European distance within The result is obtained by calculation using formula (7):

[0035] (7)

[0036] Assuming the presence of additive white Gaussian noise and a symmetric wireless channel, then the link... Free space path loss Represented as:

[0037] (8)

[0038] in Indicates the carrier wavelength. Indicates satellite and satellite In the time slot Euclidean distance within; according to Shannon's theorem, the information transmission rate of the link. It is calculated using formula (9);

[0039] (9)

[0040] in, This indicates the transmission power, measured in watts (W). and These represent the antenna gains of the receiving antenna and the transmitting antenna, respectively. Boltzmann's constant; This indicates the channel bandwidth, measured in MHz. This refers to thermal noise temperature, expressed in Kelvin (K). Represents the logarithmic function with base 2. Indicates in time slot Inner Link Free space path loss;

[0041] Next, establish the data flow. Path model, data flow The transmission path is denoted as: ,in For data stream The source satellite node, For data stream Target satellite node It is a pair of satellites that have established an inter-satellite link and meet the following requirements. ;

[0042] Finally, establish the data flow. The latency model for transmission in satellite networks considers three main types of latency: transmission latency and propagation latency incurred when data streams are transmitted on the link, and queuing latency incurred when data streams are queued for processing on the satellite; in time slots... hour gaps within In the middle, data flow In the link The transmission delay and propagation delay are calculated using formulas (10) and (11), respectively:

[0043] (10)

[0044] (11)

[0045] in, Represents the speed of light. Indicates in time slot Inner Link distance, Indicates the size of each data stream. Indicates in time slot Inner Link Information transmission rate; in hourly slots Internal, data flow In satellite Queuing delay is defined as , indicating data stream The waiting time for transmission; the transmission queue capacity for each satellite is defined as follows. ; in each hour gap Inside, satellite The occupied capacity of the transmission queue is denoted as Data stream transmission follows the first-in, first-out (FIFO) principle; if the satellite... If there are no other data streams waiting to be transmitted in the transmission queue, then the data stream... It can be transmitted out of this queue; conversely, if other data streams are waiting in the queue, it must be transmitted to the satellite. It continues to wait in the transmission queue until it becomes the next item to be processed in the queue; data stream From satellite Forwarded to adjacent satellites latency Given by formula (12):

[0046] (12)

[0047] in, and These represent the time slots. Internal, data flow In the link Transmission delay and propagation delay on the network Indicates a time slot Internal, data flow In satellite Queuing delays; data flow The end-to-end delay, i.e., the delay along its transmission path. The total delay can be expressed as:

[0048] (13)

[0049] Furthermore, the end-to-end latency threshold is set to... Data Stream Forwarding must meet the following conditions: It is carried out under the following conditions.

[0050] A further technical solution of the present invention is that the step of constructing the optimization problem for low-load routing paths in step S00 includes:

[0051] System throughput is measured by end-to-end success rate. As a metric, this metric can be calculated using formula (14):

[0052] (14)

[0053] in, Indicates a time slot Successfully transmitted data streams within the internal network Indicates a time slot Inside, satellite Forwarded traffic; this optimization problem can be described as:

[0054] (15)

[0055] in, For end-to-end successful delivery rate, Represents data stream The routing path, It is a time slot Inner satellite With cluster The mapping relationship between binary decision variables, The relationship between satellites and the constellation is defined, that is, for any single satellite... and any satellite constellation , Indicates in time slot Inside, satellite Located in the cluster Inside, otherwise ; Indicates that it refers to any one satellite In any time slot Within, it can only be located in one satellite constellation. Within, ensure in each time slot In this system, each satellite is assigned to only one cluster; This refers to any satellite constellation. In any time slot The number of satellites contained within must be greater than 1 to ensure that in each time slot... In this system, each cluster has at least one satellite to ensure the effectiveness of the satellite cluster; The capacity of the satellite transmission queue is limited to ensure that in any time slot... hour gaps within In, any satellite The traffic carried by the transmission queue Cannot exceed the transmission queue capacity ; The end-to-end transmission delay of the data stream is limited, meaning that in any time slot... hour gaps within In the middle, data flow end-to-end delay The delay threshold cannot be exceeded. This ensures that all data streams are transmitted within the specified time; optimization issues will be addressed. This can be broken down into two sub-problems: a long-scale cluster dynamic partitioning optimization problem. and load balancing routing issues on short timescales :

[0056] (16)

[0057] (17)

[0058] in, Indicates in time slot Internal Cluster Discount weight, In the time slot Internal Cluster The number of internal satellites, Indicates in time slot Inner satellite Traffic that was successfully forwarded.

[0059] A further technical solution of the present invention is that step S10 includes:

[0060] Step S101, using satellite cluster assembly This represents the alliance partitioning strategy. According to alliance game theory, for a cluster... There are three indicators:

[0061] 1) Profit The value of the alliance was quantified, and , This represents an empty set of satellites, for satellite constellations. The profit is defined as:

[0062] (18)

[0063] in, Indicates in time slot Internal Cluster Discount weight, In the time slot Internal Cluster The number of internal satellites, Indicates in time slot Inner satellite Traffic successfully forwarded;

[0064] 2) Cost The cost of quantifying cooperation is defined as follows:

[0065] (19)

[0066] There is no cost when cluster partitioning satisfies both hard constraints: (1) That is, the single-satellite single-cluster affiliation constraint, where For time slots Internal satellite constellation collection, It is a time slot Inner satellite With cluster (2) Binary decision variables for attribution relationships; That is, the cluster non-empty constraint, where It must be the set of all satellite nodes in the network; otherwise, the cost is positive infinity.

[0067] 3) Value The difference between the benefits gained from cooperation and the costs of cooperation is quantified, and this indicator is calculated according to formula (20):

[0068] (20)

[0069] Step S102: Define the set of satellites where cluster handover occurs. From any satellite node and in the same cluster It consists of adjacent satellites within;

[0070] Step S103: Define a possible switching operation. For satellite collection From the original cluster Switch to a new cluster The process, namely:

[0071] (twenty one)

[0072] Among them, if the satellite set Switch to This indicates that the satellite set has established a new satellite cluster, resulting in an increase in the number of clusters in the network; if the satellite set It is a cluster If the cluster has only one set of satellites, then after this handover operation, the cluster... Will be merged into the cluster This leads to a reduction in the number of clusters in the network;

[0073] Step S104: Define the execution of the switching operation. The switching gain that can be brought for:

[0074] (twenty two)

[0075] in and Representing satellite sets Remove from cluster Before and after, cluster Value; and Representing satellite sets Enter the cluster Before and after, cluster Value;

[0076] If satellite set If multiple switching operations are available, the operation with the highest switching gain is selected for execution; the switching preference is defined as follows:

[0077] (twenty three)

[0078] That is, if the switching operation The switching gain is higher than The switching gain indicates the switching operation. Compare High priority;

[0079] Initially, in the cluster Inside, satellite Together with its neighboring satellites in the same cluster, they form a satellite ensemble. Subsequently, the satellite assembly was examined. Neighboring clusters Perform a search, if cluster With satellite collection If there is a link connection between them, they are considered adjacent; a handover operation is established for these adjacent clusters. And calculate the corresponding switching gain; if the switching gain If the number is a finite number of positive real numbers, the switching operation is considered valid, and the switching operation is performed. Add to the candidate operation set In the middle; when the candidate operation set When not empty, select the cluster with the highest switching gain. As a collection of satellites New affiliation; in the cluster With cluster Under the premise that no handover operation is in progress, satellite assembly The cluster switchover will be performed; otherwise, the switchover operation will be discarded; in each time slot Initially, clusters in the network will be dynamically split, a process that continues until no handover operation with positive gain can be found.

