Satellite-ground resource matching method based on KM algorithm

Abstract

The invention discloses a satellite-ground resource matching method based on a KM algorithm, and belongs to the field of large-scale LEO satellite networks. The method specifically comprises the following steps: firstly, building a satellite-ground resource matching scene comprising satellites and partitions, and constructing an undirected bipartite graph model; then, establishing a multi-decision model based on an AHP hierarchical analysis method, and according to three decision factors, namely partition paging demand intensity, satellite-to-ground transmission loss and continuous service duration, attaching weight to each undirected edge in the bipartite graph model; establishing an optimization target based on the maximum weight matching of the weighted bipartite graph; and finally, adopting a KM algorithm, and solving the optimization target through vertex labeling and augmented path search to obtain global optimal matching of the satellite and the partition. According to the invention, effective matching of network resource supply and regional service requirements is realized, and in different supply-demand ratio scenes and scenes with different user distribution characteristics, the service requirement coverage satisfaction rate can be effectively improved, the transmission loss can be reduced, and the average coverage duration can be prolonged.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of large-scale LEO satellite networks, and particularly relates to a star-ground resource matching method based on a KM algorithm. BACKGROUND

[0002] Due to the characteristics of a large number of satellites, wide distribution and large scale, a large-scale LEO constellation has an unprecedented application potential in the fields of communication, remote sensing and navigation, and has a revolutionary impact on the development of global aerospace industry.

[0003] A large-scale LEO satellite network is a network system composed of satellites deployed in low earth orbit, and the number of satellites can reach hundreds, thousands or even tens of thousands. These satellites generally have inter-satellite links (ISL) and usually have on-board processing capabilities. Compared with traditional small LEO satellite networks, large-scale LEO satellite networks have greater advantages in system capacity and service capability, and can more effectively manage communication traffic and provide low-orbit satellite services.

[0004] In the development process of future 5G, 6G space internet, large-scale LEO satellite networks will become a crucial infrastructure. At the same time, driven by the rapid development of small satellite technology, rocket reuse technology and satellite communication technology, the deployment of large-scale LEO satellite networks is accelerating. So far, there have been multiple large-scale LEO satellite network system plans, and the design and deployment of these plans aim to provide high-bandwidth, low-latency and seamless internet access services worldwide, such as SpaceX, OneWeb, Telesat, Amazon's Project Kuiper, etc.

[0005] In the composite low earth orbit large-scale star-ground fusion network scenario, compared with traditional low earth orbit satellite networks, the characteristics of multi-orbit and multi-star not only significantly improve the coverage range and service quality of the network, but also bring more flexible resource scheduling capabilities. However, this complex network structure also exacerbates the high dynamicity problem inherent in traditional low earth orbit satellite networks and brings new challenges such as the difficulty of matching supply and demand of star-ground resources.

[0006] Therefore, in order to reasonably allocate and fully utilize star-ground resources, it is crucial to study a geographical zoning scheme with high scalability and a star-ground access algorithm with heterogeneous resource allocation function, in combination with the advanced nature, compatibility, robustness and economy of large-scale star-ground networks. SUMMARY

[0007] In view of the dynamic resource matching problem in the geographical zoning scenario of a large-scale star-ground network, the present application provides a star-ground resource matching method based on a KM algorithm to solve the problem of large-scale multi-layer satellite access.

[0008] The satellite-to-ground resource matching method based on the KM algorithm includes the following steps:

[0009] Step 1: Build a satellite-to-ground resource matching scenario that includes satellites and partitions, and construct an undirected bipartite graph model;

[0010] Satellite collection , For the number of satellites, the set of geographical regions , This represents the number of geographical zones.

