Satellite switching model training method, satellite switching method and related equipment

Abstract

The embodiment of the invention provides a training method of a satellite switching model, a satellite switching method and related equipment, and relates to the technical field of computers. The method comprises the steps that communication sample information is acquired, and the communication sample information is used for indicating parameter information involved in communication between terminal equipment and a satellite; dividing the communication sample information into first sample information and second sample information; training a variational automatic encoder according to the first sample information to obtain a trained variational automatic encoder; training a self-expression variational automatic encoder according to the second sample information and the trained variational automatic encoder to obtain the trained self-expression variational automatic encoder; and obtaining a satellite switching model according to the trained self-expression variational automatic encoder.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a satellite switching model training method, a satellite switching method, a satellite switching model training device, a satellite switching device, an electronic device, a computer-readable storage medium, and a computer program product. Background Art

[0002] In order to meet the growing demand for broadband wireless access, satellite constellations are developing in the direction of scale, and it is expected that a huge constellation system consisting of tens of thousands of satellites will be built in the future. In this trend, low-orbit satellites have attracted widespread attention from academia and industry due to their low cost and low latency.

[0003] However, due to the characteristics of high angular velocity and high-speed movement of low-orbit satellites, the visibility time with terminal devices is short and a single satellite cannot provide continuous service. In addition, as the number of terminal devices increases, there are problems such as difficulty in satellite selection, high cost of inter-satellite switching decision-making and unbalanced satellite load, which lead to low efficiency of inter-satellite switching and low utilization of satellite resources. Summary of the invention

[0004] The embodiments of the present disclosure provide a satellite switching model training method, a satellite switching method, a satellite switching model training device, a satellite switching device, an electronic device, a computer-readable storage medium, and a computer program product.

[0005] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0006] According to one aspect of the present disclosure, a method for training a satellite switching model is provided, the method comprising: acquiring communication sample information; the communication sample information is used to indicate parameter information involved in communication between a terminal device and a satellite; dividing the communication sample information into first sample information and second sample information; training a variational autoencoder according to the first sample information to obtain a trained variational autoencoder; training a self-expressive variational autoencoder according to the second sample information and the trained variational autoencoder to obtain a trained self-expressive variational autoencoder; and obtaining a satellite switching model according to the trained self-expressive variational autoencoder.

[0007] In some embodiments of the present disclosure, training a variational autoencoder based on the first sample information to obtain a trained variational autoencoder includes: encoding the first sample information based on the variational autoencoder to obtain reconstruction information of the first sample information; constructing a first loss function based on the first sample information and the reconstruction information of the first sample information; and adjusting parameters of the variational autoencoder based on the first loss function to obtain the trained variational autoencoder.

[0008] In some embodiments of the present disclosure, the self-expressive variational autoencoder is trained according to the second sample information and the trained variational autoencoder to obtain the trained self-expressive variational autoencoder, including: initializing the self-expressive variational autoencoder using the trained variational autoencoder to obtain the initialized self-expressive variational autoencoder; and training the initialized self-expressive variational autoencoder according to the second sample information to obtain the trained self-expressive variational autoencoder.

[0009] In some embodiments of the present disclosure, the method of training the initialized self-expressive variational autoencoder based on the second sample information to obtain the trained self-expressive variational autoencoder includes: encoding the second sample information based on the initialized self-expressive variational autoencoder to obtain reconstruction information of the second sample information; constructing a second loss function based on the second sample information and the reconstruction information of the second sample information; and adjusting the parameters of the initialized self-expressive variational autoencoder based on the second loss function to obtain the trained self-expressive variational autoencoder.

[0010] In some embodiments of the present disclosure, the communication sample information includes one or more of the following options: evaluation information of the terminal device on the satellite network service, network bandwidth information required by the terminal device, link distance information between the terminal device and the satellite, connection information between the terminal device and the satellite, transmission path information between the terminal device and the satellite, the number of idle channels of the satellite, the signal strength of the satellite, and the power spectrum density information of the satellite.

[0011] In some embodiments of the present disclosure, after acquiring the communication sample information, the method further includes: performing standardization processing on the communication sample information to obtain the standardized sample information, so as to perform training using the standardized sample information.