[0080] A further technical solution of the present invention is that step S20 includes:

[0081] When data stream Enter the current cluster At that time, cluster The control node is responsible for finding the exit node within the cluster for the data stream; there are two ways for the data stream to enter the current cluster: one is to upload from the ground station to the current cluster via satellite access; the other is to forward it from a neighboring cluster to the entry node of the current cluster; the control node will determine the data stream... destination node Is it located in the cluster? Inside; if the destination node Belongs to cluster Then the node Set as a data stream In the cluster Export node in If the destination node Not in cluster Within the cluster, the control node needs to execute an inter-cluster routing algorithm to route data from the current cluster. Neighbor cluster set Select the next-hop cluster for the data stream And determine in the current cluster Internal exit node and the entry node within the next-hop cluster In addition, data flow The current location is a cluster. Entry node ;

[0082] In the process of selecting the next-hop cluster, a weighted inter-cluster routing metric must first be calculated according to formula (24). This metric combines load and distance factors:

[0083] (twenty four)

[0084] in, Indicates the current cluster Neighbor clusters To the destination cluster The normalized distance; this normalized distance is calculated by taking the actual distance... Divide by cluster All Neighbor Clusters To the destination cluster maximum distance To obtain, Indicates a time slot Inner satellite The size of the transmission queue that can handle the traffic. The transmission queue capacity for each satellite; The weighting coefficients represent the cluster load balancing factor and the distance weighting factor, respectively; the cluster with the smallest routing index is selected from the neighboring clusters. This cluster will serve as a data stream. The next hop cluster; finally, if the cluster With cluster If multiple links exist, then choose the one that makes... The shortest link is used as the transmission link between clusters. Represents data stream From satellite node Transmitted to satellite node Single-hop transmission delay; node Will be used as data stream In the cluster The exit node within the node Then as a data stream In the cluster The entry node within.

[0085] A further technical solution of the present invention is that step S30 includes:

[0086] Step S301, Network State Awareness and Residual Topology Construction: Dynamically construct the current cluster based on real-time load status. The residual topology; specifically, it includes setting the maximum load factor of the satellite as... , In the current hourly slot Internally, for clusters Each satellite node within If its transmission queue is overloaded Then the satellite and the ISL slave cluster directly connected to it Remove it from the topology; then, assess the connectivity of the cluster topology; if topology connectivity is found, set the payload of that satellite to an unattainable upper limit. and rejoin it to the cluster. In the topology, thus obtaining the cluster In the hour gap The residual topology;

[0087] Step S302, Candidate Path Generation: Apply a depth-first traversal algorithm to the residual topology to find slave nodes. To the node All candidate routing paths are defined, and these paths are sorted in ascending order of hop count; For data stream In the cluster The set of routing paths within, sorted in ascending order of hop count. and For data stream In the cluster Internal transmission of the first and Candidate routing paths; set a threshold for the number of routing paths within the cluster. If the number of paths exceeds Then only the routes with low similarity and few hops are retained. Select one path; otherwise, retain all paths. The route similarity calculation method is as follows: Set the route similarity threshold to... If the path has fewer hops With path The common edge count accounts for The proportion of the number of sides exceeds ,Right now

[0088] (25)

[0089] Therefore, the two paths are considered to be too similar, and the path with more hops should be discarded. ,in This represents a satellite pair in any candidate path;

[0090] Step S303, Path Selection Decision: For the set of candidate routes, establish a comprehensive utility index within the cluster to plan the final transmission route within the cluster, and define... Characterize each path within the cluster The combined utility value across two dimensions: load level and transmission latency; It can be calculated using the following formula:

[0091] (26)

[0092] in, Indicate candidate path The satellite nodes that pass through, Indicate candidate path Links traversed ; and Data streams In the cluster Maximum path load and longest path delay among all paths within the network. The weighting coefficients represent the load balancing factor and latency weighting factor within the cluster, respectively. Normalization is used to eliminate dimensional differences and ensure comparability of the indicators. Finally, the following selections are made. The shortest path is used as the data flow. In the cluster The final route path within.

[0093] To achieve the above objectives, the present invention also proposes a large-scale constellation load balancing routing system based on dynamic region segmentation. The system includes a memory, a processor, and a large-scale constellation load balancing routing program based on dynamic region segmentation stored in the memory. The large-scale constellation load balancing routing program based on dynamic region segmentation is executed by the processor to perform the steps of the method described above.

[0094] To achieve the above objectives, the present invention also proposes a computer-readable storage medium storing a large-scale constellation load balancing routing program based on dynamic region segmentation, wherein the large-scale constellation load balancing routing program based on dynamic region segmentation is executed by a processor to perform the steps of the method described above.

[0095] The beneficial effects of this invention on the large-scale constellation load balancing routing method, system, and medium based on dynamic region segmentation are:

[0096] (1) Dynamic region segmentation is achieved through alliance game theory, and historical traffic characteristics are introduced into the region segmentation process. Satellites are allowed to adaptively adjust cluster affiliation according to real-time load, which overcomes the limitation that static region segmentation cannot adapt to the spatiotemporal changes of traffic. It realizes the coordinated optimization of region segmentation and routing decision, and significantly improves the network's adaptability to dynamic topology.

[0097] (2) The proposed hybrid routing mechanism combines the low signaling overhead of distributed routing with the global optimization capability of centralized routing. It achieves rapid load awareness and forwarding through distributed decision-making between clusters, and achieves fine-grained path planning through centralized control within the cluster. It effectively solves the load balancing problem while ensuring network scalability.

[0098] (3) By using the residual topology method and the backlog decision index in a collaborative design, low-load paths are dynamically constructed within the cluster to actively avoid high-load nodes and congested links. Load status and propagation delay are comprehensively considered between clusters to avoid detours, significantly reducing the probability of network congestion and queue overflow, realizing joint optimization of load and delay, and comprehensively improving transmission reliability and end-to-end delay performance. Attached Figure Description

[0099] Figure 1 This is a flowchart illustrating a preferred embodiment of the large-scale constellation load balancing routing method based on dynamic region segmentation of the present invention.

[0100] Figure 2 This is a schematic diagram of the overall process of the large-scale constellation load balancing routing method based on dynamic region segmentation of the present invention.

[0101] Figure 3 This is a scene diagram of the LEO network system of the present invention;

[0102] Figure 4 This is a diagram of a dynamic cluster;

[0103] Figure 5 This is a schematic diagram of the routing algorithm within the cluster;

[0104] Figure 6 This is a schematic diagram of the hardware architecture of the large-scale constellation load balancing routing system based on dynamic region segmentation of this invention. Detailed Implementation

[0105] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0106] This invention proposes a large-scale constellation load balancing routing method based on dynamic region segmentation. The method operates on dual time scales, including a long-timeslot-scale dynamic region segmentation phase and a short-timeslot-scale hybrid routing phase, such as... Figure 1 and Figure 2 As shown, a preferred embodiment of the present invention includes the following steps:

[0107] Step S10: In the dynamic region segmentation stage, based on the dynamic region segmentation algorithm, the cluster affiliation of satellite nodes is dynamically adjusted according to their own real-time load and historical traffic information.

[0108] This invention first constructs a LEO mega-constellation network model and then a satellite communication model considering actual channel conditions. Based on the mega-constellation's satellite network topology, the LEO mega-constellation network is divided into multiple clusters, each consisting of a cluster head node and multiple cluster member nodes. These clusters adopt a Software-Defined Network (SDN) architecture, divided into a data plane and a control plane. The core of the control plane is the cluster head node, responsible for managing the member nodes within the cluster and making routing decisions for data flows entering the cluster. The data plane consists of all member nodes in the cluster and is primarily responsible for forwarding data flows within the cluster based on the head node's decisions. Satellites within the cluster periodically report their load information to the cluster head node, which then formulates routing decisions for routing events within the cluster based on the collected link information. Simultaneously, cluster head nodes periodically exchange load information between adjacent clusters via ISL. This hybrid routing design, which divides the clusters, allows for centralized control of traffic path planning within the cluster. Based on this, the load balancing routing problem is formulated as an optimization problem that maximizes the system's successful delivery rate while satisfying end-to-end latency constraints, taking into account constraints such as traffic conservation, link capacity, and satellite access constraints.