[0011] Each satellite acts as a node, and the area it covers forms a regional cluster. A regional cluster consists of several geographical regions and satellite nodes. The corresponding partition cluster is , For partitioned clusters The first in Sub-partitions , The number of sub-partitions in a partitioned cluster, i.e., satellites. Number of geographic regions covered. Satellite nodes. It can choose to access any geographical partition within its partition cluster for service;

[0012] Similarly, each geographic region is within the coverage area of ​​multiple satellites, and the set of all satellites that can cover that geographic region is called a satellite cluster. A satellite cluster consists of several satellites, with partitioned nodes. The corresponding satellite cluster is , For satellite clusters The first in satellites , The number of satellites in the satellite cluster, i.e., the number of regions that can be accessed. Total number of satellites, geographic regions You can choose to connect to any satellite in the corresponding satellite cluster.

[0013] An undirected edge is formed between the satellite and its corresponding sub-partition, using The weight of an edge is represented by... Indicates; satellite With corresponding sub-partition The undirected edges between them are Undirected edge The weight is The state relationships between satellite and ground nodes are modeled using an undirected bipartite graph. This means that the function maps each satellite to the set of areas it covers;

[0014] Step 2: Establish a multi-decision model based on the AHP hierarchical analysis method. According to the three decision factors of partition paging demand intensity, satellite-to-ground transmission loss and continuous service duration, assign weights to each undirected edge in the bipartite graph model.

[0015] Step 201: Establish a multi-decision model including the objective layer, criterion layer, and alternative layer;

[0016] The target layer aims to maximize the overall benefits of satellite-to-ground matching.

[0017] The criteria layer includes three decision factors: The paging demand intensity for a sub-partition is determined by the user density of the sub-partition. The transmission loss between satellite and ground is determined by the distance between satellite and ground. For continuous service duration;

[0018] The scheme layer is a set of candidate satellite-to-ground connection schemes. .

[0019] Step 202: Construct a judgment matrix based on the importance of the three decision factors in the multi-decision model;

[0020]

[0021] in, It represents the first The decision factor compared to the first The importance of each decision-making factor;

[0022]

[0023] Step 203: Finally, calculate the eigenvector corresponding to the largest eigenvalue of the judgment matrix to obtain the weight vector:

[0024] ;

[0025] These represent the weighting of three decision-making factors: sub-region paging demand intensity, satellite-to-ground transmission loss, and continuous service duration.

[0026] Step 204: Construct a weighted bipartite graph based on the weight vector;

[0027] Undirected edge The weights are calculated as follows:

[0028]

[0029] in In order to target the Normalized scoring function for each decision factor;

[0030] Step three, using the maximum weight matching of the weighted bipartite graph to establish the optimization target;

[0031] First, define the decision variable:

[0032]

[0033] Then, establish the optimization target as maximizing the overall connection benefit:

[0034]

[0035] The constraint conditions include:

[0036] 1) Sub-partition service guarantee: ;

[0037] 2) Satellite single connection constraint:

[0038] 3) Capacity conservation:

[0039] Step four, using the KM algorithm, through vertex labeling and augmented path search to solve the optimization target, and get the global optimal matching of the satellite and the partition.

[0040] The specific implementation process is as follows:

[0041] Specifically:

[0042] S401: Construct the initial label of each satellite and each sub-partition, and initialize the matching set to an empty set;

[0043] For each satellite , its label is initialized to the maximum weight in all sub-partitions connected to it:

[0044]

[0045] For each sub-partition , its label is initialized to 0.

[0046] S402: Initialize search queue Q and access set to the first un-matched satellite node , and update them simultaneously;

[0047] The update means: traversing the satellite node , the neighbor nodes of the explorable edge, if the neighbor node does not belong to the access set , then the neighbor node is added to the search queue and the access set ;

[0048] S403: Based on satellite nodes Starting from the point of origin, perform an alternating path search to find an augmenting path with the destination on side P, using a breadth-first search.

[0049] An alternating path search is performed from the unmatched satellite node to the unmatched sub-partition node, and an augmenting path is found through breadth-first search.

[0050] For unmatched partition nodes First, select an unmatched satellite node. As a starting point for the search;

[0051] Then, initialize the access set as follows: , and search tree edge set .

[0052] Perform alternating path expansion: retrieve the search queue The head node iterates through its explorable edge neighbors; if a neighbor does not belong to the set... Then add the neighboring node to the search queue. Access Set Traverse and visit the collection Unvisited nodes (Initially a satellite node) ), from this node Start by determining the node. Does it belong to If it is a side, then traverse its neighboring nodes. ( express If the neighboring nodes are not visited, then the next unvisited node is revisited.

