Satellite Communication Network User Clustering and Beam Pointing Planning Method and System

Abstract

The present invention discloses a method and system for user clustering and beam pointing planning in a satellite communication network, belonging to the field of satellite communication networks. Based on the relative position, relative velocity, and relative acceleration of users and satellites at the start time of a superframe, this solution predicts the changes in the relative positions of users and satellites in each time slot within a superframe, maps the relative position relationship in three-dimensional space to a two-dimensional plane, constructs a node relationship graph model, completes user clustering based on graph theory, and determines the beam pointing of each time slot based on the clustering results. Compared with conventional clustering methods designed for low-dynamic users and fixed-pattern satellite beam pointing planning methods, this method takes into account the three-dimensional spatial distribution of users and the high-dynamic characteristics of users, and can improve the transmission capacity of satellite communication networks.

Description

Technical Field

[0001] The present invention relates to the technical field of satellite communication networks, and particularly to a method and system for user clustering and beam pointing planning in satellite communication networks. Background Art

[0002] A satellite communication network refers to a communication network formed by multiple users simultaneously accessing the same multi-beam satellite to obtain communication transmission services provided by the satellite. In a satellite communication network, users can move rapidly in three-dimensional space, thus changing the relative position relationship between the satellite and the users. Therefore, reasonably clustering moving users and adjusting the beam pointing of each antenna of the satellite helps to improve the overall communication capacity of the network, which has important research significance.

[0003] Conventional user clustering methods are mainly applicable to low-dynamic ground users distributed in a two-dimensional plane, and these methods are insufficiently applicable to the networking scenarios of high-dynamic users distributed in three-dimensional space. In conventional satellite beam pointing planning methods, the beam pattern is usually fixed; when the user dynamics are high, frequent beam switching occurs, reducing the network transmission capacity.

[0004] It can be seen that the conventional user clustering methods and beam pointing planning methods are clustering methods designed for low-dynamic users and satellite beam pointing planning methods with fixed patterns, without considering the three-dimensional space distribution of users and the high-dynamic characteristics of users.

[0005] Currently, there are few clustering designs for high-dynamic users and satellite beam pointing planning schemes for non-fixed patterns, and there are no reports at home and abroad. Summary of the Invention

[0006] In view of this, the present invention provides a method and system for user clustering and beam pointing planning in satellite communication networks, which can consider the three-dimensional space distribution of users and the high-dynamic characteristics of users, and can improve the transmission capacity of satellite communication networks.

[0007] To achieve the above object, the technical solution of the present invention is: a method for user clustering and beam pointing planning in a satellite communication network, where a multi-beam satellite in the satellite communication network is used to provide services for K clusters with K beams; determining a set of users within the satellite coverage airspace; the satellite performs clustering management on users in units of superframes, each superframe is divided into N time slots, the duration of each time slot is t, and the following steps are performed within each superframe:

[0008] Step (1) At the beginning of the superframe, the satellite obtains the initial relative dynamics between each user and the satellite through the TT&C link, including the initial relative position, initial relative velocity, and initial relative acceleration.

[0009] Step (2) Predict the relative position, relative velocity, and relative acceleration between the user and the satellite within each time slot.

[0010] Step (3) performs dimensionality reduction mapping on the predicted relative positions of users and satellites in each time slot, maps them to a two-dimensional plane, and obtains the two-dimensional positions of each user mapped on the two-dimensional plane.

[0011] Step (4) constructs a node relationship graph model based on the two-dimensional positions of each user mapped on the two-dimensional plane.

[0012] Step (5) combines the constructed node relationship graph model and completes user clustering within a superframe based on graph theory.

[0013] Step (6) adjusts the satellite beam pointing in each time slot based on the clustering results of Step (5).

[0014] Furthermore, at the beginning of the superframe in Step (1), the satellite obtains the initial relative dynamics of each user and the satellite through the TT&C link, including the initial relative position, initial relative velocity, and initial relative acceleration. Among them, for user u with number m m , its initial relative dynamics include the initial relative position p m [0], the initial relative velocity v m [0], and the initial relative acceleration a m [0], where 1 ≤ m ≤ M, and M is the number of users in the satellite coverage airspace.

[0015] Furthermore, in Step (2), predict the relative positions, relative velocities, and relative accelerations of users and satellites in each time slot. Specifically:

[0016] According to the initial relative position p m [0], the initial relative velocity v m [0], and the initial relative acceleration a m [0] of each user and the satellite, calculate the relative positions, relative velocities, and relative accelerations of user u m and the satellite in each time slot:

[0017]

[0018] v m [n] = v m [n - 1] + a m [n - 1]t

[0019] a m [n] = a m [n - 1]

[0020] where: n is the time slot serial number, p m [n], p m [n - 1] are the user u corresponding to the nth and (n - 1)th time slots respectively mRelative position with the satellite, v m [n], v m [n - 1] are the users u corresponding to the nth and (n - 1)th time slots respectively m Relative velocity with the satellite, a m [n], a m [n - 1] are the users u corresponding to the nth and (n - 1)th time slots respectively m Relative acceleration with the satellite; t is the duration of one time slot.