[0109] This invention designs a dynamic region segmentation algorithm based on coalition game theory, introducing historical traffic characteristics into the satellite cluster partitioning process for the first time. Within this algorithm framework, each satellite acts as a game participant, adaptively adjusting its cluster affiliation based on its real-time load and historical traffic information through coalition formation and switching mechanisms. Specifically, the payoff function for a satellite joining a coalition is defined as the reciprocal of the sum of historical traffic of satellites within the coalition, encouraging satellites to join coalitions with lighter loads; the cost function is defined as a linear function of the coalition size to limit management overhead; and the net payoff function is defined as the difference between payoff and cost. Switching operation rules are designed to allow satellites to switch from their current coalition to an adjacent coalition while maintaining network coverage constraints. Switching is performed only if the net payoff increment from the switching operation is positive, prioritizing the operation with the largest net payoff increment. This algorithm iteratively executes switching operations until the network reaches a stable state (i.e., no switching operations with positive gain exist), thereby achieving adaptive adjustment of cluster size and structure, enhancing the network's adaptability to dynamic topology, and reducing the risk of packet loss due to unreasonable region partitioning.

[0110] In step S20, during the hybrid routing phase, the head node of the current cluster selects the next-hop cluster for the data flow in a distributed manner based on the real-time geographical location and load status of neighboring clusters.

[0111] Step S30: The head node of the current cluster constructs low-load routing paths within the cluster for the data flow in a centralized manner based on the residual topology method to maximize system throughput.

[0112] This invention designs a hybrid routing mechanism based on dynamic region segmentation, integrating distributed inter-cluster routing and centralized intra-cluster routing strategies to achieve coordinated optimization of region segmentation and routing decisions. In the inter-cluster routing phase, a distributed decision-making mechanism is adopted. Based on the real-time geographical location and load status of neighboring clusters, a backlog decision index is introduced to autonomously select the next-hop cluster. This index is defined as the weighted sum of the neighboring cluster load and the estimated propagation delay, where the estimated propagation delay is calculated based on the spatial coordinates between control satellites. By comparing the index values ​​of candidate next-hop clusters, the one with the smallest value is selected as the forwarding direction. This ensures low propagation delay while guiding traffic to clusters with lighter loads, avoiding congestion in hotspot areas. In the intra-cluster routing phase, a centralized control strategy is adopted. The cluster head node dynamically constructs low-load paths based on the residual topology method. By setting a load threshold, high-load nodes and congested links within the cluster are eliminated to form a residual network. Based on this, a set of candidate paths from the ingress node to the egress node is calculated. Finally, the path that satisfies the end-to-end delay constraint and has the smallest load on the largest node on the path is selected as the final routing path, actively avoiding high-load nodes and achieving efficient multi-path traffic scheduling. This mechanism overcomes the limitations of traditional routing algorithms in terms of computational complexity and susceptibility to local optima by coordinating and optimizing region division and routing decisions, and significantly improves network load balancing performance while ensuring scalability.

[0113] The beneficial effects of this embodiment are:

[0114] (1) Dynamic region segmentation is achieved through alliance game theory, and historical traffic characteristics are introduced into the region segmentation process. Satellites are allowed to adaptively adjust cluster affiliation according to real-time load, which overcomes the limitation that static region segmentation cannot adapt to the spatiotemporal changes of traffic. It realizes the coordinated optimization of region segmentation and routing decision, and significantly improves the network's adaptability to dynamic topology.

[0115] (2) The proposed hybrid routing mechanism combines the low signaling overhead of distributed routing with the global optimization capability of centralized routing. It achieves rapid load awareness and forwarding through distributed decision-making between clusters, and achieves fine-grained path planning through centralized control within the cluster. It effectively solves the load balancing problem while ensuring network scalability.

[0116] (3) By using the residual topology method and the backlog decision index in a collaborative design, low-load paths are dynamically constructed within the cluster to actively avoid high-load nodes and congested links. Load status and propagation delay are comprehensively considered between clusters to avoid detours, significantly reducing the probability of network congestion and queue overflow, realizing joint optimization of load and delay, and comprehensively improving transmission reliability and end-to-end delay performance.

[0117] Furthermore, in this embodiment, the steps preceding step S10 include:

[0118] Step S00: Construct the LEO mega-constellation network model, traffic and load model, satellite communication model, and optimization problem for low-load routing paths.

[0119] like Figure 3 As shown, the routing scenario of the large-scale constellation load balancing routing method based on dynamic region segmentation in this invention includes a giant LEO satellite constellation and several ground stations. The source ground station first uploads the data stream to the satellite network, and then the satellite network is responsible for data relay and forwarding until the data is successfully delivered to the destination ground station. The entire constellation is logically divided into several satellite clusters, each consisting of a cluster head node and multiple cluster member nodes. These clusters adopt an SDN architecture divided into a data plane and a control plane. The core of the control plane is the cluster head node, which is responsible for managing the member nodes within the cluster and making routing decisions for the data streams entering the cluster. The data plane consists of all member nodes in the cluster and is mainly responsible for forwarding the data streams in the cluster according to the decisions of the head node. In addition, in order to better adapt to the traffic distribution in the network, the size and distribution of the satellite clusters can be dynamically and adaptively adjusted according to changes in network topology and traffic.

[0120] The LEO satellite constellation considered in this invention is the Walker-Delta constellation. In this embodiment, the step of constructing the LEO mega-constellation network model in step S00 includes:

[0121] Step S01, using Indicates the LEO satellite constellation. This indicates the total number of satellites in the constellation. This indicates the number of orbits in a constellation, with all orbits having the same altitude. and orbital inclination and The satellites are evenly distributed in On each track, The orbital phase factor is used to determine the phase difference between corresponding satellites in adjacent orbital planes. The phase deviation between satellites in adjacent orbits is denoted as . In a constellation, the set of satellites is denoted as... Each satellite can establish ISLs with four neighboring satellites, including two inter-plane ISLs and two intra-plane ISLs.

[0122] Considering that topology changes in giant LEO satellite constellations typically occur on a minute-by-minute basis, while data transmission occurs on a millisecond-by-millisecond basis, this invention employs a dual-time-scale method to divide time slots. First, the entire operational cycle is divided into... Discrete time slots This allows for long-term changes in satellite and cluster topology. This represents the set of time slots, and then each time slot... Further subdivided into An evenly spaced hourly gap, denoted as This allows for the relay and forwarding of traffic on a short timescale; therefore, satellite constellation topology uses... It means that, among them, For time slots Internally, the topology of the satellite constellation, For a set of satellite nodes, For time slots Internal satellite link set, and in time slots Inside, satellite and satellite The link established between them is used express.

[0123] Step S02, model the satellite constellation: the satellite constellation set is denoted as... , For time slots Number of internal clusters; clusters In the time slot The number of internal satellite nodes is used Representation; Definition of binary variables This indicates the mapping relationship between satellites and the constellation. Indicates in time slot Inside, satellite Located in the cluster Inside, otherwise Cluster Topology It means that, among them, Indicates time slot Inner Satellite Cluster The topology, Indicates in time slot Internal cluster The set of satellite nodes Indicates in time slot Internal cluster A collection of satellite links.