[0053] Explorable edges refer to all the connecting edges between partitions that a satellite node can connect to;

[0054] Determine if the neighbor node u is not matched. If it is, find an augmenting path, which is the satellite. To the node If the path is correct, update the matching M; otherwise, add the satellite node u to the search tree and continue expanding, i.e.

[0055]

[0056] Repeat the above process until all nodes have been visited or an augmenting path has been found.

[0057] S404: Determine if an augmenting path has been found. If so, update the match using a symmetric difference operation: ; enter S405; otherwise, enter S406;

[0058] S405: the next unmatched satellite node re-join the search queue Q and the access set and update, return to S403 to search for an augmented path, until all unmatched satellite nodes are matched or visited, and output the final result:

[0059]

[0060] S406: re-adjust the label to expand the search range of the augmented path, continue to perform the alternate path search augmented path on the satellite nodes until the augmented path is found or not found, end, return to S405;

[0061] The process of adjusting the label is as follows:

[0062] First, calculate the adjustment amount :

[0063]

[0064] for the access set; for the label of the satellite ; for the label of the sub-area ; represent the minimum difference between the current label and the weight, used to adjust the label to allow more edges to participate in matching.

[0065] Then, update the label using the adjustment amount:

[0066]

[0067] for the label of the neighbor satellite node of the unvisited node v in the access set .

[0068] Finally, update the weight bipartite graph model according to the new label.

[0069] The advantages of the present application are:

[0070] A star-ground resource matching method based on the KM algorithm, based on the KM algorithm to solve the global optimal star-ground resource matching scheme, realizes the effective matching of network resource supply and regional business demand, in different supply-demand ratio scenes, different user distribution characteristics scenes, can effectively improve the service demand coverage satisfaction rate, reduce transmission loss, and prolong the average coverage time. BRIEF DESCRIPTION OF DRAWINGS

[0071] Figure 1 A flow chart of a star-ground resource matching method based on a KM algorithm of the present application;

[0072] Figure 2 A bipartite graph modeling schematic provided by an embodiment of the present application;

[0073] Figure 3 A KM algorithm flow schematic provided by an embodiment of the present application; DETAILED DESCRIPTION

[0074] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions of the embodiments of the present application will be described below in detail with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the described embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art belong to the scope of protection of the present application.

[0075] In a large-scale star-ground network in a geographical partition scenario, due to the unevenness of user distribution, the paging demand intensity of each partition for satellite space resources is significantly different. Meanwhile, satellite nodes are densely distributed and have a large coverage range. The equi-latitude and longitude partition characteristics make the amount of effective coverage resources that can be allocated by different geographical partitions at any moment significantly different.

[0076] To solve the mismatch between the resources and the demand, a star-ground resource matching method based on a KM algorithm of the present application introduces the theory of bipartite graph maximum weight matching, constructs an optimal solution solving model based on a Kuhn-Munkres (KM) algorithm, calculates a weight matrix by an AHP analysis method with factors such as partition business demand intensity, transmission loss, and residual service time length, establishes a bipartite graph mapping relationship between satellite nodes and geographical partitions, solves a globally optimal star-ground resource matching scheme, and realizes a positive correlation mapping between network resource supply and regional business demand. The present application can effectively improve the satellite coverage multiplicity of a high business density area, while guaranteeing the optimal overall resource utilization rate of the system.

[0077] The star-ground resource matching method based on the KM algorithm, as shown in Figure 1 includes the following steps:

[0078] Step one, building a star-ground resource matching scenario including satellites and partitions, and constructing an undirected bipartite graph model;

[0079] The satellite set is The set of areas covered by each satellite is a sub-cluster, and the satellite can select to access any sub-cluster in the sub-cluster. Similarly, a geographical sub-cluster is in the coverage range of multiple satellites, and the set of satellites that can cover the area is the satellite cluster that can access the sub-cluster, and the sub-cluster can select to access any satellite.