[0021] Furthermore, step (3) performs dimensionality reduction mapping on the relative position, relative velocity, and relative acceleration of the user with the satellite, maps them to a two - dimensional plane, and obtains the two - dimensional positions of each user mapped on the two - dimensional plane; specifically: according to the relative position information of each user with the satellite within each time slot, calculate the maximum projection distance z in the z - axis direction of the relative positions of all users within one superframe max ; find the maximum projection distance z in the z - axis direction max After that, map the three - dimensional relative positions of each user with the satellite within each time slot onto the z max plane, and obtain the two - dimensional positions q max [n] of each user mapped on the z m plane.

[0022] Furthermore, step (4) constructs a node relationship graph model according to the two - dimensional positions of each user mapped on the two - dimensional plane; specifically: according to the two - dimensional positions q max of each user mapped on the z m [n], construct a node relationship graph model where is the vertex set of the graph, and each vertex represents a user; ε is the edge set of the graph. When the graph model is initialized, there is an edge e between each pair of vertices i,j , and the weight w i,j of the edge e i,j is the minimum distance between the two - dimensional positions of the users represented by vertex i and vertex j in the z max plane within this superframe, thus completing the establishment of the graph model.

[0023] Furthermore, step (5) combines the constructed node relationship graph model and completes user clustering within the superframe based on graph theory. The specific clustering process includes the following steps:

[0024] Step 501: Set the initial weight of each vertex i to the minimum weight of the edges connected to it:

[0025]

[0026] where is the set of edges connected to vertex i, w i,j is the weight of the edge between vertex i and j.

[0027] Set the total number of clusters to K, where the value of K is less than the total number of vertices; sort each vertex in descending order according to its initial weight, and select the K vertices with the largest initial weights. If among these K vertices, there are K' vertices that satisfy the initial weight greater than the initial clustering threshold then place these K' vertices into K' clusters respectively, where 1 ≤ K' ≤ K; if all vertices do not satisfy the initial weight greater than the initial clustering threshold value then place the vertex with the largest initial weight into the first cluster After this step is completed, if there are vertices in all clusters, execute step 503; otherwise, execute step 502.

[0028] Step 502: Divide the vertex set into a vertex set that has completed clustering and a vertex set that has not been clustered two subsets; recalculate the weights of the vertices in the set as:

[0029]

[0030] Select the vertex with the highest weight in the set and place it into the cluster that has no vertices.

[0031] Repeat step 502 until there is one vertex in all clusters, and then execute step 503.

[0032] Step 503: Delete the edges on the graph where both ends of the edge are in the set ; sort the edges with one end vertex in the set and the other end vertex in the set in ascending order according to their weights, and select the vertex i* and j* corresponding to the edge with the smallest weight, such that:

[0033]

[0034] If the number of vertices in the cluster where vertex j* is located reaches the maximum number of members F of the cluster, then delete the edge e on the graph i*,j* and then search for i* and j* again according to formula (10); if the number of vertices in the cluster where vertex j* is located is less than the maximum number of members F of the cluster, then assign vertex i* to the cluster where vertex j* is located, and move vertex i* from the set to the set

[0035] For each set The remaining vertices Repeat the following steps: Arrange all the edges with one end vertex being i and the other end vertex being in the cluster where vertex i* is located in descending order of weight, and delete all the edges except the one with the largest weight.

[0036] Repeat step 503 until the set is an empty set, and at this time, the clustering results of each cluster are obtained

[0037] Furthermore, step (6) adjusts the satellite beam pointing in each time slot based on the clustering results of step (5). Specifically: Calculate the positions of the beam centers in the two-dimensional plane according to the input clustering results and the two-dimensional positions of each user mapped in the two-dimensional plane; then inverse-map the two-dimensional positions of the beam centers in the two-dimensional plane back to the three-dimensional space to obtain the unit direction vectors of the beam pointing.

[0038] Another embodiment of the present invention also provides a satellite communication network user clustering and beam pointing planning system, including a relative dynamic acquisition module, a relative position prediction module, a dimensionality reduction mapping module, a user clustering module, and a beam pointing planning module.

[0039] The relative dynamic acquisition module is used to obtain the initial relative dynamics of each user and the satellite through the TT&C link at the beginning of the superframe and send them to the relative position prediction module. The initial relative dynamics include the initial relative position, the initial relative velocity, and the initial relative acceleration.