[0124] Step S03, Model the ground area: Divide the ground to be covered by the satellite into... Each region is defined by latitude and longitude, forming a set of ground regions, denoted as . A ground station is set up at the center of each ground area. The ground station selects the satellite with the closest geographical distance as its access point, and all traffic in that area is uploaded to the satellite network through that satellite, or the traffic is downloaded. It is assumed that when a satellite is over a certain ground area, it can communicate with the ground station in that area. Define binary variables. This indicates the relationship between the ground area and the satellite. Indicates in time slot Inside, ground area Select Satellite As an access satellite, otherwise .

[0125] In this invention, the data stream, as the basic unit of transmission, exhibits specific statistical characteristics in its generation and transmission process. The data stream is entirely generated by a ground station and ultimately transmitted to another ground station. Step S00, which involves constructing the traffic and load model, includes:

[0126] In each hour gap Inside, ground area The generated traffic Obtain the parameter as Poisson distribution:

[0127] (1)

[0128] in, Represents the Poisson distribution function. It is a natural constant. Represents ground area The present invention incorporates historical traffic data as a reference factor for cluster adaptive changes. In terms of the mathematical model, the following traffic indicators are defined: First, defined in the current time slot... Each hour gap Inside, satellite The incoming traffic is:

[0129] (2)

[0130] in, Represents ground area In the hour gap The internal flow, Represents ground area With satellite Whether connected; in time slot satellite The forwarded traffic is: Therefore, in time slots The total flow through each satellite is:

[0131] (3)

[0132] in, and They represent satellites In the hour gap Traffic accessed internally and traffic forwarded; defined in time slots. The total traffic successfully relayed by each satellite was The result is obtained by formula (4) in the time slot. Inner satellite load level :

[0133] (4)

[0134] in, This represents the sum of the total bandwidth of all satellite payloads within the same time slot. Indicates in time slot Inland flow through satellite Total flow;

[0135] To make historical traffic data more effective in the adaptive adjustment of the satellite constellation, this embodiment introduces a discount factor. Appropriate discounting is applied to the satellite payload in historical time slots to calculate the payload in the current time slot. Within this context, a discount weight is applied to each satellite and each cluster. Specifically, it reduces the impact of satellite payloads farther from the current time slot on cluster changes, while increasing the impact of satellite payloads closer to the current time slot on cluster changes; within the time slot... Inside, satellite Discount weight for:

[0136] (5)

[0137] in Indicates historical time slots, Indicates historical time slots The weighting coefficient corresponding to the load, Indicates historical time slots Inside, satellite Normalized load level; cluster Discount weight This is obtained by summing the discounted weights of all its internal satellites:

[0138] (6)

[0139] in, Indicates in time slot Inner satellite Discount weighting.

[0140] The steps in step S00 of constructing the satellite communication model include:

[0141] The satellite's position is described using a spherical coordinate system. The coordinates are represented as ,in Earth's altitude The altitude of the satellite orbit. and Satellites The longitude and latitude; based on these coordinates, the satellite and satellite In the time slot European distance within The result is obtained by calculation using formula (7):

[0142] (7)

[0143] In free-space communication environments, link loss is primarily determined by free-space path loss and thermal noise power. Assuming the presence of additive white Gaussian noise and a symmetrical wireless channel, then the link... Free space path loss Represented as:

[0144] (8)

[0145] in Indicates the carrier wavelength. Indicates satellite and satellite In the time slot Euclidean distance within; according to Shannon's theorem, the information transmission rate of the link. It is calculated using formula (9).

[0146] (9)

[0147] in, This indicates the transmission power, measured in watts (W). and These represent the antenna gains of the receiving antenna and the transmitting antenna, respectively. Boltzmann's constant; This indicates the channel bandwidth, measured in MHz. This refers to thermal noise temperature, expressed in Kelvin (K). Represents the logarithmic function with base 2. Indicates in time slot Inner Link Free space path loss;

[0148] Next, establish the data flow. Path model, data flow The transmission path is denoted as: ,in For data stream The source satellite node, For data stream Target satellite node It is a pair of satellites that have established an inter-satellite link and meet the following requirements. ;

[0149] Finally, establish the data flow. The latency model for transmission in satellite networks considers three main types of latency: transmission latency and propagation latency incurred when data streams are transmitted on the link, and queuing latency incurred when data streams are queued for processing on the satellite; in time slots... hour gaps within In the middle, data flow In the link The transmission delay and propagation delay are calculated using formulas (10) and (11), respectively:

[0150] (10)

[0151] (11)

[0152] in, Represents the speed of light. Indicates in time slot Inner Link distance, Indicates the size of each data stream. Indicates in time slot Inner Link Information transmission rate; in hourly slots Internal, data flow In satellite Queuing delay is defined as , indicating data stream The waiting time for transmission; the transmission queue capacity for each satellite is defined as follows. ; in each hour gap Inside, satellite The occupied capacity of the transmission queue is denoted as The data stream transmission follows the First In First Out (FIFO) principle. If the satellite... If there are no other data streams waiting to be transmitted in the transmission queue, then the data stream... It can be transmitted out of this queue; conversely, if other data streams are waiting in the queue, it must be transmitted to the satellite. It continues to wait in the transmission queue until it becomes the next item to be processed in the queue; data stream From satellite Forwarded to adjacent satellites latency Given by formula (12):

[0153] (12)

[0154] in, and These represent the time slots. Internal, data flow In the link Transmission delay and propagation delay on the network Indicates a time slot Internal, data flow In satellite Queuing delays; data flow The end-to-end delay, i.e., the delay along its transmission path. The total delay can be expressed as:

[0155] (13)

[0156] Furthermore, the end-to-end latency threshold is set to... Data Stream Forwarding must meet the following conditions: It is carried out under the following conditions.

[0157] This invention aims to maximize system throughput by optimizing the routing paths of all data flows, while ensuring that each data flow meets its end-to-end latency constraints. The step S00, which involves constructing an optimization problem for low-load routing paths, includes:

[0158] System throughput is measured by end-to-end success rate. As a metric, this metric can be calculated using formula (14):

[0159] (14)

[0160] in, Indicates a time slot Successfully transmitted data streams within the internal network Indicates a time slot Inside, satellite Forwarded traffic; this optimization problem can be described as:

[0161] (15)

[0162] in, For end-to-end successful delivery rate, Represents data stream The routing path, It is a time slot Inner satellite With cluster The mapping relationship between binary decision variables, The relationship between satellites and the constellation is defined, that is, for any single satellite... and any satellite constellation , Indicates in time slot Inside, satellite Located in the cluster Inside, otherwise ; Indicates that it refers to any one satellite In any time slot Within, it can only be located in one satellite constellation. Within, ensure in each time slot In this system, each satellite is assigned to only one cluster; This refers to any satellite constellation. In any time slot The number of satellites contained within must be greater than 1 to ensure that in each time slot... In this system, each cluster has at least one satellite to ensure the effectiveness of the satellite cluster; The capacity of the satellite transmission queue is limited to ensure that in any time slot... hour gaps within In, any satellite The traffic carried by the transmission queue Cannot exceed the transmission queue capacity ; The end-to-end transmission delay of the data stream is limited, meaning that in any time slot... hour gaps within In the middle, data flow end-to-end delay The delay threshold cannot be exceeded. This ensures that data streams can be transmitted within the specified time. However, since dynamic cluster partitioning and routing decisions occur at different time scales, and these two processes cannot be optimized simultaneously, this embodiment addresses the optimization problem. This can be broken down into two sub-problems. These two sub-problems are: the long-scale cluster dynamic partitioning optimization problem. and load balancing routing issues on short timescales :

[0163] (16)

[0164] (17)

[0165] in, Indicates in time slot Internal Cluster Discount weight, In the time slot Internal Cluster The number of internal satellites, Indicates in time slot Inner satellite Traffic that was successfully forwarded.