[0012] According to another aspect of the present disclosure, a satellite switching method is provided, the method comprising: acquiring communication information of a terminal device to be switched; transmitting the communication information of the terminal device to be switched to a satellite switching model to obtain a satellite switching decision result of the device to be switched; the satellite switching model is trained according to the training method of the satellite switching model described in the above embodiment; and executing the satellite switching decision result of the device to be switched.

[0013] According to another aspect of the present disclosure, a training device for a satellite switching model is provided, the device comprising: a sample acquisition module, configured to acquire communication sample information; the communication sample information is used to indicate parameter information involved in communication between a terminal device and a satellite; a sample processing module, configured to divide the communication sample information into first sample information and second sample information; a first training module, configured to train a variational autoencoder according to the first sample information to obtain a trained variational autoencoder; a second training module, configured to train a self-expressive variational autoencoder according to the second sample information and the trained variational autoencoder to obtain a trained self-expressive variational autoencoder; and a model generation model, configured to obtain a satellite switching model according to the trained self-expressive variational autoencoder.

[0014] According to another aspect of the present disclosure, a satellite switching device is provided, the device comprising: an information acquisition module, configured to acquire communication information of a terminal device to be switched; a satellite switching module, configured to transmit the communication information of the terminal device to be switched to a satellite switching model, and obtain a satellite switching decision result of the device to be switched; the satellite switching model is trained according to the training method of the satellite switching model described in the above embodiment; and the satellite switching decision result of the device to be switched is executed.

[0015] According to another aspect of the present disclosure, an electronic device is provided, comprising: one or more processors; and a storage device configured to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement a training method for a satellite switching model as described above, or implement the above-mentioned satellite switching method.

[0016] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the program implements the training method of the satellite switching model as described above, or implements the satellite switching method as described above.

[0017] According to another aspect of the present disclosure, a computer program product is provided, including: a computer program or instructions, which, when executed by a processor, implements the training method of the satellite switching model or the satellite switching method.

[0018] In the training method of the satellite switching model of the embodiment of the present disclosure, the variational autoencoder is first trained, and then the trained variational autoencoder is used to further train the self-expressive variational autoencoder, so that the self-expressive variational autoencoder obtains effective parameters through the initialization parameters in its network architecture (provided by the trained variational autoencoder) and subsequent training, and can efficiently learn the features of the high-dimensional vector constructed by the terminal device and the satellite. When the self-expressive variational autoencoder reaches convergence, it can not only generate samples with similar features to the constructed vector, but also successfully capture the effective feature association between the terminal device and the satellite. The satellite switching model finally generated in this way can intelligently generate satellite switching decision results, thereby effectively solving the problems of difficult satellite selection, high cost of inter-satellite switching decisions, and unbalanced satellite load, and improving the inter-satellite switching efficiency and satellite resource utilization.

[0019] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.

[0021] Figure 1 A schematic diagram of an exemplary application system architecture to which the data processing method in the embodiment of the present disclosure can be applied is shown;

[0022] Figure 2 A flow chart of a training method for a satellite switching model in an embodiment of the present disclosure is shown;

[0023] Figure 3 A diagram showing the training process of VAE in an embodiment of the present disclosure is shown;

[0024] Figure 4 A diagram showing the training process of SE-VAE in an embodiment of the present disclosure is shown;

[0025] Figure 5 A flow chart of another satellite switching model training method according to an embodiment of the present disclosure is shown;

[0026] Figure 6 A flow chart of a satellite switching method according to an embodiment of the present disclosure is shown;

[0027] Figure 7 A schematic diagram of the structure of a satellite switching model training device in an embodiment of the present disclosure is shown;

[0028] Figure 8A schematic structural diagram of a satellite switching device in an embodiment of the present disclosure is shown;

[0029] Fig. 9 A structural block diagram of an electronic device in an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0030] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the disclosure will be more comprehensive and complete and to fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0031] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0032] It should be noted that the ordinal numbers such as "first" and "second" mentioned in the embodiments of the present disclosure are used to distinguish multiple objects, and are not used to limit the order, timing, priority or importance of multiple objects, and the descriptions of "first" and "second" do not limit the objects to be different.

