Satellite communication scheduling method, device, equipment and medium based on dual link

Abstract

The present application discloses a satellite communication scheduling method, apparatus, device and medium based on dual links. The method includes: constructing a first multidimensional matrix based on the position information of multiple satellites; extracting features from the first multidimensional matrix to determine a second multidimensional matrix; constructing first graph data based on the second multidimensional matrix, wherein the nodes of the first graph data correspond to each satellite respectively, and the node attribute information is used to indicate the data transmission volume and data reception volume of the cell corresponding to each satellite, and the edges of the first graph data are used to indicate that two satellites can communicate with each other, and the edge attributes are used to indicate the data transmission volume and the communication strategy that can be implemented between the two satellites; using a pre-trained graph neural network to transmit messages to the first graph data, thereby determining the corresponding second graph data; and uniquely determining the communication strategy for unidirectional communication between each satellite based on the edge feature information of the unidirectional edges in the second graph data.

Description

Technical Field

[0001] The present application relates to the field of satellite communication technology, and in particular to a dual-link-based satellite communication scheduling method, apparatus, device, and medium. Background Art

[0002] Currently, low-orbit satellites can carry both microwave phased array antennas and laser communication terminals, enabling simultaneous communication via microwave and laser links. Laser links offer the advantages of ultra-high bandwidth, high transmission rates, and low latency, and can achieve extremely long single-hop distances (theoretically, up to 5,400 kilometers). However, lasers are susceptible to atmospheric haze and turbulence, and tracking and aiming techniques are complex when satellites are in high-speed motion. In contrast, microwave links offer the advantages of high reliability and robustness against adverse weather conditions, but they suffer from limited bandwidth and transmission rates, and relatively high latency.

[0003] Therefore, in actual low-orbit satellite communication services, how to schedule microwave links and laser links so as to take into account the transmission advantages of both microwave links and laser links is a technical problem that the low-orbit satellite communication system urgently needs to solve. Summary of the Invention

[0004] The embodiments of the present disclosure provide a dual-link-based satellite communication scheduling method, apparatus, device, and medium to at least solve the technical problems existing in the prior art.

[0005] According to one aspect of an embodiment of the present disclosure, a dual-link-based satellite communication scheduling method is provided, comprising: constructing a first multidimensional matrix based on position information of multiple satellites, the first multidimensional matrix being used to indicate the positions between each satellite and position environment information related to signal communication; performing feature extraction on the first multidimensional matrix to determine a second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate communication strategies that can be implemented between each satellite, wherein the communication strategies include: communication via microwave links and communication via laser links; constructing first graph data based on the second multidimensional matrix, wherein nodes of the first graph data correspond to each satellite respectively, node attribute information is used to indicate the data transmission volume and data reception volume of the cell corresponding to each satellite, and edges of the first graph data are used to indicate that communication can be carried out between two satellites, and edge attributes are used to indicate the data transmission volume between the two satellites and the communication strategies that can be implemented; performing message passing on the first graph data using a pre-trained graph neural network to determine the corresponding second graph data; and uniquely determining the communication strategy for unidirectional communication between each satellite based on edge feature information of unidirectional edges in the second graph data.

[0006] According to another aspect of the embodiments of the present disclosure, a storage medium is further provided. The storage medium includes a stored program, wherein the above method is executed by a processor when the program is running.

[0007] According to another aspect of an embodiment of the present disclosure, a dual-link satellite communication scheduling device is provided, comprising: a first matrix construction module, configured to construct a first multidimensional matrix based on position information of a plurality of satellites, the first multidimensional matrix being used to indicate the positions between the satellites and position environment information related to signal communication; a second matrix determination module, configured to perform feature extraction on the first multidimensional matrix to determine a second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate communication strategies that can be implemented between the satellites, wherein the communication strategies include: communication via microwave links and communication via laser links; a first graph data construction module, configured to construct first graph data based on the second multidimensional matrix, wherein nodes of the first graph data correspond to satellites, node attribute information is used to indicate data transmission and data reception amounts of cells corresponding to the satellites, edges of the first graph data are used to indicate that two satellites can communicate, and edge attributes are used to indicate data transmission amounts between the two satellites and implementable communication strategies; a message passing module, configured to perform message passing on the first graph data using a pre-trained graph neural network, thereby determining corresponding second graph data; and a communication strategy scheduling module, configured to uniquely determine a communication strategy for unidirectional communication between the satellites based on edge feature information of unidirectional edges in the second graph data.

[0008] According to another aspect of an embodiment of the present disclosure, a dual-link-based satellite communication scheduling device is also provided, comprising: a processor; and a memory connected to the processor, configured to provide the processor with instructions for processing the following processing steps: constructing a first multidimensional matrix based on position information of multiple satellites, the first multidimensional matrix being used to indicate the positions between each satellite and position environment information related to signal communication; performing feature extraction on the first multidimensional matrix to determine a second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate communication strategies that can be implemented between each satellite, wherein the communication strategies include: communication via microwave links and communication via laser links; constructing first graph data based on the second multidimensional matrix, wherein nodes of the first graph data correspond to each satellite respectively, node attribute information is used to indicate the data transmission volume and data reception volume of the cell corresponding to each satellite, and edges of the first graph data are used to indicate that communication can be performed between two satellites, and edge attributes are used to indicate the data transmission volume and the communication strategies that can be implemented between the two satellites; using a pre-trained graph neural network to perform message passing on the first graph data to determine the corresponding second graph data; and uniquely determining the communication strategy for unidirectional communication between each satellite based on edge feature information of the unidirectional edges in the second graph data.