[0040] The relative position prediction module is used to predict the relative position, relative velocity, and relative acceleration of the user and the satellite in each time slot according to the initial relative dynamics of each user and the satellite, and send them to the dimensionality reduction mapping module.

[0041] The dimensionality reduction mapping module is used to perform dimensionality reduction mapping on the relative position, relative velocity, and relative acceleration of the user and the satellite, map them to the two-dimensional plane, and obtain the two-dimensional positions of each user mapped in the two-dimensional plane, and send them to the user clustering module.

[0042] The user clustering module is used to construct a node relationship graph model according to the two-dimensional positions of each user mapped in the two-dimensional plane, and complete user clustering within the superframe based on graph theory. The clustering results are sent to the beam pointing planning module.

[0043] The beam pointing planning module is used to adjust the satellite beam pointing in each time slot based on the clustering results.

[0044] Further, the dimensionality reduction mapping module is specifically: according to the relative position information between each user and the satellite within each time slot, calculate the maximum projection distance z in the z-axis direction of the relative positions of all users within a superframe max ; After finding the maximum projection distance z in the z-axis direction max , map the three-dimensional relative positions of each user and the satellite within each time slot onto the z max plane to obtain the two-dimensional position q max of each user mapped on the z m [n].

[0045] Further, the user clustering module, the specific clustering process includes the following steps:

[0046] Step 501: Set the initial weight of each vertex i to the minimum weight of the connecting edges connected to it

[0047]

[0048] where is the set of connecting edges connected to vertex i, and w i,j is the weight of the connecting edge between vertex i and j.

[0049] Set the total number of clusters to K, and the value of K is less than the total number of vertices; sort all vertices in descending order according to their initial weights, and take the K vertices with the largest initial weights. If among these K vertices, there are K' vertices whose initial weights are greater than the initial clustering threshold , then put these K' vertices into K' clusters respectively, where 1 ≤ K' ≤ K; if all vertices do not satisfy the initial weight being greater than the initial clustering threshold value , then put the vertex with the largest initial weight into the first cluster After this step is completed, if there are vertices in all clusters, execute step 503, otherwise execute step 502.

[0050] Step 502: Divide the vertex set into two subsets: the vertex set that has completed clustering and the vertex set that has not been clustered yet; recalculate the weights of the vertices in the set as:

[0051]

[0052] Take the vertex with the highest weight in the set and put it into the cluster without vertices.

[0053] Repeat step 502 until there is one vertex in each cluster, and then execute step 503.

[0054] Step 503: Delete the edges on the graph where both vertices at the two ends of the connected edge are in the set ; Sort the edges with one vertex at each end in the sets and in ascending order of weight, and select the vertices i* and j* corresponding to the edge with the smallest weight, such that:

[0055]

[0056] If the number of vertices in the cluster where vertex j* is located reaches the maximum number of members F of the cluster, then after deleting the edge e i*,j* on the graph, search for i* and j* again according to formula (10); if the number of vertices in the cluster where vertex j* is located is less than the maximum number of members F of the cluster, then assign vertex i* to the cluster where vertex j* is located, and move vertex i* from the set to the set

[0057] For each set of the remaining vertices Repeat the following steps: Sort all the edges with one end vertex being i and the other end vertex being in the cluster where vertex i* is located in descending order of weight, and delete all the edges except the one with the largest weight.

[0058] Repeat step 503 until the set is an empty set, and at this time, the clustering results of each cluster are obtained

[0059] Beneficial effects:

[0060] A method and system for user clustering and beam pointing planning in a satellite communication network provided by the present invention. Based on the relative position, relative velocity, and relative acceleration of the user and the satellite at the start time of the superframe, this method predicts the change in the relative position of the user and the satellite in each time slot within a superframe, maps the relative position relationship in three-dimensional space to a two-dimensional plane, constructs a node relationship graph model, completes user clustering based on graph theory, and determines the beam pointing of each time slot based on the clustering results. Compared with the conventional clustering methods designed for low-dynamic users and the fixed-pattern satellite beam pointing planning methods, this method takes into account the three-dimensional spatial distribution of users and the high-dynamic characteristics of users, and can improve the transmission capacity of the satellite communication network. Description of the Drawings

[0061] Figure 1It is a method and system block diagram for user clustering and beam pointing planning in a satellite communication network. Detailed implementation mode

[0062] The present invention will be described in detail below with reference to the accompanying drawings and by way of examples.