[0033] It should be noted that the collection, collection, update, analysis, processing, use, transmission, storage and other aspects of the customer information involved in this disclosure are in compliance with the relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken for customer information to prevent illegal access to customer information data and maintain customer information security, network security and national security.

[0034] Figure 1 FIG. 1 is a schematic diagram showing an exemplary application system architecture to which the training method of the satellite switching model or the satellite switching method in the embodiment of the present disclosure can be applied. Figure 1 As shown, the system architecture may include a satellite ground station system 110 , a terminal device 120 , and a satellite 130 .

[0035] The satellite ground station system 110 can be a vital component of the satellite communication network, acting as a bridge between the satellite and the ground network. The satellite ground station system 110 can be located in a fixed geographical location and has powerful communication equipment and antennas for stable and efficient communication with the satellite in orbit. The main functions of the satellite ground station system 110 include: sending data, instructions or control signals from the ground network to the satellite 130; receiving downlink signals from the satellite 130, which may contain data collected by the satellite (such as remote sensing images, meteorological data), user communication data (such as telephone calls, Internet data packets) or satellite status information. The satellite ground station system 110 is also responsible for the management and maintenance of the satellite network, including satellite orbit tracking, status monitoring, troubleshooting, and coordination with other ground stations or network management centers.

[0036] The terminal device 120 can be various electronic devices, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, smart speakers, smart watches, wearable devices, augmented reality devices, virtual reality devices, etc.

[0037] Optionally, the client of the application installed in different terminal devices 120 is the same, or the client of the same type of application based on different operating systems. Based on the different terminal platforms, the specific form of the client of the application can also be different, for example, the application client can be a mobile client, a PC client, etc.

[0038] Satellite 130 may be a space vehicle orbiting the Earth, equipped with communication equipment, antennas and other necessary sensors or instruments for establishing a communication link with a ground station, other satellites or terminal equipment. Satellite 130 plays a vital role in the satellite communication network, which can cover a wide area on the Earth and provide a variety of services such as telecommunications, broadcasting, navigation, telemetry and remote sensing.

[0039] Those skilled in the art will know that Figure 1 The number of satellite ground station systems, terminal devices, and satellites in the embodiment is only for illustration, and any number of satellite ground station systems, terminal devices, and satellites may be provided according to actual needs. The embodiments of the present disclosure are not limited to this.

[0040] Under the above system architecture, a training method for a satellite switching model is provided in an embodiment of the present disclosure. The method can be executed by any electronic device with computing and processing capabilities.

[0041] Figure 2 A flow chart of a training method for a satellite switching model in an embodiment of the present disclosure is shown. Figure 2 In the embodiment, the execution subject of the method can be any electronic device with data processing and transmission capabilities, such as Figure 1 The satellite ground station system 110 in the embodiment is also as follows Figure 1 In the embodiment, the satellite ground station system 110, the terminal device 120 and the satellite 130 jointly implement the training method of the satellite switching model, but the present disclosure is not limited thereto.

[0042] like Figure 2 As shown, the training method of the satellite switching model provided in the embodiment of the present disclosure includes the following steps.

[0043] Step S210, obtaining communication sample information; the communication sample information is used to indicate parameter information involved in the communication between the terminal device and the satellite.

[0044] In the disclosed embodiment, the communication sample information refers to various parameter information involved when the terminal device communicates with the satellite, and the parameter information can reflect the communication status, quality and demand between the terminal device and the satellite.

[0045] In some embodiments of the present disclosure, the communication sample information includes one or more of the following options: evaluation information of the terminal device on the satellite network service, network bandwidth information required by the terminal device, link distance information between the terminal device and the satellite, connection information between the terminal device and the satellite, transmission path information between the terminal device and the satellite, the number of idle channels of the satellite, the signal strength of the satellite, and the power spectrum density information of the satellite.