[0009] In the disclosed embodiment, a multidimensional matrix-based feature extraction operation is first used to determine feasible communication strategies between multiple low-orbit satellites based on the distribution of multiple low-orbit satellites and the signal transmission environment. Then, a graph neural network-based message passing operation is used to further determine a unique communication strategy for one-way communication between each satellite based on the data transmission volume of the cell corresponding to each satellite and the one-way data transmission volume between each satellite, combined with the already determined feasible communication strategies between each low-orbit satellite. Thus, through two decisions, the present application combines the distribution of multiple low-orbit satellites and the signal transmission environment, as well as the data transmission volume and data reception volume of each cell, thereby taking into account the transmission advantages of microwave links and laser links and rationally scheduling microwave links and laser links. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The drawings described herein are used to provide a further understanding of the present disclosure and constitute a part of this application. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation of the present disclosure. In the drawings:

[0011] Figure 1 is a schematic diagram for implementing a satellite communication scheduling system based on microwave-laser dual links according to Embodiment 1 of the present disclosure;

[0012] Figure 2 1 is a flow chart of a satellite communication scheduling method based on a microwave-laser dual link according to the first aspect of Embodiment 1 of the present disclosure;

[0013] Figure 3 is a schematic diagram of the first multidimensional matrix according to Embodiment 1 of the present disclosure;

[0014] Figure 4 is a schematic diagram of the second multidimensional matrix according to embodiment 1 of the present disclosure;

[0015] Figure 5 is a schematic diagram of a graph structure based on multiple low-orbit satellites according to embodiment 1 of the present disclosure;

[0016] Figure 6 This is a schematic diagram of obtaining second graph data by performing message transmission on first graph data according to the first aspect of embodiment 1 of the present disclosure;

[0017] Figure 7 is a schematic diagram of determining communication strategy information for unidirectional communication between satellites based on edge features of second graph data according to the first aspects of embodiment 1 of the present disclosure;

[0018] Figure 8 is a schematic diagram of dividing the atmospheric space into grids according to Example 1 of the present disclosure;

[0019] Figure 9 is a schematic diagram of determining an inter-satellite grid between satellites according to embodiment 1 of the present disclosure;

[0020] Figure 10 is a schematic diagram of generating a second multidimensional matrix based on a first multidimensional matrix according to embodiment 1 of the present disclosure;

[0021] Figure 11 is a schematic diagram of a matrix conversion model according to embodiment 1 of the present disclosure;

[0022] Figure 12 is a schematic diagram of a satellite communication scheduling device based on a microwave-laser dual link according to Embodiment 2 of the present disclosure; and

[0023] Figure 13 This is a schematic diagram of a satellite communication scheduling device based on microwave-laser dual links according to Example 3 of the present disclosure. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present disclosure.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] Example 1

[0027] According to this embodiment, a method embodiment of a satellite communication scheduling method based on a microwave-laser dual link is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0028] Figure 1 The schematic diagram of the satellite communication scheduling system based on microwave-laser dual link according to this embodiment is shown. The system includes: a plurality of low-orbit satellites S1~S m And a ground station 100. Among them, multiple low-orbit satellites S1~S m They can communicate with the ground station 100 respectively.

[0029] Based on the above system, this embodiment proposes a satellite communication scheduling method based on microwave-laser dual link. Figure 2 A schematic flow chart of the method is shown. Figure 2 As shown, the method includes:

[0030] S202: Constructing a first multidimensional matrix based on position information of a plurality of satellites, where the first multidimensional matrix is ​​used to indicate positions between the satellites and position environment information related to signal communication;

[0031] S204: extracting features from the first multidimensional matrix to determine a second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate communication strategies that can be implemented between satellites, wherein the communication strategies include: communication via microwave links and communication via laser links;

[0032] S206: Constructing first graph data based on the second multidimensional matrix, wherein nodes of the first graph data correspond to satellites, and node attribute information is used to indicate data transmission and data reception amounts of cells corresponding to the satellites. Edges of the first graph data are used to indicate that two satellites can communicate with each other, and edge attributes are used to indicate data transmission amounts and implementable communication strategies between the two satellites.

[0033] S208: Using a pre-trained graph neural network to perform message passing on the first graph data, thereby determining corresponding second graph data; and

[0034] S210: Uniquely determine a communication strategy for unidirectional communication between satellites based on edge feature information of the unidirectional edges in the second graph data.

[0035] Specifically, the ground station 100 can be based on multiple satellites S1~S m The ephemeris information determines multiple satellites S1~S m Then, the ground station 100 can obtain meteorological data of various regions from websites such as meteorological data platforms, and thus m The first multidimensional matrix for characterizing the environmental element information related to the signal communication between each satellite is determined based on the location information and the acquired meteorological data. , where c represents the number of environmental element types. Figure 3 A schematic diagram of a first multidimensional matrix is ​​shown.

[0036] In this embodiment, using x i,j,u represents each element data in the first multidimensional matrix, where i, j = 1 to m, and u = 1 to c. Thus x i,j,u Indicates the location environment information between the i-th and j-th satellites relative to the u-th information type. Specifically, information types may include: straight-line distance, satellite orbit inclination difference, atmospheric visibility, precipitation intensity, refractive index structure constant, etc. This location environment information affects not only the attenuation of laser link communications, but also the attenuation of microwave link communications.

[0037] Thus, the ground station 100 can detect multiple satellites S1~S m Each pair of satellites S i and S j , determine the various types of position environment information x between the pair of satellites i,j,u , thus forming the first multidimensional matrix (S202). In addition, Can be regarded as satellite S i and S j The position environment information vector between Figure 3 As shown), each element of the position environment information vector is used to indicate the satellite S i and S j The different types of location environment information between the two will not be described in detail here. The specific method for determining the first multi-dimensional matrix X will be described in detail later.