[0080] The set of areas composed of all sub-clusters is A satellite and a matching sub-cluster form an undirected edge, and the set of all undirected edges is represented by The set of weights of all edges is represented by The state relationship between the satellite and the ground node is represented by an undirected bipartite graph model This function maps each satellite to the set of areas it covers.

[0081] The undirected edge between the satellite and the corresponding sub-cluster is The weight of the undirected edge is .

[0082] The satellite node can select to access any sub-cluster in the sub-cluster to provide services. Similarly, the sub-cluster can select to access any satellite in the satellite cluster .

[0083] Step 2: Establish a multi-decision model based on the AHP hierarchical analysis method. According to the sub-cluster paging demand intensity, satellite-ground transmission loss, and continuous service duration, attach weights to each undirected edge in the bipartite graph model.

[0084] Step 201: Establish a multi-decision model including a target layer, a criterion layer, and a scheme layer.

[0085] The target layer is to maximize the comprehensive benefits of satellite-ground matching.

[0086] The criterion layer defines three evaluation criteria for ranking, including the impact factors: : sub-cluster paging demand intensity, : satellite-ground transmission loss, : continuous service duration. For the first item , the paging demand intensity is determined by the sub-cluster user density. For the second item , it is assumed that the satellite transmission scenario conditions are similar, i.e., other factors such as satellite-ground transmission atmospheric loss are ignored, and the satellite-ground transmission loss is determined by the satellite-ground distance. For the third item , it can be obtained by the following formula:

[0087]

[0088] wherein, is the duration of continuous service, is the duration of coverage of the pth satellite; and is a constant related to the same-satellite network, is the satellite elevation angle of the region where the user is located.

[0089]

[0090]

[0091] Suppose that the user needs to perform satellite switching at time , which is counted from the time when the user initiates a call; suppose that the number of satellites visible to the user at time is P, that is, the user has P candidate satellites for switching at time . represents the “trace angle” of the pth satellite.

[0092] Among the three, the paging demand intensity is more important than the other two, and therefore has the largest weight in the priority division.

[0093] The scheme layer is a candidate set of satellite-ground connection schemes .

[0094] Step 202, according to the importance of the three decision factors in the multi-decision model, a judgment matrix is constructed;

[0095]

[0096] wherein

[0097] The judgment matrix can be constructed by using the 1-9 scale method, and the typical assignment rules are as follows: B=1: equal importance; B=3: slightly important; B=5: obviously important; B=7: strongly important; B=9: extremely important.

[0098] It can be seen that the three decision conditions of , and have different importance for the final decision.

[0099] Step 203, the maximum eigenvalue of the judgment matrix and the corresponding eigenvector are finally calculated, and the weight vector is obtained, and consistency check is performed;

[0100] ;

[0101] The importance coefficients of the three decision factors, i.e. sub-division paging demand intensity, satellite-ground transmission loss and continuous service duration, in the total evaluation system respectively represent;

[0102] Step 204, according to the weight vector Constructing a weighted bipartite graph;

[0103] Undirected edges The weight is calculated as follows:

[0104]

[0105] Wherein is a normalized score function for the first decision factor;

[0106] Step three, after completing the weight assignment, the maximum weight matching of the weighted bipartite graph is used to establish the optimization target;

[0107] Define the decision variable:

[0108]

[0109] The optimization target is to maximize the overall connection benefit:

[0110]

[0111] The constraint conditions include:

[0112] 1) Sub-division service guarantee: ;

[0113] 2) Satellite single connection constraint: ;

[0114] 3) Capacity conservation: ;

[0115] The above constraint conditions represent that a satellite can only be connected to one region in the corresponding sub-division cluster; but for a region, in order to meet the various needs of regional communication, it can be connected to one or several satellites in the corresponding satellite cluster.

[0116] Step four, KM algorithm is used to solve the optimization target through vertex label and augmented path search, and the global optimal matching of satellite and sub-division is obtained.