[0063] Example 1:

[0064] The present invention provides a method for user clustering and beam pointing planning in a satellite communication network. A multi-beam satellite in the satellite communication network is used to provide services for K clusters with K beams; a user set within the satellite coverage airspace is determined; the satellite manages user clustering in units of superframes. Each superframe is divided into N time slots, and the duration of each time slot is t. The following steps are executed within each superframe:

[0065] Step (1) At the beginning of the superframe, the satellite obtains the initial relative dynamics between each user and the satellite through the TT&C link, including the initial relative position, initial relative velocity, and initial relative acceleration; among them, for the user u numbered m m , its initial relative dynamics include the initial relative position p m [0], initial relative velocity v m [0], and initial relative acceleration a m [0], where 1 ≤ m ≤ M, and M is the number of users within the satellite coverage airspace.

[0066] Step (2) Predict the relative position, relative velocity, and relative acceleration between the user and the satellite within each time slot; specifically:

[0067] Based on the initial relative position p m [0], initial relative velocity v m [0], and initial relative acceleration a m [0] of each user and the satellite, calculate the relative position, relative velocity, and relative acceleration between the user u m and the satellite within each time slot:

[0068]

[0069] v m [n] = v m [n - 1] + a m [n - 1]t

[0070] a m [n] = a m [n - 1]

[0071] Where: n is the time slot serial number, p m [n], p m [n - 1] are the relative positions of the user u corresponding to the nth and (n - 1)th time slots respectively mRelative position with the satellite, v m [n], v m [n - 1] are the users u corresponding to the nth and (n - 1)th time slots respectively m Relative velocity with the satellite, a m [n], a m [n - 1] are the users u corresponding to the nth and (n - 1)th time slots respectively m Relative acceleration with the satellite; t is the time

[0072] In step (3), the predicted relative positions of the users and the satellite in each time slot are dimensionally reduced and mapped onto a two - dimensional plane to obtain the two - dimensional positions of each user mapped on the two - dimensional plane; specifically: According to the relative position information of each user and the satellite in each time slot, calculate the maximum projection distance z of all users' relative positions in the z - axis direction within a super - frame max ; find the maximum projection distance z in the z - axis direction max After that, map the three - dimensional relative positions of each user and the satellite in each time slot onto the z max plane to obtain the two - dimensional positions q max of each user mapped on the z m plane, [n]

[0073] In step (4), according to the two - dimensional positions of each user mapped on the two - dimensional plane, construct a node - relationship graph model; specifically: According to the two - dimensional positions q max of each user mapped on the z m plane, [n], construct a node - relationship graph model where is the vertex set of the graph, and each vertex represents a user; ε is the edge set of the graph. When the graph model is initialized, there is an edge e between every pair of vertices i,j , and the weight w i,j of the edge e i,j is the minimum distance between the two - dimensional positions of the users represented by vertex i and vertex j in the z max plane within this super - frame, thus completing the establishment of the graph model

[0074] In step (5), combined with the constructed node - relationship graph model, complete the user clustering within the super - frame based on graph theory; the specific clustering process includes the following steps:

[0075] Step 501: Set the initial weight of each vertex i to the minimum weight of the edges connected to it:

[0076]

[0077] where is the set of edges connected to vertex i, w i,jis the weight of the edge connecting vertices i and j;

[0078] Set the total number of clusters to K, where K is smaller than the total number of vertices; sort the vertices in descending order according to their initial weights, and take the K vertices with the largest initial weights. If among these K vertices, there are K' vertices that meet the initial weights Greater than the initial clustering threshold Then put these K' vertices into K' clusters respectively, where 1≤K'≤K; if all vertices do not meet the initial weight Greater than the initial clustering threshold Then put the vertex with the largest initial weight into the first cluster After this step is completed, if there are vertices in all clusters, then execute step 503, otherwise execute step 502;

[0079] Step 502: Assemble the vertices Divide into a set of vertices that have completed clustering and the set of vertices that have not yet been clustered Two subsets; recalculate the set The vertex weights in are:

[0080]

[0081] Get Collection The vertex with the highest weight is placed in the cluster that has no vertices yet;

[0082] Repeat step 502 until there is a vertex in all clusters, and then execute step 503;

[0083] Step 503: In the diagram Delete the edge whose two ends are both in the set The edges in the set and collection The edge Arrange them in ascending order from small to large weights, and select the vertices i* and j* corresponding to the edge with the smallest weight so that they satisfy:

[0084]

[0085] If the number of vertices in the cluster where vertex j* is located reaches the maximum number of members F of the cluster, then in the graph Delete the edge e i*,j* After that, search for i* and j* again according to formula (10); if the number of vertices in the cluster where vertex j* is located is less than the maximum number of cluster members F, then assign vertex i* to the cluster where vertex j* is located, and remove vertex i* from the set

[0086] Move to Collection

[0087] For each set The remaining vertices Repeat the following steps: Arrange all the edges with one end vertex being i and the other end vertex being in the cluster where vertex i* is located in descending order of weight, and delete all the edges except the one with the largest weight;

[0088] Repeat step 503 until the set Is an empty set, and at this time, the clustering results of each cluster are obtained

[0089] Step (6) adjusts the satellite beam pointing in each time slot based on the clustering results of step (5). Specifically, in this step, the positions of the beam centers in the two-dimensional plane are calculated according to the input clustering results and the two-dimensional positions of each user mapped in the two-dimensional plane; then the two-dimensional positions of the beam centers in the two-dimensional plane are inversely mapped back to the three-dimensional space to obtain the unit direction vectors of the beam pointings.