[0166] For the aforementioned optimization problem, designing an algorithm to solve it remains a challenging task due to multiple challenges. Since the optimization variable is a binary integer, the problem can be simplified to a bin packing problem and a shortest path problem, proven to be a binary integer programming problem, and has been proven to be NP-hard. Solving it directly as an integer programming problem is extremely expensive. Secondly, because low-Earth orbit satellites operate at high speeds, this dynamic leads to topology fluctuations and frequent changes in visibility between satellites, resulting in frequent updates to the satellite link visibility matrix. Therefore, this optimization problem needs to be solved efficiently within each short interval while maintaining a short-term static topology. Finally, although the trajectories of LEO satellites in a satellite constellation are predictable, the specific arrival times of user flows in actual user flow arrival models are uncertain. Therefore, it is difficult to predetermine and allocate high-bandwidth paths, making it difficult to solve using traditional algorithms such as greedy algorithms or heuristic algorithms.

[0167] Considering that there are numerous equivalent paths with similar delays within a megacons, and that using these equivalent paths for flow transmission does not significantly increase latency, the bandwidth competition problem between different flows can be transformed into reducing bandwidth contention during flow transmission by scheduling routing paths to different equivalent paths. The proposed megacons load balancing routing strategy based on dynamic region segmentation comprises three stages: a dynamic region segmentation algorithm based on federated game theory, a distributed inter-cluster routing algorithm, and an intra-cluster routing algorithm based on residual topology.

[0168] This embodiment designs a dynamic region segmentation algorithm based on coalition game theory, such as... Figure 4As shown. This algorithm allows each satellite to autonomously choose to join a neighboring satellite cluster or remain in the current cluster, based on its own circumstances, thereby achieving adaptive adjustment of the cluster size and structure. Before proposing the algorithm, this invention provides relevant definitions based on the framework of coalition game theory.

[0169] It should be noted that the term "alliance" used in game theory is synonymous with "cluster" in this embodiment. Therefore, the satellite cluster set defined above is used. This indicates the alliance division strategy.

[0170] In this embodiment, step S10 includes:

[0171] Step S101, using satellite cluster assembly This represents the alliance partitioning strategy. According to alliance game theory, for a cluster... There are three indicators:

[0172] 1) Profit The value of the alliance was quantified, and , This represents an empty set of satellites; in this embodiment, it refers to a satellite cluster. The profit is defined as:

[0173] (18)

[0174] in, Indicates in time slot Internal Cluster Discount weight, In the time slot Internal Cluster The number of internal satellites, Indicates in time slot Inner satellite Traffic that was successfully forwarded.

[0175] 2) Cost The cost of quantifying cooperation typically takes into account the constraints imposed on the alliance, and the cost is defined as follows:

[0176] (19)

[0177] There is no cost when cluster partitioning satisfies both hard constraints: (1) That is, the single-satellite single-cluster affiliation constraint, where For time slots Internal satellite constellation collection, It is a time slot Inner satellite With cluster (2) Binary decision variables for attribution relationships; That is, the cluster non-empty constraint, where It is the set of all satellite nodes in the network; otherwise, the cost is positive infinity.

[0178] 3) Value The difference between the benefits gained from cooperation and the costs of cooperation is quantified, and this indicator is calculated according to formula (20):

[0179] (20)

[0180] Step S102: Define players as a set of satellites that can undergo cluster switching. This set consists of any satellite node and in the same cluster It consists of adjacent satellites within the region.

[0181] Step S103: Define a possible switching operation. For satellite collection From the original cluster Switch to a new cluster The process, namely:

[0182] (twenty one)

[0183] Among them, if the satellite set Switch to This indicates that the satellite set has established a new satellite cluster, resulting in an increase in the number of clusters in the network; if the satellite set It is a cluster If the cluster has only one set of satellites, then after this handover operation, the cluster... Will be merged into the cluster This leads to a reduction in the number of clusters in the network;

[0184] Step S104: Define the execution of the switching operation. The switching gain that can be brought for:

[0185] (twenty two)

[0186] in and Representing satellite sets Remove from cluster Before and after, cluster Value; and Representing satellite sets Enter the cluster Before and after, cluster The value of.

[0187] If satellite set If multiple switching operations are available, the operation with the highest switching gain is selected for execution. Therefore, the switching preference is defined as follows:

[0188] (twenty three)

[0189] That is, if the switching operation The switching gain is higher than The switching gain indicates the switching operation. Compare High priority.

[0190] Initially, in the cluster Inside, satellite Together with its neighboring satellites in the same cluster, they form a satellite ensemble. Subsequently, the satellite assembly was examined. Neighboring clusters Perform a search, if cluster With satellite collection If there is a link connection between them, they are considered adjacent; a handover operation is established for these adjacent clusters. And calculate the corresponding switching gain; if the switching gain If the number is a finite number of positive real numbers, the switching operation is considered valid, and the switching operation is performed. Add to the candidate operation set In the middle; when the candidate operation set When not empty, select the cluster with the highest switching gain. As a collection of satellites New affiliation; in the cluster With cluster Under the premise that no handover operation is in progress, satellite assembly The cluster switchover will be performed; otherwise, the switchover operation will be discarded; in each time slot Initially, clusters in the network will be dynamically split, a process that continues until no handover operation with positive gain can be found.

[0191] In this embodiment, the pseudocode of the dynamic region segmentation algorithm based on coalition game theory is shown below.

[0192]

[0193] Further, in this embodiment, step S20 includes:

[0194] When data stream Enter the current cluster At that time, cluster The control node is responsible for finding the exit node within the cluster for the data stream; there are two ways for the data stream to enter the current cluster: one is to upload from the ground station to the current cluster via satellite access; the other is to forward it from a neighboring cluster to the entry node of the current cluster; the control node will determine the data stream... destination node Is it located in the cluster? Inside; if the destination node Belongs to cluster Then the node Set as a data stream In the cluster Export node in If the destination node Not in cluster Within the cluster, the control node needs to execute an inter-cluster routing algorithm to route data from the current cluster. Neighbor cluster set Select the next-hop cluster for the data stream And determine in the current cluster Internal exit node and the entry node within the next-hop cluster In addition, data flow The current location is a cluster. Entry node .

[0195] In the process of selecting the next-hop cluster, a weighted inter-cluster routing metric must first be calculated according to formula (24). This metric combines load and distance factors:

[0196] (twenty four)

[0197] in, Indicates the current cluster Neighbor clusters To the destination cluster The normalized distance; this normalized distance is calculated by taking the actual distance... Divide by cluster All Neighbor Clusters To the destination cluster maximum distance To obtain, Indicates a time slot Inner satellite The size of the transmission queue that can handle the traffic. The transmission queue capacity for each satellite; The weighting coefficients represent the cluster load balancing factor and the distance weighting factor, respectively; the cluster with the smallest routing index is selected from the neighboring clusters. This cluster will serve as a data stream. The next hop cluster; finally, if the cluster With cluster If multiple links exist, then choose the one that makes... The shortest link is used as the transmission link between clusters. Represents data stream From satellite node Transmitted to satellite node Single-hop transmission delay; node Will be used as data stream In the cluster The exit node within the node Then as a data stream In the cluster The entry node within.

[0198] The pseudocode for the inter-cluster routing algorithm is shown below:

[0199]

[0200] After the inter-cluster routing algorithm has been executed, the data flow can be determined. In the current cluster Entry node inside and export nodes Next, the cluster routing algorithm needs to be configured for the data flow. Generate it in the cluster The routing path within. The entry node... As a temporary source node for the data stream, the exit node As a temporary destination node for data flow, this algorithm aims to achieve a balanced distribution of traffic within the cluster by dynamically avoiding high-load nodes and links. Its design concept is as follows: Figure 5 As shown, it mainly includes three stages: network state awareness, candidate path generation, and path selection decision.