[0117] The specific implementation process is as follows:

[0118] Specifically:

[0119] S401: Construct the initial label of each satellite and each sub-division, and initialize the matching set to an empty set;

[0120] For each satellite its label Initialize it to the maximum weight among all its connected sub-partitions:

[0121]

[0122] For each sub-partition its label Initialize to 0.

[0123] S402: Initialize the search queue Q and the access set All are unmatched first satellite nodes And update it simultaneously;

[0124] Update refers to: traversing satellite nodes The neighbor node of an explorable edge, if the neighbor node is not in the visited set. Then add the neighboring node to the search queue at the same time. Access Set ;

[0125] S403: Based on satellite nodes Starting from the point of origin, perform an alternating path search to find an augmenting path with the destination on side P, using a breadth-first search.

[0126] An alternating path search is performed from the unmatched satellite node to the unmatched sub-partition node, and an augmenting path is found through breadth-first search.

[0127] For unmatched partition nodes First, select an unmatched satellite node. As a starting point for the search;

[0128] Then, initialize the access set as follows: Search queue and search tree edge set .

[0129] Perform alternating path expansion: retrieve the search queue The head node iterates through its explorable edge neighbors; if a neighbor does not belong to the set... Then add the neighboring node to the search queue. Access Set .

[0130] Traverse and visit collection Unvisited nodes (Initially a satellite node) ), from this node Start by determining the node. Does it belong to If it is a side, then traverse its neighboring nodes. ( express If the neighboring nodes are not visited, then the next unvisited node is revisited.

[0131] Explorable edges refer to all the connecting edges between partitions that a satellite node can connect to;

[0132] Determine if the neighbor node u is not matched. If it is, find an augmenting path, which is the satellite. To the node If the path is correct, update the matching M; otherwise, add the satellite node u to the search tree and continue expanding, i.e.

[0133]

[0134] Repeat the above process until all nodes have been visited or an augmenting path has been found.

[0135] S404: Determine if an augmenting path has been found. If so, update the match using a symmetric difference operation: Proceed to S405; otherwise, proceed to S406.

[0136] S405: Move the next unmatched satellite node Re-add to search queue Q and access set Then update and return to S403 to search for augmenting paths until all unmatched satellite nodes have been matched or visited, and output the final result:

[0137]

[0138] S406: Relabeling to expand the search range of augmented paths, continuing with satellite nodes. Perform an alternating path search for an augmenting path until an augmenting path is found or the search ends if no path is found, then return to S405.

[0139] This will include all nodes visited in failed searches (the visit set). The labeling needs to be adjusted as follows:

[0140] First, calculate the adjustment amount. :

[0141]

[0142] For accessing a collection; For satellite The label; Sub-partition The label; This represents the minimum difference between the current label and the weight, used to adjust the label to allow more edges to participate in the matching.

[0143] Then, update the label using the adjustment amount:

[0144]

[0145] For accessing the collection Neighboring satellite nodes of unvisited node v The labels. For all visited satellite nodes, subtract the labels from the total number of nodes visited. All visited partition nodes, labeled with their numbers. This allows for the adjustment of the weight of each edge without disrupting the original matching, thereby expanding the explorable edges.

[0146] Finally, the weighted bipartite graph model is updated according to the new label. .

[0147] Example:

[0148] This invention allocates a large number of satellite resources (denoted as S) to a limited number of geographical areas (denoted as P). Each area has different service requirements, while each satellite has a certain coverage capability. Therefore, the goal is to achieve a reasonable resource matching scheme that ensures the needs of each area are met while taking into account factors such as the service duration of the service satellites, transmission latency, and the distribution of service satellites in multi-layer and multi-domain scenarios, so as to make full use of satellite resources.

[0149] The above problem can be abstracted into an allocation model, where partitions are considered as sets, and each region is considered as an element of that set. Correspondingly, each satellite can be considered as another set. In this model, positive integers and are defined. Let f(S) represent the number of areas each satellite can cover and the number of satellites required for each area to meet its coverage needs, respectively. The mapping relationship between satellites and regions can then be represented by the function f: S → P, where: for any subset of regions... The size of its corresponding satellite coverage set S′ must satisfy... This ensures that the coverage needs of the subset are met; at the same time, the sum of the coverage needs of all areas equals the total number of satellite resources, i.e. .