[0090] Embodiment 2:

[0091] Another embodiment of the present invention also provides a satellite communication network user clustering and beam pointing planning system, the structural block diagram is as Figure 1 Shown, including a relative dynamic acquisition module, a relative position prediction module, a dimensionality reduction mapping module, a user clustering module, and a beam pointing planning module.

[0092] Without loss of generality, the satellite communication network user clustering and beam pointing planning are described as follows: Assume that a multi-beam satellite can generate K beams to serve K clusters, and each cluster can accommodate at most F users. The set of users in the satellite coverage airspace is where u m Is the user numbered m, and M is the number of users in the satellite coverage airspace. The satellite manages user clustering in units of superframes. Each superframe is divided into N time slots, and the duration of each time slot is t. The satellite needs to predict the relative positions p m Of the users in each time slot of this superframe through the initial relative position p m [0], the initial relative velocity v m [0] and the initial relative acceleration a m [0] of the user u m [n], where 1 ≤ n ≤ N, and allocate the user u m To the cluster And determine the unit direction vector e k [n] of the beam k pointing in each time slot, where 1 ≤ k ≤ K.

[0093] At the beginning of the superframe, the relative dynamic acquisition module obtains the initial relative dynamics between each user and the satellite through the TT&C link. For user u numbered m m , its initial relative dynamics include the initial relative position p m [0] of the satellite, the initial relative velocity v m [0], and the initial relative acceleration a m [0], where 1 ≤ m ≤ M, and M is the number of users within the satellite coverage area. The initial relative dynamics are then passed to the relative position prediction module.

[0094] Based on the initial relative position p m [0], the initial relative velocity v m [0], and the initial relative acceleration a m [0] of each user and the satellite, the relative position prediction module calculates the relative position p m [n], relative velocity v m [n], and relative acceleration a m [n] between user u m and the satellite in each time slot. The calculation method is as follows: n is the time slot number;

[0095]

[0096] v m [n] = v m [n - 1] + a m [n - 1]t

[0097] a m [n] = a m [n - 1]

[0098] where: n is the time slot sequence number, p m [n] and p m [n - 1] are the relative positions between user u m and the satellite corresponding to the nth and (n - 1)th time slots respectively, v m [n] and v m [n - 1] are the relative velocities between user u m and the satellite corresponding to the nth and (n - 1)th time slots respectively, a m [n] and a m [n - 1] are the relative accelerations between user u m and the satellite corresponding to the nth and (n - 1)th time slots respectively; t is the time.

[0099] After the calculation is completed, the relative positions between each user and the satellite in each time slot are passed to the dimensionality reduction mapping module.

[0100] The dimensionality reduction mapping module calculates the maximum projection distance of the relative positions of all users within a superframe in the z-axis direction based on the relative position information of each user and the satellite in each time slot.

[0101]

[0102] Where (p m [n]) i is the i-th element of the vector p m [n].

[0103] After finding the maximum projection distance z max in the z-axis direction, the three-dimensional relative positions of each user and the satellite in each time slot are mapped onto the z max plane. The mapping method is as follows:

[0104]

[0105]

[0106] Where (q m [n])1 is the horizontal axis value of the mapped two-dimensional coordinate, and (q m [n])2 is the vertical axis value of the mapped two-dimensional coordinate.

[0107] After obtaining the two-dimensional positions q max mapped on the z m plane, the maximum projection distance z max in the z-axis direction is passed to the beam pointing planning module, and the two-dimensional positions q max mapped on the z m plane of each user are passed to the user clustering module and the beam pointing planning module.

[0108] The user clustering module constructs a node relationship graph model based on the two-dimensional positions q max mapped on the z m plane of each input user Where is the vertex set of the graph, and each vertex represents a user; ε is the edge set of the graph. When the graph model is initialized, there is an edge e between each pair of vertices i,j , and the weight w i,j of the edge e i,j is the minimum distance between the two-dimensional positions mapped on the z max plane of the users represented by vertex i and vertex j in this superframe, expressed as:

[0109]

[0110] After completing the establishment of the graph model, perform the clustering steps:

[0111] Step 1: Set the initial weight of each vertex \(i\) to the minimum weight of the connected edges:

[0112]

[0113] where is the set of connected edges to vertex \(i\).