[0201] Specifically, in this embodiment, step S30 includes:

[0202] Step S301, Network State Awareness and Residual Topology Construction: Dynamically construct the current cluster based on real-time load status. The residual topology; specifically, it includes setting the maximum load factor of the satellite as... ( ), in the current hour slot Internally, for clusters Each satellite node within If its transmission queue is overloaded Then the satellite and the ISL slave cluster directly connected to it Remove it from the topology; then, assess the connectivity of the cluster topology; if topology connectivity is found, set the payload of that satellite to an unattainable upper limit. and rejoin it to the cluster. In the topology, thus obtaining the cluster In the hour gap The residual topology.

[0203] Step S302, Candidate Path Generation: Apply a depth-first traversal algorithm to the residual topology to find slave nodes. To the node All candidate routing paths are defined, and these paths are sorted in ascending order of hop count; For data stream In the cluster The set of routing paths within, sorted in ascending order of hop count. and For data stream In the cluster Internal transmission of the first and Candidate routing paths; set a threshold for the number of routing paths within the cluster. If the number of paths exceeds Then only the routes with low similarity and few hops are retained. Select one path; otherwise, retain all paths. The route similarity calculation method is as follows: Set the route similarity threshold to... If the path has fewer hops With path The common edge count accounts for The proportion of the number of sides exceeds ,Right now

[0204] (25)

[0205] Therefore, the two paths are considered to be too similar, and the path with more hops should be discarded. ,in This represents a satellite pair in any candidate path.

[0206] Step S303, Path Selection Decision: For the set of candidate routes, establish a comprehensive utility index within the cluster to plan the final transmission route within the cluster, and define... Characterize each path within the cluster The combined utility value across two dimensions: load level and transmission latency; It can be calculated using the following formula:

[0207] (26)

[0208] in, Indicate candidate path The satellite nodes that pass through, Indicate candidate path Links traversed ; and Data streams In the cluster Maximum path load and longest path delay among all paths within the network. The weighting coefficients represent the load balancing factor and latency weighting factor within the cluster, respectively. Normalization is used to eliminate dimensional differences and ensure comparability of the indicators. Finally, the following selections are made. The shortest path is used as the data flow. In the cluster The final route path within.

[0209] The overall process of the large-scale constellation load balancing routing method based on dynamic region segmentation in this invention is as follows: Figure 2 As shown.

[0210] Based on the above description, the large-scale constellation load balancing routing method based on dynamic region segmentation of this invention includes three stages: a dynamic region segmentation algorithm based on federated game theory, a distributed inter-cluster routing algorithm, and a centralized intra-cluster routing algorithm. The overall flowchart of the algorithm is shown below. Figure 2 As shown. First, the system executes a dynamic region segmentation algorithm at the beginning of each time slot, adaptively adjusting the cluster partitioning based on network topology and historical load information. This stage utilizes federated game theory to optimize the cluster size and composition to achieve load balancing. Then, given that data flows require routing, it determines whether they should enter a new cluster: if not, forwarding is performed directly; if they do enter, a distributed inter-cluster routing algorithm and a centralized intra-cluster routing algorithm are triggered sequentially. Inter-cluster routing determines the next-hop cluster and egress node in a distributed manner based on the location and load status of neighboring clusters; intra-cluster routing is achieved by the control satellite constructing a residual topology based on real-time load, centrally planning internal cluster paths, and balancing traffic distribution while meeting latency constraints. The algorithm iteratively executes routing decisions and state updates within each time slot to achieve load balancing within the cluster.

[0211] The beneficial effects of the large-scale constellation load balancing routing method based on dynamic region segmentation in this invention are:

[0212] (1) Dynamic region segmentation is achieved through alliance game theory, and historical traffic characteristics are introduced into the region segmentation process. Satellites are allowed to adaptively adjust cluster affiliation according to real-time load, which overcomes the limitation that static region segmentation cannot adapt to the spatiotemporal changes of traffic. It realizes the coordinated optimization of region segmentation and routing decision, and significantly improves the network's adaptability to dynamic topology.

[0213] (2) The proposed hybrid routing mechanism combines the low signaling overhead of distributed routing with the global optimization capability of centralized routing. It achieves rapid load awareness and forwarding through distributed decision-making between clusters, and achieves fine-grained path planning through centralized control within the cluster. It effectively solves the load balancing problem while ensuring network scalability.

[0214] (3) By using the residual topology method and the backlog decision index in a collaborative design, low-load paths are dynamically constructed within the cluster to actively avoid high-load nodes and congested links. Load status and propagation delay are comprehensively considered between clusters to avoid detours, significantly reducing the probability of network congestion and queue overflow, realizing joint optimization of load and delay, and comprehensively improving transmission reliability and end-to-end delay performance.

[0215] To achieve the above objectives, this invention also proposes a large-scale constellation load balancing routing system based on dynamic region segmentation, such as... Figure 6 As shown, the system includes a processor 1001, a CPU, a network interface 1004, a user interface 1003, a memory 1005, a communication bus 1002, and a large-scale constellation load balancing routing program based on dynamic region segmentation stored in the memory. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard. Optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0216] Those skilled in the art will understand that Figure 6 The system structure shown does not constitute a limitation on the system and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0217] like Figure 6 As shown, the memory 1005, which serves as a computer storage medium, may include an operating device, a network communication module, a user interface module, and a large-scale constellation load balancing routing program based on dynamic region segmentation.

[0218] exist Figure 6In the system shown, network interface 1004 is mainly used to connect to the network server and communicate with the network server; user interface 1003 is mainly used to interact with user terminals and receive user input commands; and processor 1001 can be used to call the large-scale constellation load balancing routing program based on dynamic region segmentation stored in memory 1005.

[0219] To achieve the above objectives, the present invention also proposes a computer-readable storage medium storing a large-scale constellation load balancing routing program based on dynamic region segmentation. When the processor runs the large-scale constellation load balancing routing program based on dynamic region segmentation, it executes the steps of the method described above, which will not be repeated here.