[0038] Then, the ground station 100 performs feature extraction on the first multidimensional matrix X, thereby generating a second multidimensional matrix corresponding to the first multidimensional matrix .in, Figure 4 A schematic diagram of the second multidimensional matrix Y is shown.

[0039] In this embodiment, y i,j,v Represents satellite S i and S j The probability of implementing the vth communication strategy between satellites S, where v = 1~2. i and S j The second communication strategy is satellite S i and S j They communicate via microwave links. i,j,v When ≥50%, it means satellite S i and S j They can communicate through the vth communication strategy, otherwise it means that the satellite Si and S j They cannot communicate through the vth communication strategy.

[0040] For example, when y i,j,1 When ≥50%, it means satellite S i and S j They can communicate via microwave links, otherwise it means that the satellite S i and S j They cannot communicate via microwave links; when y i,j,1 When ≥50%, it means satellite S i and S j They can communicate through laser links, otherwise it means that the satellite S i and S j They cannot communicate via laser link; when y i,j,1 and y i,j,2 When both are ≥50%, it means satellite S i and S j They can communicate via microwave links as well as laser links; when y i,j,1 and y i,j,2 When both are less than 50%, it means satellite S i and S j They cannot communicate with each other via microwave links or laser links.

[0041] Thus, the ground station 100 can detect multiple satellites S1~S m Each pair of satellites S i and S j , determine the probability y of the communication strategy that can be implemented between the pair of satellites i,j,v , thus forming the second multidimensional matrix (S204). In addition, Can be regarded as satellite S i and S j Communication strategy information vector between Figure 4 As shown), the elements of the information vector are used to indicate S i and S j The probability information of each communication strategy can be implemented between them will not be described in detail here. The specific method of determining the second multidimensional matrix Y will be described in detail later.

[0042] Then, the ground station 100 defines a graph structure according to the second multi-dimensional matrix. Specifically, the ground station 100 constructs a graph structure with each satellite S1~S m Corresponding nodes N1~N m Then, the ground station 100 builds the nodes N1~N according to the following rules. m The corresponding adjacency matrix :

[0043] 1) w i,j =1, when y i,j,1 ≥50% or y i,j,2 ≥50%;

[0044] 2) w i,j =0, when y i,j,1 and y i,j,2 When both are <50%.

[0045] Then, the ground station 100 constructs the edges of the graph structure according to the adjacency matrix, thereby defining the graph structure. Considering the bidirectionality of data transmission between satellites, the graph structure is a bidirectional graph, where two nodes are connected by two edges in opposite directions. Figure 5 A schematic diagram showing the structure of the graph is shown in Figure 5 In the graph structure shown, there are L unidirectional edges.

[0046] Then, further reference Figure 6 As shown, the ground station 100 determines the first graph data corresponding to the graph structure. The first graph data includes the first graph data corresponding to each node N1~N m Corresponding node attribute information A1~A m and with each edge E1~E L Corresponding edge attribute information B1~B L . Among them, the node attribute information A1~A m Used to indicate each satellite S1~S m Corresponding cells C1~C m The amount of data transmission and data reception. L It is used to indicate the data transmission volume between the two corresponding satellites and the communication strategy that can be implemented. m and edge attribute information B1~B L This will be described in detail later. Thus, in this way, the ground station 100 constructs the first graph data (S206).

[0047] Then, the ground station 100 uses a message passing model based on a graph neural network to perform message passing on the first graph data to extract features of the first graph data, and then generates second graph data corresponding to the first graph data. The second graph data is feature information corresponding to the first graph data (S208). Figure 6 As shown, the second graph data includes node features F1~F corresponding to each graph node. m And each one-way E1~E L Corresponding edge features G1~G L The specific message transmission operations will be described in detail later.

[0048] Finally, reference Figure 7 As shown, the ground station 100 sets each unidirectional edge E k The corresponding edge feature G k Input to the pre-set fully connected layer and binary classifier to uniquely determine the unidirectional edge E k The corresponding communication type. For example, the binary classifier outputs a two-dimensional classification vector Q k , the two-dimensional classification vector contains two elements, namely the probability of communication through the microwave link and the probability of communication through the laser link (S210). Therefore, the strategy with higher probability can be used as the unidirectional communication strategy corresponding to the unidirectional edge. For example, Figure 6 In the classification vector Q1 corresponding to the unidirectional edge E1, if the probability corresponding to microwave link communication is higher, it indicates that microwave link communication is adopted from satellite S2 to satellite S1. If the probability corresponding to laser link communication is higher in the classification vector Q2 corresponding to the unidirectional edge E2, it indicates that laser link communication is adopted from satellite S1 to satellite S2. Therefore, the two-bit classification vector Q k , which can also be regarded as a one-way edge E k Communication strategy information.

[0049] Then, the ground station 100 sends the communication strategy information corresponding to the communication type of each unidirectional edge to each low-orbit satellite S1~S m . Thus, low-orbit satellites S1~S m Microwave link communication or laser link communication can be selected for data transmission according to corresponding communication strategy information.

[0050] In this way, the present application first uses a feature extraction operation based on a multidimensional matrix to determine the feasible communication strategy between each low-orbit satellite based on the distribution of multiple low-orbit satellites and the signal transmission environment. Then, it uses a message passing operation based on a graph neural network to further determine the unique communication strategy for one-way communication between each satellite based on the data transmission volume of the cell corresponding to each satellite and the one-way data transmission volume between each satellite, combined with the feasible communication strategy between each low-orbit satellite. Thus, the present application makes two decisions, combining the distribution of multiple low-orbit satellites and the signal transmission environment, and combining the data transmission volume and data reception volume of each cell, so as to take into account the transmission advantages of microwave links and laser links and reasonably schedule microwave links and laser links.