[0046] The evaluation information of the terminal device on the satellite network service may include evaluation indicators such as the terminal device's satisfaction with the satellite network service, network delay, packet loss rate, etc. This information helps to understand the terminal device's evaluation and demand for satellite network services.

[0047] The network bandwidth information required by the terminal device is used to indicate the size of the network bandwidth required by the terminal device when communicating. This information helps the satellite network allocate appropriate network resources to the terminal device.

[0048] The link distance information between the terminal device and the satellite refers to the physical distance between the terminal device and the satellite. This information is of great significance for evaluating communication quality, predicting signal attenuation, etc., and may affect signal strength and communication quality.

[0049] The connection information between the terminal device and the satellite may include connection status (such as connected, disconnected, etc.), connection type (such as TCP connection, UDP connection, direct connection, relay connection, etc.) and connection quality (such as signal strength, bit error rate, etc.). This information helps to understand the communication connection between the terminal device and the satellite.

[0050] The transmission path information between the terminal device and the satellite refers to the channel for data transmission between the terminal device and the satellite, which affects the communication delay and stability.

[0051] The number of idle channels of a satellite is used to indicate the number of channels currently available to the satellite and is used to evaluate the communication capacity of the satellite.

[0052] The satellite's signal strength and power spectrum density information reflects the strength and spectrum characteristics of the satellite's transmitted signals, affecting communication quality and coverage.

[0053] It should be noted that, in the embodiments of the present disclosure, the communication sample information may include other information in addition to the above information, such as the moving speed information of the terminal device, and the present disclosure does not limit this.

[0054] In the disclosed embodiment, the communication sample information includes parameter information such as the terminal device's evaluation of the satellite network service, the required network bandwidth, link distance, connection status, transmission path, number of satellite idle channels, signal strength and power spectrum density. By integrating these diverse parameter information into a high-dimensional vector, a comprehensive and rich data foundation is provided for the training of the satellite switching model, which can significantly improve the model's prediction accuracy and decision-making ability.

[0055] In some embodiments of the present disclosure, after acquiring the communication sample information, the method further includes: performing standardization processing on the communication sample information to obtain the standardized sample information, so as to perform training using the standardized sample information.

[0056] After obtaining the communication sample information, the communication sample information can be standardized, which may include but is not limited to data cleaning, data conversion, data normalization and data standardization. Among them, data cleaning refers to removing duplicate, invalid or abnormal data to ensure the accuracy and reliability of the data. Data conversion refers to converting raw data into a format or type suitable for subsequent processing and analysis. Data normalization refers to converting data of different dimensions into the same dimension for comparison and analysis. Data standardization refers to adjusting the distribution of data to a standard normal distribution or other specified distribution form for subsequent statistical analysis or machine learning algorithm training.

[0057] In the disclosed embodiment, the acquired communication sample information is standardized, and the standardized sample information is subsequently used for model training to ensure the consistency of different information on the numerical scale and reduce the impact of dimensional differences on model training.

[0058] Step S220: dividing the communication sample information into first sample information and second sample information.

[0059] In an embodiment of the present disclosure, after obtaining the communication sample information, it is divided into first sample information and second sample information. This division is usually based on a preset division strategy, such as random division, time window-based division, etc., to ensure that the two subsets are as consistent as possible in data distribution, while being independent of each other for subsequent model training.

[0060] Step S230: training a variational autoencoder according to the first sample information to obtain a trained variational autoencoder.

[0061] Among them, the Variational AutoEncoder (VAE) is a generative model based on deep learning. It operates through a dual process of encoding and decoding to learn and understand the latent variables in a given data set. By learning the latent distribution of the data, it can generate samples similar to the original data.

[0062] In the disclosed embodiment, the first sample information is used to train the VAE. During the training process, the VAE learns the latent space representation of the first sample information and learns how to generate new samples similar to the first sample information from this latent space.

[0063] Step S240: training a self-expressive variational autoencoder according to the second sample information and the trained variational autoencoder to obtain a trained self-expressive variational autoencoder.

[0064] Among them, Self-Expressive Variational AutoEncoder (SE-VAE) is a model that introduces self-expressive properties based on VAE, so that the model can not only learn the potential representation of the data, but also capture the correlation or similarity between data samples.