[0114] Set the number of clusters to \(K\), where the value of \(K\) is less than the total number of vertices; sort all vertices in descending order according to their initial weights, and select the \(K\) vertices with the largest initial weights. If among these \(K\) vertices, there are \(K'\) vertices that satisfy the initial weight greater than the initial clustering threshold then place these \(K'\) vertices into \(K'\) clusters respectively, where \(1\leq K'\leq K\); if all vertices do not satisfy the initial weight greater than the initial clustering threshold ( Set an initial value, for example, it can be set to half of the beam width), then place the vertex with the largest initial weight into the first cluster After this step is completed, if there are vertices in all clusters, execute Step 3, otherwise execute Step 2.

[0115] Step 2: Divide the vertex set into two subsets, the vertex set that has been clustered and the vertex set that has not been clustered,. Recalculate the weights of the vertices in the set as:

[0116]

[0117] Select the vertex with the highest weight in the set and place it into the cluster that has no vertices.

[0118] Repeat Step 2 until there is one vertex in all clusters, and then execute Step 3.

[0119] Step 3: Delete the edges on the graph whose both end vertices are in the set . Sort the edges whose one end vertex is in the set and the other end vertex is in the set in ascending order according to their weights, and select the vertex \(i^*\) and \(j^*\) corresponding to the edge with the smallest weight, such that:

[0120]

[0121] ​If the number of vertices in the cluster where vertex j* is located reaches the maximum number of members F of the cluster, then on the graph delete the connecting edge e i*,j* After that, search for i* and j* again according to Equation (10); if the number of vertices in the cluster where vertex j* is located is less than the maximum number of members F of the cluster, then assign vertex i* to the cluster where vertex j* is located, and move vertex i* from the set to the set

[0122] For each set the remaining vertices Repeat the following: Arrange all the connecting edges with one end vertex being i and the other end vertex being in the cluster where vertex i* is located in descending order of weight, and delete all the connecting edges except the one with the largest weight.

[0123] Repeat Step 3 until the set

[0124] is an empty set. At this time, the clustering results of each cluster are obtained Pass the clustering results to the beam pointing planning module.

[0125] The beam pointing planning module first calculates the positions c max of the beam centers in the z max plane according to the input clustering results and the two-dimensional positions of each user mapped in the z k [n]

[0126]

[0127] where is the number of vertices in the kth cluster.

[0128] Then, inverse-map the positions of the beam centers in the z max plane back to the three-dimensional space to obtain the unit direction vectors e k [n] of each beam pointing. This vector can indicate the pointing direction:

[0129]

[0130] In summary, the above is only a preferred embodiment of the present invention and is not used to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for user clustering and beam pointing planning in a satellite communication network, characterized in that The multi-beam satellite in the satellite communication network serves K clusters with K beams; determine the set of users within the satellite coverage airspace; the satellite manages user clustering in units of superframes. Each superframe is divided into N time slots, and the duration of each time slot is t. The following steps are performed within each superframe: Step (1) At the beginning of the superframe, the satellite obtains the initial relative dynamics between each user and the satellite through the TT&C link, including the initial relative position, initial relative velocity, and initial relative acceleration; Step (2) Predict the relative position, relative velocity, and relative acceleration between the user and the satellite within each time slot; Step (3) Perform dimensionality reduction mapping on the predicted relative position between the user and the satellite within each time slot, map it to a two-dimensional plane, and obtain the two-dimensional positions of each user mapped on the two-dimensional plane; Step (4) Construct a node relationship graph model based on the two-dimensional positions of each user mapped on the two-dimensional plane; Step (5) Combine the constructed node relationship graph model and complete user clustering within the superframe based on graph theory; Step (6) Adjust the satellite beam pointing within each time slot based on the clustering result of Step (5).

2. The satellite communication network user clustering and beam pointing planning method according to claim 1, characterized in that At the start of the superframe in step (1), the satellite obtains the initial relative dynamics between the satellite and each user through the TT&C link, including the initial relative position, the initial relative velocity, and the initial relative acceleration. Among them, for user u numbered m m , its initial relative dynamics include the initial relative position p m [0], the initial relative velocity v m [0], and the initial relative acceleration a m [0], where 1 ≤ m ≤ M and M is the number of users within the airspace covered by the satellite.

3. A method for satellite communication network user clustering and beam pointing planning according to claim 1, characterized in that The specific method for Step (2) to predict the relative position, relative velocity, and relative acceleration between the user and the satellite within each time slot is as follows: According to the initial relative positions p m [0] of each user and the satellite, the initial relative velocities v m [0], and the initial relative accelerations a m [0], calculate the relative positions, relative velocities, and relative accelerations of the user u m and the satellite within each time slot: v m [n] = v m [n - 1]+a m [n - 1]t a m [n] = a m [n - 1] Where: n is the time slot number, p m [n]、p m [n-1] are user u corresponding to the nth and n-1th time slots respectively m Relative position to the satellite, v m [n]、v m [n-1] are user u corresponding to the nth and n-1th time slots respectively m Relative speed to the satellite, a m [n]、a m [n-1] are user u corresponding to the nth and n-1th time slots respectively m Relative acceleration to the satellite; t is the duration of a time slot.