[0051] Optionally, the operation of constructing a first multidimensional matrix based on the position information of multiple satellites includes: determining the distances and orbital inclination differences between the satellites based on the position information of the multiple satellites; determining the atmospheric environment information between the satellites based on the position information of the multiple satellites; and constructing the first multidimensional matrix based on the distances, orbital inclination differences, and atmospheric environment information between the multiple satellites. Furthermore, the operation of determining the atmospheric environment information between the satellites based on the position information of the multiple satellites includes: dividing the atmospheric space into multiple grids; determining the grid corresponding to each satellite based on the position information of the multiple satellites; determining the atmospheric environment information of the inter-satellite grids through which the straight line connecting the satellites passes; and determining the atmospheric environment information between the satellites based on the atmospheric environment information of the inter-satellite grids.

[0052] Specifically, in this embodiment, the ground station 100 can use the distances between the satellites as the first channel matrix of the first multidimensional matrix X, that is, x i,j,1 Used to indicate satellite S i and S j The distance between them.

[0053] In addition, the ground station 100 can use the orbital inclination differences between the satellites as the second channel matrix of the first multidimensional matrix X, that is, x i,j,2 Used to indicate satellite S i and S j The difference in orbital inclination.

[0054] Then, refer to Figure 8 As shown, the ground station 100 can divide the atmospheric space into a plurality of grids Z1~Z 16 , so that the atmospheric environment information in each grid can be determined, such as atmospheric visibility, precipitation intensity, refractive index structure constant, etc. 16 The corresponding atmospheric environment information can be represented in the form of a vector, for example. Different elements of each vector represent numerical values ​​of different types of atmospheric environment information, such as atmospheric visibility, precipitation intensity, refractive index structure constant, etc.

[0055] Then, refer to Figure 9 As shown, the ground station 100 can be based on each satellite S1~S m The location information of each satellite S1~S m The corresponding grid.

[0056] Then, further reference Figure 9 As shown, the ground station 100 determines the grid through which the straight lines between the satellites pass, i.e., the inter-satellite grid. Figure 9In FIG, the inter-satellite grids between satellites S1 and S2 are grids Z2 and Z3, the inter-satellite grid between satellites S1 and S3 is grid Z6, and the inter-satellite grid between satellites S2 and S3 is grid Z7 and Z8.

[0057] The ground station 100 then determines the atmospheric environment information between each satellite based on the atmospheric environment information of the inter-satellite grid. When the inter-satellite grid consists of multiple grids (for example, the inter-satellite grids between satellites S1 and S2 are grids Z2 and Z3, and the inter-satellite grids between satellites S2 and S3 are grids Z7 and Z8), the atmospheric environment information of the multiple grids can be averaged to provide the atmospheric environment information between the satellites. For example, for satellites S1 and S2, the atmospheric visibility of grids Z2 and Z3 can be averaged to provide the atmospheric visibility between satellites S1 and S2, the precipitation intensity of grids Z2 and Z3 can be averaged to provide the precipitation intensity between satellites S1 and S2, and the refractive index structure constant of grids Z2 and Z3 can be averaged to provide the refractive index structure constant between satellites S1 and S2, and so on. This process can be repeated to determine the atmospheric environment information between each satellite.

[0058] Then, the ground station 100 sends each satellite S i and S j Different types of atmospheric environment information between the two are used as the data of channel 3 to channel c of the first multidimensional matrix, that is, x i,j,3 ~x i,j,c Corresponding to satellite S i and S j Different types of atmospheric environment information between.

[0059] In this way, the ground station 100 is constructed to indicate the i and S j The first dimension matrix of the position and atmospheric environment .

[0060] Optionally, the operation of performing feature extraction on the first multidimensional matrix to determine the second multidimensional matrix includes: inputting the first multidimensional matrix into a matrix conversion model, and generating a third multidimensional matrix corresponding to the first multidimensional matrix through the matrix conversion model, wherein the third multidimensional matrix includes four channels, the first channel and the second channel are used to indicate whether microwave link communication can be implemented between each satellite, and the third channel and the fourth channel are used to indicate whether laser link communication can be implemented between each satellite; using a first classifier to classify the elements corresponding to the first channel and the second channel to determine the probability that microwave link communication can be implemented between each satellite; using a second classifier to classify the elements corresponding to the third channel and the fourth channel to determine the probability that laser link communication can be implemented between each satellite; and using the results output by the first classifier and the second classifier as the second multidimensional matrix.

[0061] Specifically, refer to Figure 10 As shown, the ground station 100 inputs the first multidimensional matrix X into the pre-trained matrix conversion model to generate a third multidimensional matrix with four channels . The element d of the first channel of the third multidimensional matrix i,j,1 Used to indicate the satellite S i and S j The integral value of the microwave link communication can be implemented between them; the element d of the second channel i,j,2 Used to indicate the satellite S i and S j The integral value of the microwave link communication cannot be implemented between them; the element d of the third channel i,j,3 Used to indicate the satellite S i and S j The integral value of the laser link communication can be implemented between them; the element d of the fourth channel i,j,4 Used to indicate the satellite S i and S j The integral value at which laser link communication cannot be implemented.

[0062] Then, the first classifier based on binary classification is used to classify the elements corresponding to the first channel and the second channel to determine the probability that microwave link communication can be implemented between the satellites, for example:

[0063] (h1,h2)=softmax(d i,j,1 , d i,j,2 ); and y i,j,1 =h1.

[0064] In addition, the second classifier based on binary classification is used to classify the corresponding elements of the third and fourth channels to determine the probability that laser link communication can be implemented between the satellites, for example:

[0065] (h3,h4)=softmax(d i,j,3 , d i,j,4 ); and y i,j,1 =h3.