[0065] In the disclosed embodiment, the second sample information and the trained VAE are used to train SE-VAE. During the training process, SE-VAE not only learns the potential representation of the data, but also learns how to represent other samples with linear or nonlinear combinations in a latent space, thereby capturing the potential relationship between samples.

[0066] Step S250, obtaining a satellite switching model according to the trained self-expressive variational autoencoder.

[0067] In the disclosed embodiment, the trained SE-VAE is used to construct a satellite switching model. The satellite switching model can predict and decide the optimal satellite switching strategy based on the communication parameter information between the terminal device and the satellite (such as link quality, network bandwidth requirements, etc.). Exemplarily, the satellite switching model can make a switching decision by comparing the network service quality provided by different satellites and the communication requirements of the terminal device.

[0068] In the training method of the satellite switching model of the embodiment of the present disclosure, VAE is trained first, and then the trained VAE is used to further train SE-VAE, so that SE-VAE can efficiently learn the features of the high-dimensional vector constructed by the terminal device and the satellite through the initialization parameters in its network architecture (provided by the trained VAE) and subsequent training. When SE-VAE reaches convergence, it can not only generate samples with similar features to the constructed vector, but also successfully capture the effective feature association between the terminal device and the satellite. The satellite switching model finally generated in this way can intelligently generate satellite switching decision results, thereby effectively solving the problems of difficult satellite selection, high cost of inter-satellite switching decisions, and unbalanced satellite load, and improving the inter-satellite switching efficiency and satellite resource utilization.

[0069] Figure 3 FIG. 2 shows a diagram of the training process of VAE in an embodiment of the present disclosure. Figure 3 As shown, the following steps are included.

[0070] Step S310: encoding the first sample information based on VAE to obtain reconstructed information of the first sample information.

[0071] In the disclosed embodiment, the VAE includes an encoder and a decoder. The encoder is responsible for mapping the first sample information to a latent space and generating a latent vector (or latent vector), which is a low-dimensional representation of the first sample information and captures important features of the first sample information. Then, the decoder receives the latent vector as input and attempts to reconstruct an approximate version of the first sample information, that is, the reconstructed information of the first sample information.

[0072] Step S320: construct a first loss function according to the first sample information and the reconstruction information of the first sample information.

[0073] In the disclosed embodiment, a first loss function is constructed to evaluate the performance of VAE. Exemplarily, the first loss function may include reconstruction error and KL divergence (Kullback-Leibler Divergence). The reconstruction error is used to measure the difference between the reconstructed information and the original first sample information. The reconstruction error metric includes mean square error (MSE) or cross entropy loss. KL divergence is used to measure the difference between the distribution of the latent vector generated by VAE and the standard normal distribution.

[0074] The reconstruction error and KL divergence are combined to obtain the first loss function. The goal of this loss function is to minimize the reconstruction error and KL divergence, so that VAE can reconstruct the input data more accurately and generate a latent vector with good properties.

[0075] Step S330, adjust the parameters of the VAE based on the first loss function to obtain a trained VAE.

[0076] In the disclosed embodiment, an optimization algorithm is used to minimize the first loss function, thereby adjusting the parameters of the VAE, which include weights and biases in the encoder and decoder.

[0077] By iteratively updating these parameters, the value of the loss function is gradually reduced, so that the performance of VAE is improved. When the loss value of the calculated first loss function is less than the preset loss threshold p, the VAE is considered to have been trained, and the trained VAE is subsequently used to further train SE-VAE.

[0078] In the training method of the satellite switching model provided in the embodiment of the present disclosure, the first sample information is first used to train VAE. As a neural network model, the encoder of VAE can effectively map the complex multi-dimensional feature vectors constructed by the terminal device and the satellite into a low-dimensional latent space, thereby realizing dimensionality reduction and feature extraction, avoiding the problems of excessive computational complexity and dimensionality disaster caused by direct clustering in high-dimensional space, and providing strong support for the subsequent further training of SE-VAE, thereby improving the efficiency and accuracy of the overall model.