4. A method for satellite communication network user clustering and beam pointing planning according to claim 3, characterized in that, Step (3) performs dimensionality reduction mapping on the relative position, relative velocity, and relative acceleration between the user and the satellite, maps them to a two-dimensional plane, and obtains the two-dimensional positions of each user mapped on the two-dimensional plane. Specifically, according to the relative position information between each user and the satellite in each time slot, calculate the maximum projection distance z of the relative positions of all users in the z-axis direction within one superframe max ; Find the maximum projection distance \(z\) in the \(z\)-axis direction max After that, map the three-dimensional relative positions of each user and the satellite within each time slot onto the \(z\)- max plane to obtain the two-dimensional positions \(q\) max mapped by each user on the \(z\)- m [n].

5. A method for satellite communication network user clustering and beam pointing planning according to claim 4, characterized in that, The step (4) constructs a node relationship graph model according to the two-dimensional positions of each user mapped on the two-dimensional plane; specifically: according to the two-dimensional positions q max [n] of each user mapped on the z m plane, a node relationship graph model G(V, E) is constructed, where V is the set of vertices of the graph, and each vertex represents a user; E is the set of connecting edges of the graph. When the graph model is initialized, there is a connecting edge e i,j between every pair of vertices i, j ∈ V. The weight w i,j of the connecting edge e i,j is the minimum distance between the two-dimensional positions of the users represented by vertices i and j in the z max plane within this superframe, thereby completing the establishment of the graph model.

6. A method for satellite communication network user clustering and beam pointing planning according to claim 5, characterized in that, The specific clustering process for Step (5) to combine the constructed node relationship graph model and complete user clustering within the superframe includes the following steps: Step 501: Set the initial weight of each vertex i to the minimum weight of the connected edges; where L i is the set of edges connected to vertex i, and w i,j is the weight of the edge between vertex i and j; Set the total number of clusters to K, where the value of K is less than the total number of vertices; sort each vertex in descending order according to its initial weight, and select the K vertices with the largest initial weights. If among these K vertices, there are K' vertices that satisfy the initial weight greater than the initial clustering threshold then put these K' vertices into K' clusters respectively, where 1 ≤ K' ≤ K; if all vertices do not satisfy the initial weight greater than the initial clustering threshold then put the vertex with the largest initial weight into the first cluster C1; after this step is completed, if there are vertices in all clusters, execute step 503, otherwise execute step 502; Step 502: Divide the vertex set V into a vertex set that has been clustered and a vertex set that has not been clustered yet; recalculate the vertex weights in the set into two subsets; The vertex weights in the set are recalculated as follows: Take the set the vertex with the highest weight and place it in a cluster that has no vertices yet; Repeat Step 502 until there is one vertex in each cluster, and then execute Step 503; Step 503: Delete the edges in the graph G(V, E) whose both end vertices are in the set ; For the edges e with one end vertex in the set and the other end vertex in the set i,j , i ∈ V nc , j ∈ V c , sort them in ascending order of weights from small to large, and select the vertices i* and j* corresponding to the edge with the smallest weight, such that: If the number of vertices in the cluster where vertex j* is located reaches the maximum number of members F of the cluster, then delete the edge e in the graph G(V, E). i*,j* After that, search for i* and j* again according to Equation (10); if the number of vertices in the cluster where vertex j* is located is less than the maximum number of members F of the cluster, then assign vertex i* to the cluster where vertex j* is located, and move vertex i* from the set to the set V c _; For each set for the remaining vertex \(i\in V\) nc , repeat the following steps: Arrange all the edges with one end vertex being \(i\) and the other end vertex being the vertex in the cluster where \(i^*\) is located in descending order of weight, and delete all the edges except the edge with the largest weight; Repeat step 503 until the set is an empty set, and at this time, the clustering results C1,..., C K of each cluster are obtained.

7. A method for satellite communication network user clustering and beam pointing planning according to claim 1, 2, 3, 5 or 6, characterized in that, The specific method for Step (6) to adjust the satellite beam pointing within each time slot based on the clustering result of Step (5) is as follows: Calculate the position of the center of each beam in the two-dimensional plane according to the input clustering result and the two-dimensional positions of each user mapped on the two-dimensional plane; then inverse-map the two-dimensional positions of the centers of each beam in the two-dimensional plane back to the three-dimensional space to obtain the unit direction vector of each beam pointing.