[0066] Therefore, the ground station 100 can determine the second multi-dimensional matrix Y through the above method.

[0067] Furthermore, the matrix conversion model includes an encoding network and a decoding network, wherein the encoding network is used to encode the first multidimensional matrix to generate a corresponding feature map, and the decoding network is used to decode the feature map to generate a second multidimensional matrix, and wherein the encoding network and the decoding network also perform feature fusion via a jump connection. Furthermore, the encoding network includes a plurality of encoding modules arranged in cascade, wherein each encoding module includes a convolution layer and a pooling layer; the decoding network includes a plurality of decoding modules arranged in cascade, wherein each decoding module includes a deconvolution layer and an upsampling layer; and the feature maps output by each encoding module are fused with the feature maps to be input to the corresponding decoding module via a jump connection.

[0068] Specifically, refer to Figure 11 As shown, the matrix conversion model includes an encoding network and a decoding network, wherein the encoding network is used to encode the first multidimensional matrix X to generate a feature map, and the decoding network is used to decode the feature map generated by the encoding network to generate a third multidimensional matrix D. In addition, the encoding network and the decoding network are connected by skipping to perform feature fusion.

[0069] In addition, further reference Figure 11 As shown, the encoding network includes multiple encoding modules, each of which includes multiple convolutional layers and pooling layers. Thus, each encoding module can be regarded as a convolutional neural network unit, thereby using convolution to extract features from the first multidimensional matrix X. Each encoding module will output a corresponding feature map. Thus, after the encoding network uses multiple encoding modules to encode the first multidimensional matrix X multiple times, it outputs the final feature map to the decoding network.

[0070] Although the figure shows that the encoding network includes three encoding modules and each encoding module includes three convolutional layers, the specific number of encoding modules and the number of convolutional layers in each encoding module can be deployed according to actual conditions. Similarly, the downsampling capability of the pooling layer can also be deployed according to actual conditions and will not be detailed here.

[0071] Further references Figure 11 As shown, the decoding network includes multiple decoding modules. Decoding modules 1 to 3 include their own deconvolution layers and upsampling layers. They are used to decode the feature maps. Figure 11 As shown, the last decoding module includes not only a deconvolution layer, but also a convolution layer. The convolution layer of the structure includes 4 convolution kernels, each of which has only one weight parameter. Thus, the final decoding network can output the third multidimensional matrix .

[0072] Therefore, in the present application, by utilizing convolution-based encoding operations and inverse convolution-based decoding operations, a first multidimensional matrix for indicating location environment information between satellites can be mapped to a third multidimensional matrix for indicating the feasibility of microwave link communication and the feasibility of laser link communication between satellites.

[0073] In existing technologies, encoding and decoding networks are often used for image processing, such as image region segmentation. However, this application uses encoding and decoding network models to implement mappings between multidimensional matrices representing different meanings (for example, mapping from a location-indicating environment information space to a communication strategy information space). This allows the spatial data processing capabilities of the encoding and decoding networks to accurately evaluate and predict communication strategies between satellites.

[0074] Furthermore, the operation of constructing the first graph data based on the second multidimensional matrix includes: concatenating the following information of the cells corresponding to each satellite to obtain node attribute information corresponding to each satellite: the current data transmission request volume of the cell; the data transmission volume transmitted by the cell in multiple consecutive cycles before the current moment; and the data reception volume received by the cell in multiple consecutive cycles before the current moment. Furthermore, the operation of constructing the first graph data based on the second multidimensional matrix also includes: concatenating the following information corresponding to each unidirectional edge to obtain edge attribute information corresponding to each unidirectional edge: communication strategy information corresponding to the unidirectional edge and the data transmission volume transmitted via the unidirectional edge in multiple consecutive cycles before the current moment.

[0075] Specifically, refer to Figure 6 As shown, in this application, each node attribute information A1~A m Including the corresponding satellites S1~S m Corresponding cells C1~C m The following information: 1) the current data transmission request amount a0 of the cell; 2) the data transmission amount a1~a1 transmitted by the cell in the n consecutive cycles before the current moment n ; 3) The amount of data received by the cell in n consecutive cycles before the current time a n+1 ~a 2n+1 .

[0076] For example, A i =[a i,0 , a i,1 , a i,2 , a i,3 , ..., a i,2n+1 ] T , where a i,0 Indicates cell C i The current amount of data transfer requests; ai,1 ~a i,n Indicates cell C i The amount of data transmitted in n consecutive cycles before the current moment; and i,n+1 ~a i,2n+1 Indicates cell C i The amount of data received in n consecutive cycles before the current time.

[0077] In addition, for each one-way edge E k (k=1~L), its edge attribute B k Includes n+2 elements b0~b n+1 The elements b0 and b1 correspond to the communication strategy information between the two satellites corresponding to the two nodes connected by the unidirectional edge, and the elements b2~b n+1 They correspond to the amount of data transmitted from the cell corresponding to the source node of the unidirectional edge to the cell corresponding to the destination node in the n consecutive cycles before the current moment.

[0078] For example, for Figure 5 The unidirectional edge E1 shown in the figure has attribute B1 as [b 1,0 , b 1,1 , b 1,2 , ..., b 1,n+1 ] T Among them, element b 1,0 and b 1,1 Respectively with y 1,2,1 and y 1,2,2 Correspondingly, element b 1,2 ~b 1,n+1 Corresponds to the amount of data transmitted from cell C2 to cell C1 in the n consecutive cycles before the current moment. For the unidirectional edge E2, its attribute B2 is [b 2,0 , b 2,1 , b 2,2 ,..., b 2,n+1 ] T Among them, element b 2,0 and b 2,1 Respectively with y 1,2,1 and y 1,2,2 Correspondingly, element b 2,2 ~b 2,n+1 Corresponds to the amount of data transmitted from cell C1 to cell C2 in the n consecutive cycles before the current moment. Similarly, we can determine Figure 5 The unidirectional edges E1~E L Edge attributes B1~B L .