[0079] Figure 4 FIG. 4 shows a diagram of the training process of SE-VAE in an embodiment of the present disclosure. Figure 4 As shown, the following steps are included.

[0080] Step S410, using the trained VAE to initialize SE-VAE to obtain the initialized SE-VAE.

[0081] In the disclosed embodiment, the trained VAE is used to initialize SE-VAE, that is, the initial parameters of SE-VAE are set according to the parameters of the trained VAE.

[0082] Since SE-VAE not only inherits the dimensionality reduction and feature extraction capabilities of VAE, but also introduces self-expression characteristics, it further enhances the model's ability to capture the relationship between data points in the latent space, so that the feature vectors of terminal devices and satellites can be represented by linear combinations of other data points in the latent space, thereby achieving the effect of subspace clustering.

[0083] Step S420: train the initialized SE-VAE according to the second sample information to obtain a trained SE-VAE.

[0084] In some embodiments of the present disclosure, the initialized SE-VAE is trained according to the second sample information to obtain the trained SE-VAE, including: encoding the second sample information based on the initialized SE-VAE to obtain reconstruction information of the second sample information; constructing a second loss function according to the second sample information and the reconstruction information of the second sample information; adjusting the parameters of the initialized SE-VAE based on the second loss function to obtain the trained SE-VAE.

[0085] In the disclosed embodiment, the initialized self-SE-VAE is applied to the second sample information. The high-dimensional second sample information is mapped to a low-dimensional latent space through the SE-VAE encoder to generate a latent vector (also called a latent vector), which captures the key features in the second sample information while reducing the dimension of the data. The SE-VAE decoder receives this latent vector as input and attempts to map it back to the original high-dimensional space to generate reconstructed information of the second sample information. This reconstructed information should be as close to the original second sample information as possible, but may differ slightly due to information loss during encoding and decoding.

[0086] In order to evaluate the performance of SE-VAE to determine whether it can accurately reconstruct the second sample information, a second loss function is constructed, which is used to measure the difference between the original second sample information and its reconstructed information. Among them, the loss function type can include mean square error (MSE), cross entropy loss, etc.

[0087] In addition, since SE-VAE introduces self-expressive properties, it is based on the assumption that each data point can be represented by a linear combination of other data points in the latent space. During the training process of SE-VAE, the self-expressive property can be achieved by optimizing a loss function containing a self-expressive regularization term, that is, adding a regularization term to the second loss function to encourage the linear relationship between latent vectors through the regularization term. In other words, this regularization term can help SE-VAE better capture the intrinsic relationship between data points, thereby achieving more accurate subspace clustering.

[0088] After determining the second loss function, an optimization algorithm is used to minimize the value of the second loss function. By iteratively updating the parameters of SE-VAE, the value of the loss function can be gradually reduced, thereby improving the performance of SE-VAE. Specifically, in each iteration, the gradient of the second loss function under the current parameters is calculated, and this gradient is used to update the parameters. This process will continue until the loss value of the second loss function is less than the preset loss threshold q, then SE-VAE is considered to have been trained, and a trained SE-VAE is obtained.

[0089] In the training method of the satellite switching model of the embodiment of the present disclosure, SE-VAE can effectively learn the features of the constructed high-dimensional vector. The VAE initialization parameters learned through pre-training are first used, and then SE-VAE is trained to obtain its effective parameters. When SE-VAE reaches convergence, samples with features similar to the constructed vector can be generated. In this way, SE-VAE can learn the effective features of the terminal device and the satellite, and then through the self-expression characteristics of SE-VAE, the linear relationship between the potential vectors can be used to formulate a switching strategy.

[0090] Figure 5 FIG. 2 is a flow chart showing another satellite switching model training method according to an embodiment of the present disclosure. Figure 5 As shown, the training method of the satellite switching model provided in the embodiment of the present disclosure includes the following steps.

[0091] Step S501: Acquire communication sample information, where the communication sample information is used to indicate parameter information involved in communication between a terminal device and a satellite.