8. A satellite communication network user clustering and beam pointing planning system, characterized in that, It includes a relative dynamics acquisition module, a relative position prediction module, a dimensionality reduction mapping module, a user clustering module, and a beam pointing planning module; The relative dynamics acquisition module is used to obtain the initial relative dynamics between each user and the satellite through the TT&C link at the beginning of the superframe and send it to the relative position prediction module. The initial relative dynamics include the initial relative position, initial relative velocity, and initial relative acceleration; The relative position prediction module is used to predict the relative position, relative velocity, and relative acceleration between the user and the satellite within each time slot according to the initial relative dynamics between each user and the satellite and send it to the dimensionality reduction mapping module; The dimensionality reduction mapping module is used to perform dimensionality reduction mapping on the relative position, relative velocity, and relative acceleration between the user and the satellite, map it to a two-dimensional plane, and obtain the two-dimensional positions of each user mapped on the two-dimensional plane and send it to the user clustering module; The user clustering module is used to construct a node relationship graph model according to the two-dimensional positions of each user mapped on the two-dimensional plane and complete user clustering within the superframe based on graph theory. The clustering result is sent to the beam pointing planning module; The beam pointing planning module is used to adjust the satellite beam pointing within each time slot based on the clustering result.

9. A satellite communication network user clustering and beam pointing planning system according to claim 8, characterized in that, The dimensionality reduction mapping module specifically calculates, according to the relative position information between each user and the satellite in each time slot, the maximum projection distance z of the relative positions of all users within a superframe in the z-axis direction. max ; Find the maximum projection distance z in the z-axis direction max After that, map the three-dimensional relative positions of each user and the satellite within each time slot onto the z max plane to obtain the two-dimensional positions q max of each user mapped on the z m plane, where m is the user number within the satellite coverage airspace and n is the time slot number.

10. A satellite communication network user clustering and beam pointing planning system according to claim 8 or 9, characterized in that, The user clustering module constructs a node relationship graph model according to the two-dimensional positions of each user mapped on the two-dimensional plane; specifically: according to the two-dimensional positions q max [n] of each user mapped on the z m plane, a node relationship graph model G(V, E) is constructed, where V is the set of vertices of the graph, and each vertex represents a user; E is the set of connecting edges of the graph. When the graph model is initialized, there is a connecting edge e i,j between every pair of vertices i, j ∈ V. The weight w i,j of the connecting edge e i,j is the minimum distance between the two-dimensional positions of the users represented by vertices i and j in the z max plane within this superframe, thus completing the establishment of the graph model; The specific clustering process includes the following steps: Step 501: Set the initial weight of each vertex i to the minimum weight of the connected edges thereto where L i is the set of edges connected to vertex i, and w i,j is the weight of the edge between vertex i and j; Set the total number of clusters to K, where the value of K is less than the total number of vertices; sort each vertex in descending order according to its initial weight, and select the K vertices with the largest initial weights. If among these K vertices, there are K' vertices that satisfy the initial weight greater than the initial clustering threshold then put these K' vertices into K' clusters respectively, where 1 ≤ K' ≤ K; if all vertices do not satisfy the initial weight greater than the initial clustering threshold value then put the vertex with the largest initial weight into the first cluster C1; after this step is completed, if there are vertices in all clusters, execute step 503, otherwise execute step 502; Step 502: Divide the vertex set V into a vertex set that has been clustered and a vertex set that has not been clustered yet; recalculate the vertex weights in the set into two subsets; The vertex weights in the set are as follows: Select the vertex with the highest weight from the set and place it in a cluster that currently has no vertices; Repeat step 502 until there is one vertex in each cluster, and then execute step 503; Step 503: Delete the edges in the graph G(V, E) whose both end vertices are in the set ; For the edges e whose one end vertex is in the set and the other end vertex is in the set i,j , i ∈ V nc , j ∈ V c , sort them in ascending order according to their weights from smallest to largest, and select the vertices i* and j* corresponding to the edge with the smallest weight, such that: If the number of vertices in the cluster where vertex j* is located reaches the maximum number of members F of the cluster, then the connecting edge e is deleted in the graph G(V, E). i*,j* After that, i* and j* are searched again according to Equation (10); if the number of vertices in the cluster where vertex j* is located is less than the maximum number of members F of the cluster, then vertex i* is assigned to the cluster where vertex j* is located, and vertex i* is moved from the set V_ nc to the set V c _; For each set V_ nc for the remaining vertices i ∈ V nc , repeat the following steps: Arrange all the edges with one end vertex being i and the other end vertex being in the cluster where vertex i* is located in descending order of weight, and delete all the edges except the one with the largest weight; Repeat step 503 until the set V_ nc is an empty set, and at this time, the clustering results C1,..., C K of each cluster are obtained.

Images