[0079] In this way, the ground station 100 determines the relationship with each graph node N1~Nm Corresponding node attribute information A1~A m , and each unidirectional edge E1~E L Corresponding edge attribute information B1~B L , thereby constructing the first graph data.

[0080] Optionally, a pre-trained graph neural network is used to perform message passing on the first graph data to determine the corresponding second graph data, including performing multi-layer message passing on the first graph data to determine the corresponding second graph data. And wherein, a single message passing includes: based on the node information of the target node to be transmitted and the edge information of the unidirectional edge with the target node as the source node, a first fusion feature is generated by fusing; and based on the first fusion feature, updated node information corresponding to the target node is generated by using a first multi-layer perceptron. And wherein, a single message passing also includes: based on the edge information of the target unidirectional edge to be transmitted and the node information of the source node and the destination node associated with the unidirectional edge, a second fusion feature is generated by fusing; and based on the second fusion feature, updated edge information corresponding to the target unidirectional edge is generated by using a second multi-layer perceptron.

[0081] Specifically, although Figure 6 The message transmission process described in the preceding text may be a single round of message transmission for the first graph data, or multiple rounds of message transmission for the first graph data. The following takes the first message transmission as an example to illustrate a single message transmission:

[0082] 1. About message delivery of node data

[0083] 1) For the target node to be transferred, the node information of the target node to be transferred and the edge information of the unidirectional edge with the target node as the source node are fused to generate a first fused feature. For example, for graph node N1, the node attribute information A1 corresponding to the graph node in the first graph data and the edge attribute information B2 and B3 of the unidirectional edges E2 and E3 with graph node N1 as the source node can be fused to generate the first fused feature.

[0084] Specifically, for example, the edge attribute information B2 and B3 may be averaged to obtain edge attribute mean information, and then the edge attribute mean information and the node attribute information A1 may be concatenated to obtain the first fusion feature.

[0085] 2) The first fused feature is input into the first MLP layer to obtain the updated node information corresponding to the graph node N1. The first MLP layer is dedicated to performing message passing operations on graph nodes.

[0086] Thus, through the above method, the node attribute information A of each graph node is i to update.

[0087] 2. About message passing for edge data

[0088] 1) For a target unidirectional edge with transfer, based on the edge information of the target unidirectional edge to be transferred and the node information of the source and destination nodes associated with the unidirectional edge, for example, taking unidirectional edge E1 as an example, the edge attribute information B1 of unidirectional edge E1, the node attribute information A2 of source node N2, and the node attribute information A1 of destination node N1 are concatenated to generate the second fused feature.

[0089] 2) The second fused feature is input into the second MLP layer to obtain the updated edge information corresponding to the unidirectional edge E1. The second MLP layer is dedicated to performing message passing operations on edge nodes.

[0090] Thus, through the above method, the edge attribute information B of each unidirectional edge is i to update.

[0091] Thus, one round of message passing is completed through the above operations. Then, based on the graph data output by the first round of message passing, the next round of message passing is performed in the same manner, and this iteration is repeated until the second graph data is generated.

[0092] Therefore, through the above method, the first graph data is constructed according to the data transmission information of the cells corresponding to each satellite and the one-way communication between each satellite, and then the second graph data is generated according to the first graph data using the message passing operation based on the graph neural network. Therefore, the characteristics of the second graph data can not only reflect the feasible one-way communication strategy between each satellite, but also reflect the data transmission volume of the cells corresponding to each satellite and the data transmission volume of the one-way communication between each satellite, so that the communication of multiple satellites can be scheduled more accurately.

[0093] Optionally, based on the edge feature information of the unidirectional edge in the second graph data, the operation of uniquely determining the communication strategy of the unidirectional communication between each satellite includes: inputting the edge feature of the unidirectional edge into a fully connected layer and a communication strategy classification unit based on binary classification to determine the communication strategy uniquely corresponding to the unidirectional edge.

[0094] As mentioned earlier, reference Figure 7 As shown, the ground station 100 sets each unidirectional edge E k The corresponding edge feature G k Input to the pre-set fully connected layer and binary classifier to uniquely determine the unidirectional edge E k The corresponding communication type. For example, the binary classifier outputs a two-dimensional classification vector Q k ,The 2D classification vector contains two elements, namely the probability of communicating through microwave link and the probability of communicating through laser link.

[0095] In addition, reference Figure 1 As shown, according to a second aspect of this embodiment, a storage medium is provided, wherein the storage medium includes a stored program, wherein when the program is run, a processor executes any one of the above methods.

[0096] Therefore, according to this embodiment, a multidimensional matrix-based feature extraction operation is first used to determine feasible communication strategies between each low-orbit satellite based on the distribution of multiple low-orbit satellites and the signal transmission environment. Then, a graph neural network-based message passing operation is used to further determine a unique communication strategy for one-way communication between each satellite based on the data transmission volume of the cell corresponding to each satellite and the one-way data transmission volume between each satellite, combined with the already determined feasible communication strategies between each low-orbit satellite. Therefore, through two decisions, this application combines the distribution of multiple low-orbit satellites and the signal transmission environment, as well as the data transmission volume and data reception volume of each cell, thereby taking into account the transmission advantages of microwave links and laser links and reasonably scheduling microwave links and laser links.