[0092] The communication sample information includes one or more of the following options: evaluation information of the terminal device on the satellite network service, network bandwidth information required by the terminal device, link distance information between the terminal device and the satellite, connection information between the terminal device and the satellite, transmission path information between the terminal device and the satellite, number of idle channels of the satellite, signal strength of the satellite and power spectrum density information of the satellite.

[0093] Step S502: standardize the communication sample information and then divide it into first sample information and second sample information.

[0094] Step S503, setting the initial parameters of the VAE, and inputting the first sample information into the VAE for pre-training.

[0095] During the process of inputting the first sample information into the VAE for pre-training, the first sample information is encoded based on the VAE to obtain reconstruction information of the first sample information, and the loss value of the first loss function is calculated based on the first sample information and the reconstruction information of the first sample information.

[0096] Step S504, determining whether the loss value of the first loss function is less than a preset loss threshold p. If the loss value of the first loss function is greater than or equal to the preset loss threshold p, returning to step S503.

[0097] Step S505, if the loss value of the first loss function is less than the preset loss threshold p, it is confirmed that the VAE training is completed, and the parameters of the trained VAE are used as the initial parameters of SE-VAE.

[0098] Step S506: input the second sample information into SE-VAE for pre-training.

[0099] During the process of inputting the second sample information into SE-VAE for training, reconstruction information of the second sample information is obtained, and the loss value of the second loss function is calculated according to the second sample information and the reconstruction information of the second sample information.

[0100] Step S507, determining whether the loss value of the second loss function is less than a preset loss threshold q. If the loss value of the second loss function is greater than or equal to the preset loss threshold q, returning to step S506.

[0101] Step S508: If the loss value of the second loss function is less than the preset loss threshold q, it is confirmed that the SE-VAE training is completed, and the satellite switching model is generated using the trained SE-VAE.

[0102] In summary, the communication sample information includes parameter information such as the terminal equipment's evaluation of satellite network services, required network bandwidth, link distance, connection status, transmission path, number of satellite idle channels, signal strength and power spectrum density. By integrating these diverse parameter information into high-dimensional vectors, a comprehensive and rich data foundation is provided for the training of the satellite switching model, which can significantly improve the model's prediction accuracy and decision-making ability.

[0103] Also, first train the VAE, and then use the trained VAE to further train the SE-VAE, so that the SE-VAE can efficiently learn the features of the high-dimensional vector constructed by the terminal device and the satellite through the initialization parameters in its network architecture (provided by the trained VAE) and subsequent training to obtain effective parameters. When the SE-VAE reaches convergence, it can not only generate samples with similar features to the constructed vector, but also successfully capture the effective feature associations between the terminal device and the satellite. The satellite switching model finally generated can intelligently generate satellite switching decision results, thereby effectively solving the problems of difficult satellite selection, high cost of inter-satellite switching decisions, and unbalanced satellite load, and improving the inter-satellite switching efficiency and satellite resource utilization.

[0104] Under the above system architecture, a satellite switching method is provided in an embodiment of the present disclosure, and the method can be executed by any electronic device with computing and processing capabilities.

[0105] Figure 6 A flow chart of a satellite switching method according to an embodiment of the present disclosure is shown. Figure 6 In the embodiment, the execution subject of the method can be any electronic device with data processing and transmission capabilities, such as Figure 1 The satellite ground station system 110 in the embodiment is also as follows Figure 1 The satellite ground station system 110 , the terminal device 120 , and the satellite 130 in the embodiment jointly implement the satellite switching method, but the present disclosure is not limited thereto.

[0106] like Figure 6 As shown, the satellite switching method provided in the embodiment of the present disclosure includes the following steps.

[0107] Step S610, obtaining communication information of the terminal device to be switched.

[0108] In the disclosed embodiment, the terminal device to be switched refers to a terminal device whose current connection status with the satellite can no longer meet its service requirements.

[0109] In the disclosed embodiment, the communication information of the terminal device may include location information, motion status, currently connected satellite information, and communication demand information. Among them, the location information of the terminal device helps to determine the current geographical area of ​​the terminal device, so as to select the most suitable satellite for communication. The motion status of the terminal device may affect the timing and method of satellite switching. The currently connected satellite information may include the identification, signal strength, communication quality, etc. of the current satellite, which is helpful in evaluating the urgency of switching. The communication demand information may include the current communication data volume, data transmission rate requirements, delay requirements, etc. of the terminal device, which will affect the formulation of satellite switching decisions.