[0150] The specific steps are as follows:

[0151] Step 1: Establish the task model;

[0152] like Figure 2 As shown, the undirected bipartite graph model The function represents the state relationship between satellite and ground nodes, which maps each satellite to the set of areas it covers, and is applied to the satellite-ground resource matching scenario: in the selection process, both the satellite and the partition will consider factors such as partition paging demand, transmission loss, serviceable duration, and business balance.

[0153] Step 2: Weight assignment based on AHP multi-criteria decision model, according to the state information and distribution between different satellites and ground stations, the priority of satellite-ground selection and access is sorted by AHP.

[0154] Step 3: After completing the weight assignment based on AHP, the satellite-ground matching problem is transformed into a maximum weight matching problem of a weighted bipartite graph.

[0155] Step 4: Use the KM algorithm to solve the optimal matching problem of the above weighted bipartite graph;

[0156] The bipartite graph includes two disjoint vertex sets S and P, and each edge in the graph connects two vertices, one in S and the other in P. The problem requires selecting the highest priority matching item between nodes to provide more efficient services while meeting the above constraints. The core idea is to achieve the global optimal solution through vertex labeling and augmented path search. As shown in the following implementation process: Figure 3

[0157] Input: weighted bipartite graph G = {S, P, E, W}, where S is the set of satellite nodes, P is the set of geographical partitions, E is the edge set, and W is the weight matrix.

[0158] Output: maximum perfect matching , i.e., the global optimal matching scheme of satellite and partition.

[0159] S401: initialization;

[0160] 1) Construct the initial label: for each satellite node , its label is initialized to the maximum weight of all partition nodes connected to it, thus ensuring that the initial label of the satellite node reflects its potential maximum contribution value; for each partition node , its label is initialized to 0.

[0161] 2) Set the initial matching set: , which represents that no satellite and partition connection has been established.

[0162] 3) Initialize the search queue: create an empty queue Q for subsequent breadth-first search (BFS). ​​

[0163] S402: Alternating path search (BFS), find an augmenting path (an alternating path from an unmatched satellite node to an unmatched partition node) by breadth-first search to extend the current matching.

[0164] When there exists an unmatched node , perform the following steps: 1) select an unmatched satellite node as the search starting point. 2) initialize BFS: set the visited set to record the visited nodes) and the search tree edge set . 3) BFS extension: when there exists an unvisited node, start from the node , if belongs to side, traverse its adjacent point ( denotes the adjacent node of ). If the adjacent satellite node u is not matched, find an augmenting path and update the matching M; if u is matched, add u to the search tree and continue to extend, repeat the above process until all nodes are visited or an augmenting path is found.

[0165] S403: Augmenting operation. In S402, if an augmenting path is found , update the matching by symmetric difference operation: . If no augmenting path is found, go to S404.

[0166] S404: Re-labeling, adjust the label to expand the range of possible augmenting paths.

[0167] S405: If all nodes are matched, output the maximum matching: .