[0220] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A large-scale constellation load balancing routing method based on dynamic region segmentation, characterized in that, The method operates on dual time scales, including a dynamic region segmentation phase at a long time slot scale and a hybrid routing phase at a short time slot scale. The method includes the following steps: Step S10: In the dynamic region segmentation stage, based on the dynamic region segmentation algorithm, the cluster affiliation of satellite nodes is dynamically adjusted according to their own real-time load and historical traffic information. Step S20: In the hybrid routing phase, the head node of the current cluster selects the next-hop cluster for the data flow in a distributed manner based on the real-time geographical location and load status of the neighboring clusters. Step S30: The head node of the current cluster constructs low-load routing paths within the cluster for the data flow in a centralized manner based on the residual topology method to maximize system throughput; Before step S10, the following are included: Step S00 involves constructing a LEO mega-constellation network model, a traffic and load model, a satellite communication model, and an optimization problem for low-load routing paths. The steps in step S00 for constructing the LEO mega-constellation network model include: Step S01, using Indicates the LEO satellite constellation. This indicates the total number of satellites in the constellation. This indicates the number of orbits in a constellation, with all orbits having the same altitude. and orbital inclination and The satellites are evenly distributed in On each track, The orbital phase factor is used to determine the phase difference between corresponding satellites in adjacent orbital planes. The phase deviation between satellites in adjacent orbits is denoted as . In a constellation, the set of satellites is denoted as... Each satellite can establish ISLs with four neighboring satellites, including two inter-plane ISLs and two intra-plane ISLs; A dual-time-scale method is used to divide the time slots. First, the entire operating cycle is divided into... Discrete time slots This allows for long-term changes in satellite and cluster topology. This represents the set of time slots, and then each time slot... Further subdivided into An evenly spaced hourly gap, denoted as This allows for the relay and forwarding of traffic on a short timescale; therefore, satellite constellation topology uses... It means that, among them, For time slots Internally, the topology of the satellite constellation, For a set of satellite nodes, For time slots Internal satellite link set, and in time slots Inside, satellite and satellite The link established between them is used express; Step S02, model the satellite constellation: the satellite constellation set is denoted as... , For time slots Number of internal clusters; clusters In the time slot The number of internal satellite nodes is used Representation; Definition of binary variables This indicates the mapping relationship between satellites and the constellation. Indicates in time slot Inside, satellite Located in the cluster Inside, otherwise Cluster Topology It means that, among them, Indicates time slot Inner Satellite Cluster The topology, Indicates in time slot Internal cluster The set of satellite nodes Indicates in time slot Internal cluster A collection of satellite links; Step S03, Model the ground area: Divide the ground to be covered by the satellite into... Each region is defined by latitude and longitude, forming a set of ground regions, denoted as . A ground station is set up at the center of each ground area. The ground station selects the satellite with the closest geographical distance as its access point, and all traffic in that area is uploaded to the satellite network through that satellite, or the traffic is downloaded. It is assumed that when a satellite is over a certain ground area, it can communicate with the ground stations in that area. Define binary variables. This indicates the relationship between the ground area and the satellite. Indicates in time slot Inside, ground area Select satellite As an access satellite, otherwise ; As the basic unit of transmission, data streams exhibit specific statistical characteristics in their generation and transmission processes. Data streams are entirely generated by ground stations and ultimately transmitted to another ground station. The steps in step S00 involving constructing the traffic and load model include: In every hour gap Inside, ground area The generated traffic Obtain the parameter as Poisson distribution: (1) in, Represents the Poisson distribution function. It is a natural constant. Represents ground area The traffic arrival rate; taking historical traffic data into consideration as a reference factor for cluster adaptive changes, the following traffic indicators are defined in the mathematical model: First, defined in the current time slot Each hour gap Inside, satellite The incoming traffic is: (2) in, Represents ground area In the hour gap The internal flow, Represents ground area With satellite Whether connected; in time slot hour gap China Satellite The forwarded traffic is: Therefore, in time slots The total flow through each satellite is: (3) in, and They represent satellites In the time slot hour gap Traffic accessed and traffic forwarded in the middle; Indicates in time slot Inner satellite The traffic that was successfully forwarded; calculated using formula (4) in the time slot. Inner satellite load level : (4) in, This represents the sum of the total bandwidth of all satellite payloads within the same time slot. Indicates in time slot Inland flow through satellite Total flow; Introducing discount factors Appropriate discounting is applied to the satellite payload in historical time slots to calculate the payload in the current time slot. Within each satellite and cluster, the discount weights reduce the impact of satellite loads farther from the current time slot on cluster changes, while satellite loads closer to the current time slot have a greater impact on cluster changes; within the time slot... Inside, satellite Discount weight for: (5) in Indicates historical time slots, Indicates historical time slots The weighting coefficient corresponding to the load, Indicates historical time slots Inside, satellite Normalized load level; cluster Discount weight This is obtained by summing the discounted weights of all its internal satellites: (6) in, Indicates in time slot Inner satellite Discount weighting.

2. The large-scale constellation load balancing routing method based on dynamic region segmentation according to claim 1, characterized in that, The steps in step S00 of constructing the satellite communication model include: The satellite's position is described using a spherical coordinate system. The coordinates are represented as ,in Earth's altitude The altitude of the satellite orbit. and Satellites The longitude and latitude; based on these coordinates, in time slots Inside, satellite With satellite Links between Euclidean distance The result is obtained by calculation using formula (7): (7) Assuming the presence of additive white Gaussian noise and a symmetric wireless channel, then the link... Free space path loss Represented as: (8) in Indicates the carrier wavelength. Indicates in time slot Inside, satellite With satellite Links between The Euclidean distance; according to Shannon's theorem, the information transmission rate of the link. It is calculated using formula (9); (9) in, This indicates the transmission power, measured in watts (W). and These represent the antenna gains of the transmitting and receiving antennas, respectively. Boltzmann's constant; This indicates the channel bandwidth, measured in MHz. This refers to the thermal noise temperature, expressed in Kelvin (K). Represents the logarithmic function with base 2. Indicates in time slot Inner Link Free space path loss; Next, establish the data flow. Path model, data flow The routing path is denoted as: ,in For data stream The source satellite node, For data stream Target satellite node It is a pair of satellites that have established an inter-satellite link and meet the following requirements. ; Finally, establish the data flow. The latency model for transmission in satellite networks considers three main types of latency: transmission latency and propagation latency incurred when data streams are transmitted on the link, and queuing latency incurred when data streams are queued for processing on the satellite; in time slots... hour gaps within In the middle, data flow In the link The transmission delay and propagation delay are calculated using formulas (10) and (11), respectively: (10) (11) in, Represents the speed of light. Indicates in time slot Inner satellite With satellite Links between Euclidean distance, Indicates the size of each data stream. Indicates in time slot Inner Link Information transmission rate; in hourly slots Internal, data flow In satellite Queuing delay is defined as , indicating data stream The waiting time for transmission; the transmission queue capacity for each satellite is defined as follows. ; Indicates a time slot Inner satellite The transmission queue carries the traffic; data stream transmission follows the first-in, first-out principle, if the satellite If there are no other data streams waiting to be transmitted in the transmission queue, then the data stream... It can be transmitted out of this queue; conversely, if other data streams are waiting in the queue, it must be transmitted to the satellite. It continues to wait in the transmission queue until it becomes the next item to be processed in the queue; data stream From satellite Forwarded to adjacent satellites latency Given by formula (12): (12) in, and These represent the time slots. Internal, data flow In the link Propagation delay and transmission delay on the surface Indicates a time slot Internal, data flow In satellite Queuing delays; data flow The end-to-end delay, i.e., the delay along its routing path. The total delay is expressed as: (13) Furthermore, the end-to-end latency threshold is set to... Data Stream Forwarding must meet the following conditions: It is carried out under the following conditions.

3. The large-scale constellation load balancing routing method based on dynamic region segmentation according to claim 2, characterized in that, The steps in step S00 for constructing the optimization problem regarding low-load routing paths include: System throughput is measured by end-to-end success rate. As a metric, the metric is calculated using formula (14): (14) in, Indicates a time slot Data streams successfully transmitted within the internal network Indicates a time slot Inside, satellite The incoming traffic; this optimization problem is described as: (15) in, For end-to-end successful delivery rate, Represents data stream The routing path, It is a time slot Inner satellite With cluster The mapping relationship between binary decision variables, The relationship between satellites and the constellation is defined, that is, for any single satellite... and any satellite constellation , Indicates in time slot Inside, satellite Located in the cluster Inside, otherwise ; Indicates that it refers to any one satellite In any time slot Within, it can only be located in one satellite constellation. Within, ensure in each time slot In this system, each satellite is assigned to only one cluster; This refers to any satellite constellation. In any time slot The number of satellites contained within is at least 1, ensuring that in each time slot... In this system, each cluster has at least one satellite to ensure the effectiveness of the satellite cluster; The capacity of the satellite transmission queue is limited to ensure that in any time slot... hour gaps within In, any satellite The traffic carried by the transmission queue Cannot exceed the transmission queue capacity ; The end-to-end transmission delay of the data stream is limited, meaning that in any time slot... hour gap In the middle, data flow end-to-end delay The delay threshold cannot be exceeded. This ensures that all data streams are transmitted within the specified time; optimization issues will be addressed. This can be broken down into two sub-problems: a long-scale cluster dynamic partitioning optimization problem. and load balancing routing issues on short timescales : (16) (17) in, Indicates in time slot Internal Cluster Discount weight, In the time slot Internal Cluster The number of internal satellites, Indicates in time slot Inner satellite Traffic that was successfully forwarded.