[0097] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that the present invention is not limited by the order of the actions described, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0098] Through the description of the above embodiments, those skilled in the art will clearly understand that the methods according to the above embodiments can be implemented using software plus the necessary general-purpose hardware platform. Of course, hardware can also be used, but in many cases the former is a more preferred embodiment. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, or optical disk) and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0099] Example 2

[0100] Figure 12 FIG2 shows a satellite communication scheduling device 1200 based on a microwave-laser dual link according to this embodiment, which corresponds to the method described in the first aspect of embodiment 1. Figure 12As shown, the device 1200 includes: a first matrix construction module 1210, configured to construct a first multidimensional matrix based on position information of multiple satellites, wherein the first multidimensional matrix is ​​used to indicate the positions between the satellites and position environment information related to signal communication; a second matrix determination module 1220, configured to perform feature extraction on the first multidimensional matrix to determine a second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate a communication strategy that can be implemented between the satellites, wherein the communication strategy includes: communication via microwave link and communication via laser link; a first graph data construction module 1230, configured to construct first graph data based on the second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate a communication strategy that can be implemented between the satellites, wherein the communication strategy includes: communication via microwave link and communication via laser link; The nodes of a graph data correspond to each satellite respectively, and the node attribute information is used to indicate the data transmission volume and data reception volume of the cell corresponding to each satellite, and the edges of the first graph data are used to indicate that communication can be carried out between two satellites, and the edge attributes are used to indicate the data transmission volume between the two satellites and the communication strategy that can be implemented; a message passing module 1240 is used to use a pre-trained graph neural network to perform message passing on the first graph data, thereby determining the corresponding second graph data; and a communication strategy scheduling module 1250 is used to uniquely determine the communication strategy for unidirectional communication between each satellite based on the edge feature information of the unidirectional edge in the second graph data.

[0101] Therefore, according to this embodiment, a multidimensional matrix-based feature extraction operation is first used to determine feasible communication strategies between each low-orbit satellite based on the distribution of multiple low-orbit satellites and the signal transmission environment. Then, a graph neural network-based message passing operation is used to further determine a unique communication strategy for one-way communication between each satellite based on the data transmission volume of the cell corresponding to each satellite and the one-way data transmission volume between each satellite, combined with the already determined feasible communication strategies between each low-orbit satellite. Therefore, through two decisions, this application combines the distribution of multiple low-orbit satellites and the signal transmission environment, as well as the data transmission volume and data reception volume of each cell, thereby taking into account the transmission advantages of microwave links and laser links and reasonably scheduling microwave links and laser links.

[0102] Example 3

[0103] Figure 13 FIG2 shows a satellite communication scheduling device 1300 based on a microwave-laser dual link according to this embodiment, which corresponds to the method described in the first aspect of embodiment 1. Figure 13As shown, the device 1300 includes: a processor 1310; and a memory 1320, connected to the processor 1310, for providing the processor 1310 with instructions for processing the following processing steps: constructing a first multidimensional matrix based on position information of multiple satellites, the first multidimensional matrix being used to indicate the positions between each satellite and position environment information related to signal communication; performing feature extraction on the first multidimensional matrix to determine a second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate communication strategies that can be implemented between each satellite, wherein the communication strategies include: communication via microwave links and communication via laser links; constructing first graph data based on the second multidimensional matrix, wherein nodes of the first graph data correspond to each satellite, node attribute information is used to indicate the data transmission volume and data reception volume of the cell corresponding to each satellite, and edges of the first graph data are used to indicate that two satellites can communicate, and edge attributes are used to indicate the data transmission volume between the two satellites and the communication strategies that can be implemented; using a pre-trained graph neural network to perform message passing on the first graph data to determine corresponding second graph data; and uniquely determining the communication strategy for unidirectional communication between each satellite based on edge feature information of unidirectional edges in the second graph data.

[0104] Therefore, according to this embodiment, a multidimensional matrix-based feature extraction operation is first used to determine feasible communication strategies between each low-orbit satellite based on the distribution of multiple low-orbit satellites and the signal transmission environment. Then, a graph neural network-based message passing operation is used to further determine a unique communication strategy for one-way communication between each satellite based on the data transmission volume of the cell corresponding to each satellite and the one-way data transmission volume between each satellite, combined with the already determined feasible communication strategies between each low-orbit satellite. Therefore, through two decisions, this application combines the distribution of multiple low-orbit satellites and the signal transmission environment, as well as the data transmission volume and data reception volume of each cell, thereby taking into account the transmission advantages of microwave links and laser links and reasonably scheduling microwave links and laser links.

[0105] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0106] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0107] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0108] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0109] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0110] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), a mobile hard drive, a magnetic disk, or an optical disk.

[0111] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A satellite communication scheduling method based on dual links, characterized in that: include: constructing a first multidimensional matrix based on position information of a plurality of satellites, wherein the first multidimensional matrix is ​​used to indicate positions between the satellites and position environment information related to signal communication; Performing feature extraction on the first multidimensional matrix to determine a second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate communication strategies that can be implemented between satellites, wherein the communication strategies include: communication via microwave links and communication via laser links; Constructing first graph data based on the second multidimensional matrix, wherein nodes of the first graph data correspond to respective satellites, and node attribute information is used to indicate a data transmission volume and a data reception volume of a cell corresponding to each satellite, and edges of the first graph data are used to indicate that communication between two satellites is possible, and edge attributes are used to indicate a data transmission volume and an implementable communication strategy between the two satellites; Performing message passing on the first graph data using a pre-trained graph neural network to determine corresponding second graph data; and A communication strategy for unidirectional communication between satellites is uniquely determined based on the edge feature information of the unidirectional edges in the second graph data.