Claims

1. A method for satellite-ground resource matching based on a KM algorithm, characterized in that, Comprising the following steps: Step one, build a satellite and subzone resource matching scenario, and construct an undirected bipartite graph model; A set of satellites A set of clusters of partitions consisting of sub-partitions A set of undirected edges between satellites and corresponding sub-partitions A set of weights of edges An undirected bipartite graph model of state relations between satellite and ground nodes ; Step two, establish a multi-decision model based on AHP hierarchical analysis method, and according to the three decision factors of subzone paging demand intensity, satellite-ground transmission loss and continuous service time length, attach weights to each undirected edge in the bipartite graph model; to satellite with corresponding sub-partition undirected edge between The attached weight is calculated as follows: wherein respectively represent the proportion of three decision factors of sub-division paging demand intensity, satellite-ground transmission loss and continuous service duration; is a normalized score function for the first decision factor. Step three, use the maximum weight matching of weighted bipartite graph to establish the optimization target; The optimization target is to maximize the overall connection benefit: is a decision variable; Step four, use the KM algorithm to solve the optimization target through vertex label and augmented path search to obtain the global optimal matching of satellite and subzone; Specifically: S401: Construct initial labels for each satellite and each subpartition, and initialize the matching set is empty; S402: initialize search queue Q and access set Both are the first satellite node of unmatched And update synchronously; Update means: traverse the satellite node Explore the neighbor node of the edge, if the neighbor node does not belong to the access set Then, the neighbor node is added to the search queue at the same time And the access set ; S403: With the satellite node As a starting point, an alternate path search is performed to find an augmented path with the end point on the P side, by breadth-first search; S404: judge whether an augmented path is found If yes, update the match by symmetric difference operation: ; enter S405; otherwise, enter S406; S405: the next unmatched satellite node is added to the search queue Q and the access set is updated re-join the search queue Q and the access set and the search for the augmented path is returned to S403 until all unmatched satellite nodes are matched or visited, and the final result is output S406: Re-adjust the label to expand the search range of the augmented path, continue to search for satellite nodes Perform alternate path search augmented path until the augmented path is found or not found, return to S405; The process of adjusting the label is as follows: First, the adjustment amount is calculated : to access the collection; to a satellite of the label; to a sub-partition of the label; Then, the adjustment amount is used to update the label: to access unvisited nodes v in the set of neighbor satellite nodes of v of v. 2.The satellite-ground resource matching method based on the KM algorithm of claim 1, wherein, The satellite set is Each satellite is a node, the covered area is a sub-cluster, and the area set of the sub-cluster is ; is the number of satellites, is the number of sub-clusters in the sub-cluster; Satellite node Any one of the sub-zones in the access coverage zone cluster can be selected for service, each sub-zone as a node; Similarly, each sub-zone is within the coverage of multiple satellites, and the set of satellites that can cover the sub-zone is a satellite cluster, and the sub-zone Any one satellite in the corresponding satellite cluster can be selected for access. 3.The satellite-ground resource matching method based on the KM algorithm of claim 1, wherein, The step two is specifically: Step 201, establish a multi-decision model including target layer, criterion layer and scheme layer; Among them, the target layer is to maximize the comprehensive benefit of satellite-ground matching; The criterion layer includes three decision factors: The paging demand intensity for the subzone is determined by the user density of the subzone; The satellite-ground transmission loss is determined by the satellite-ground distance; The continuous service duration; The scheme layer is a candidate set of satellite-to-ground connection schemes ; Step 202, construct a judgment matrix according to the importance between the three decision factors in the multi-decision model; wherein, represents the importance of the first decision factor compared to the importance of the second decision factor; Step 203, finally calculate the characteristic vector corresponding to the maximum eigenvalue of the judgment matrix to obtain the weight vector: ； Step 204, construct a weighted bipartite graph according to the weight vector. 4.The satellite-ground resource matching method based on the KM algorithm of claim 1, wherein, The constraint condition corresponding to the optimization target includes: 1) Sub-division service guarantee: ; 2) Satellite single connection constraint: 3) Conservation of volume: decision variable .

5. The method of claim 1, wherein, In S401, for each satellite its label is initialized to the maximum weight among all sub-partitions connected to it: For each sub-partition its index is initialized to 0.

6. The method of claim 1, wherein the method is based on a KM algorithm. The S403 is specifically: Step 4031, initialize the access set as , search queue and search tree edge set ; Step 4032, traversing the satellite node The neighbor node of the edge can be explored, if the neighbor node does not belong to the visited set The neighbor node is added into the search queue simultaneously The visited set ; Step 4043, traverse and visit the set Unvisited nodes From this node Start by determining the node. Does it belong to If it is a side, then traverse its neighboring nodes. Proceed to step 4024; Otherwise, reiterate through the next unvisited node in the access set ; representing a neighbor node of the node Step 4034, judging whether the neighbor node u is unmatched, if yes, finding an augmenting path, otherwise, adding the satellite node u to the search tree and continuing to expand, i.e. .

Images