4. The large-scale constellation load balancing routing method based on dynamic region segmentation according to claim 3, characterized in that, Step S10 includes: Step S101, using satellite cluster assembly This represents the alliance partitioning strategy. According to alliance game theory, for a cluster... There are three indicators: 1) Profit The value of the alliance was quantified, and , This represents an empty set of satellites, for satellite constellations. The profit is defined as: (18) in, Indicates in time slot Internal Cluster Discount weight, In the time slot Internal Cluster The number of internal satellites, Indicates in time slot Inner satellite Traffic successfully forwarded; 2) Cost The cost of quantifying cooperation is defined as follows: (19) There is no cost when cluster partitioning satisfies both hard constraints: (1) That is, the single-satellite single-cluster affiliation constraint, where For time slots Internal satellite constellation collection, It is a time slot Inner satellite With cluster The mapping relationship between binary decision variables; (2) That is, the cluster non-empty constraint, where It must be the set of all satellite nodes in the network; otherwise, the cost is positive infinity. 3) Value The value quantifies the difference between the benefits gained from cooperation and the costs of cooperation. The index is calculated according to formula (20): (20) Step S102: Define the set of satellites where cluster handover occurs. From any satellite node and in the same cluster It consists of adjacent satellites within; Step S103: Define a possible switching operation. For satellite collection From the original cluster Switch to a new cluster The process, namely: (21) Among them, if the satellite set Switch to This indicates that the satellite set has established a new satellite cluster, resulting in an increase in the number of clusters in the network; if the satellite set It is a cluster If the cluster has only one set of satellites, then after this handover operation, the cluster... Will be merged into the cluster This leads to a reduction in the number of clusters in the network; Step S104: Define the execution of the switching operation. The switching gain that can be brought for: (22) in and Representing satellite sets Remove from cluster Before and after, cluster Value; and Representing satellite sets Enter the cluster Before and after, cluster Value; If satellite set If multiple switching operations are available, the operation with the highest switching gain is selected for execution; the switching preference is defined as follows: (23) That is, if the switching operation The switching gain is higher than The switching gain indicates the switching operation. Compare High priority; Initially, in the cluster Inside, satellite Together with its neighboring satellites in the same cluster, they form a satellite ensemble. Subsequently, the satellite assembly was examined. Neighboring clusters Perform a search, if cluster With satellite collection If there is a link connection between them, they are considered adjacent; a handover operation is established for these adjacent clusters. And calculate the corresponding switching gain; if the switching gain If the number is a finite number of positive real numbers, the switching operation is considered valid, and the switching operation is performed. Add to the candidate operation set In the middle; when the candidate operation set When not empty, select the cluster with the highest switching gain. As a collection of satellites New affiliation; in the cluster With cluster Under the premise that no handover operation is in progress, satellite assembly The cluster switchover will be performed; otherwise, the switchover operation will be discarded; in each time slot Initially, clusters in the network will be dynamically split, a process that continues until no handover operation with positive gain can be found.

5. The large-scale constellation load balancing routing method based on dynamic region segmentation according to claim 4, characterized in that, Step S20 includes: When data stream Enter the current cluster At that time, cluster The control node is responsible for finding the exit node within the cluster for the data stream; there are two ways for the data stream to enter the current cluster: one is to upload from the ground station to the current cluster via satellite access; the other is to forward it from a neighboring cluster to the entry node of the current cluster; the control node will determine the data stream... destination node Is it located in the cluster? Inside; if the destination node Belongs to cluster Then the node Set as a data stream In the cluster Export node in If the destination node Not in cluster Within the cluster, the control node needs to execute an inter-cluster routing algorithm to route data from the current cluster. Neighbor cluster set Select the next-hop cluster for the data stream And determine in the current cluster Internal exit node and the entry node within the next-hop cluster In addition, data flow The current location is a cluster. Entry node ; In the process of selecting the next-hop cluster, a weighted inter-cluster routing metric must first be calculated according to formula (24). The weighted inter-cluster routing metric combines load and distance factors: (24) in, Indicates the current cluster Neighbor clusters To the destination cluster The normalized distance; this normalized distance is calculated by taking the actual distance... Divide by cluster All Neighbor Clusters To the destination cluster maximum distance To obtain, Indicates a time slot Inner satellite The traffic carried by the transmission queue The transmission queue capacity for each satellite; The weighting coefficients represent the cluster load balancing factor and the distance weighting factor, respectively; the cluster with the smallest routing index is selected from the neighboring clusters. This cluster will serve as a data stream. The next hop cluster; finally, if the cluster With cluster If multiple links exist, then choose the one that makes... The shortest link is used as the transmission link between clusters. Represents data stream From satellite node Transmitted to satellite node Single-hop transmission delay; node Will be used as data stream In the cluster The exit node within the node Then as a data stream In the cluster The entry node within.

6. The large-scale constellation load balancing routing method based on dynamic region segmentation according to claim 5, characterized in that, Step S30 includes: Step S301, Network State Awareness and Residual Topology Construction: Dynamically construct the current cluster based on real-time load status. The residual topology; specifically, it includes setting the maximum load factor of the satellite as... , In the current hourly slot Internally, for clusters Each satellite node within If it is in a small gap Traffic carried by the internal transmission queue Then the satellite and the ISL slave cluster directly connected to it Remove from the topology; then, evaluate the connectivity of the cluster topology; if the topology is found to be disconnected, then remove the cluster. Satellite nodes within The load is set to an unattainable upper limit. and rejoin it to the cluster. In the topology, thus obtaining the cluster In the hour gap The residual topology; Step S302, Candidate Path Generation: Apply a depth-first traversal algorithm to the residual topology to find slave nodes. To the node All candidate routing paths are defined, and these paths are sorted in ascending order of hop count; For data stream In the cluster The set of routing paths within, sorted in ascending order of hop count. and For data stream In the cluster Internal transmission of the first and Candidate routing paths; set a threshold for the number of routing paths within the cluster. If the number of paths exceeds Then only the routes with low similarity and few hops are retained. Select one path; otherwise, retain all paths. The route similarity calculation method is as follows: Set the route similarity threshold to... If the path has fewer hops With path The common edge count accounts for The proportion of the number of sides exceeds ,Right now (25) Therefore, the two paths are considered to be too similar, and the path with more hops should be discarded. ,in This represents a satellite pair in any candidate path; Step S303, Path Selection Decision: For the set of candidate routes, establish a comprehensive utility index within the cluster to plan the final transmission route within the cluster, and define... Characterize each path within the cluster The combined utility value across two dimensions: load level and transmission latency; The calculation is performed using the following formula: (26) in, Indicate candidate path The satellite nodes that pass through, Indicate candidate path Links traversed ; and Data streams In the cluster Maximum path load and longest path delay among all paths within the network. The weighting coefficients represent the load balancing factor and latency weighting factor within the cluster, respectively. Normalization is used to eliminate dimensional differences and ensure comparability of the indicators. Finally, the following selections are made. The shortest path is used as the data flow. In the cluster The final route path within.

7. A large-scale constellation load balancing routing system based on dynamic region segmentation, characterized in that, The system includes a memory, a processor, and a large-scale constellation load balancing router based on dynamic region segmentation stored in the memory, wherein the large-scale constellation load balancing router based on dynamic region segmentation is executed by the processor to perform the steps of the method as described in any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a large-scale constellation load balancing routing program based on dynamic region segmentation, which, when executed by a processor, performs the steps of the method as described in any one of claims 1 to 6.

Images