2. The method according to claim 1, characterized in that The operation of constructing a first multi-dimensional matrix according to position information of a plurality of satellites includes: Determine the distance between each satellite and the orbital inclination difference based on the position information of the plurality of satellites; Determining atmospheric environment information between the satellites based on the position information of the plurality of satellites; and The first multidimensional matrix is ​​constructed according to the distances between the plurality of satellites, the orbital inclination differences and the atmospheric environment information, and wherein, The operation of determining atmospheric environment information between the satellites based on the position information of the plurality of satellites includes: Divide the atmospheric space into multiple grids; Determining a grid corresponding to each satellite according to the position information of the plurality of satellites; Determine the atmospheric environment information of the inter-satellite grid through which the straight line connecting each satellite passes; and The atmospheric environment information between the satellites is determined according to the atmospheric environment information of the inter-satellite grid.

3. The method according to claim 1, characterized in that The operation of performing feature extraction on the first multidimensional matrix to determine a second multidimensional matrix includes: Inputting the first multidimensional matrix into a matrix conversion model, and generating a third multidimensional matrix corresponding to the first multidimensional matrix through the matrix conversion model, wherein the third multidimensional matrix includes four channels, a first channel and a second channel are used to indicate whether microwave link communication can be implemented between satellites, and a third channel and a fourth channel are used to indicate whether laser link communication can be implemented between satellites; Using a first classifier to classify elements corresponding to the first channel and the second channel, and determine a probability that microwave link communication can be implemented between the satellites; Using a second classifier to classify the elements corresponding to the third channel and the fourth channel, and determine the probability that laser link communication can be implemented between the satellites; and The results output by the first classifier and the second classifier are used as the second multi-dimensional matrix.

4. The method according to claim 3, characterized in that The matrix conversion model includes an encoding network and a decoding network, wherein the encoding network is used to encode the first multidimensional matrix to generate a corresponding feature map, and the decoding network is used to decode the feature map to generate the third multidimensional matrix, and wherein The encoding network and the decoding network also perform feature fusion via skip connections.

5. The method according to claim 4, characterized in that The encoding network includes a plurality of encoding modules arranged in cascade, wherein each encoding module includes a convolutional layer and a pooling layer; The decoding network includes a plurality of decoding modules arranged in cascade, wherein each decoding module includes a deconvolution layer and an upsampling layer; and The feature maps output by each encoding module are fused with the feature maps to be input to the corresponding decoding module through jump connections.

6. The method according to claim 4, characterized in that An operation of performing message passing on the first graph data using a pre-trained graph neural network to determine corresponding second graph data includes performing multi-layer message passing on the first graph data to determine corresponding second graph data, and wherein a single message passing includes: Fusion generates a first fused feature based on the node information of the target node to be transferred and the edge information of the unidirectional edge with the target node as the source node; and According to the first fusion feature, a first multilayer perceptron is used to generate updated node information corresponding to the target node, and wherein A single message delivery also includes: Fusion generates a second fused feature based on the edge information of the target unidirectional edge to be transferred and the node information of the source node and the destination node associated with the unidirectional edge; and A second multilayer perceptron is used to generate updated edge information corresponding to the target unidirectional edge based on the second fused feature.

7. The method according to claim 1, characterized in that The operation of uniquely determining a communication strategy for unidirectional communication between satellites based on edge feature information of the unidirectional edges in the second graph data includes: The edge features of the unidirectional edge are input into the fully connected layer and the communication strategy classification unit based on binary classification to determine the communication strategy uniquely corresponding to the unidirectional edge.

8. A storage medium, characterized in that: The storage medium includes a stored program, wherein when the program is executed, the processor executes the method according to any one of claims 1 to 7.

9. A satellite communication scheduling device based on dual links, characterized in that: include: A first matrix construction module is configured to construct a first multidimensional matrix based on position information of a plurality of satellites, wherein the first multidimensional matrix is ​​configured to indicate positions between the satellites and position environment information related to signal communication; a second matrix determination module, configured to perform feature extraction on the first multidimensional matrix to determine a second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate communication strategies that can be implemented between satellites, wherein the communication strategies include: communication via microwave links and communication via laser links; a first graph data construction module, configured to construct first graph data based on the second multidimensional matrix, wherein nodes of the first graph data correspond to respective satellites, node attribute information is used to indicate a data transmission volume and a data reception volume of a cell corresponding to each satellite, and edges of the first graph data are used to indicate that two satellites can communicate with each other, and edge attributes are used to indicate a data transmission volume and an implementable communication strategy between the two satellites; a message passing module, configured to perform message passing on the first graph data using a pre-trained graph neural network, thereby determining corresponding second graph data; and The communication strategy scheduling module is used to uniquely determine the communication strategy for unidirectional communication between each satellite based on the edge feature information of the unidirectional edge in the second graph data.

10. A satellite communication scheduling device based on dual links, characterized in that: include: processor; as well as A memory, connected to the processor, configured to provide the processor with instructions for processing the following processing steps: constructing a first multidimensional matrix based on position information of a plurality of satellites, wherein the first multidimensional matrix is ​​used to indicate positions between the satellites and position environment information related to signal communication; Performing feature extraction on the first multidimensional matrix to determine a second multidimensional matrix, wherein the second multidimensional matrix is ​​used to indicate communication strategies that can be implemented between satellites, wherein the communication strategies include: communication via microwave links and communication via laser links; Constructing first graph data based on the second multidimensional matrix, wherein nodes of the first graph data correspond to respective satellites, and node attribute information is used to indicate a data transmission volume and a data reception volume of a cell corresponding to each satellite, and edges of the first graph data are used to indicate that communication between two satellites is possible, and edge attributes are used to indicate a data transmission volume and an implementable communication strategy between the two satellites; Performing message passing on the first graph data using a pre-trained graph neural network to determine corresponding second graph data; and A communication strategy for unidirectional communication between satellites is uniquely determined based on the edge feature information of the unidirectional edges in the second graph data.

Images