[0110] Step S620, transmitting the communication information of the terminal device to be switched to the satellite switching model to obtain the satellite switching decision result of the device to be switched; the satellite switching model is trained according to the satellite switching model training method of the above embodiment.

[0111] In the disclosed embodiment, the communication information of the terminal device to be switched is transmitted to the satellite switching model, and the satellite switching model outputs the satellite switching decision result corresponding to the device to be switched. The satellite switching model is obtained by training according to the training method of the satellite switching model of the above embodiment. The above embodiment has described the model training process in detail, which will not be repeated here.

[0112] In some embodiments of the present disclosure, the satellite switching decision result of the device to be switched includes one or more of the following options: identification information of the target satellite, switching time information of the target satellite, switching parameter information corresponding to the target satellite, and service quality estimation information corresponding to the target satellite; wherein, the target satellite is the satellite to be accessed by the device to be switched by performing a switching operation.

[0113] Among them, the identification information of the target satellite is used to determine which satellite to switch to. The switching time information of the target satellite indicates when the satellite switching operation will be performed, which may be an immediate switch or at a future time point. The switching parameter information corresponding to the target satellite includes the parameters required in the switching process, including but not limited to technical parameters such as channel allocation and transmission path adjustment used in the switching process. The service quality estimation information corresponding to the target satellite refers to the prediction of the communication quality after switching to the target satellite, including but not limited to the estimation of indicators such as signal strength, data transmission rate, and delay.

[0114] Step S630, executing the satellite switching decision result of the device to be switched.

[0115] Exemplarily, a switching instruction is sent to the terminal device, instructing it to switch to a specified target satellite. During the switching process, changes in communication quality can be monitored to ensure that the switching proceeds smoothly. After the switching is completed, the status information of the terminal device is updated, including the currently connected satellite, communication quality, etc.

[0116] In the satellite switching method of the disclosed embodiment, a satellite switching decision result is intelligently generated for the device to be switched based on a satellite switching model, ensuring that the terminal device always enjoys the best communication connection, helping to balance the load between satellites, and effectively solving problems such as difficult satellite selection, high cost of inter-satellite switching decisions, and unbalanced satellite loads, thereby improving inter-satellite switching efficiency and satellite resource utilization.

[0117] Figure 7 FIG. 2 is a schematic diagram showing the structure of a satellite switching model training device according to an embodiment of the present disclosure. Figure 7 As shown, the device 700 includes: a sample acquisition module 710, a sample processing module 720, a first training module 730, a second training module 740 and a model generation module 750.

[0118] The sample acquisition module 710 is configured to: acquire communication sample information; wherein the communication sample information is used to indicate parameter information involved in the communication between the terminal device and the satellite. The sample processing module 720 is configured to: divide the communication sample information into first sample information and second sample information. The first training module 730 is configured to: train the VAE according to the first sample information to obtain the trained VAE. The second training module 740 is configured to: train the SE-VAE according to the second sample information and the trained VAE to obtain the trained SE-VAE. The model generation model 750 is configured to: obtain the satellite switching model according to the trained SE-VAE.

[0119] In some embodiments of the present disclosure, the first training module 730 is further configured to: encode the first sample information based on VAE to obtain reconstruction information of the first sample information; construct a first loss function based on the first sample information and the reconstruction information of the first sample information; adjust the parameters of VAE based on the first loss function to obtain a trained VAE.

[0120] In some embodiments of the present disclosure, the second training module 740 is further configured to: use the trained VAE to initialize the SE-VAE to obtain the initialized SE-VAE; and train the initialized SE-VAE according to the second sample information to obtain the trained SE-VAE.

[0121] In some embodiments of the present disclosure, the second training module 740 is further configured to: encode the second sample information based on the initialized SE-VAE to obtain reconstruction information of the second sample information; construct a second loss function based on the second sample information and the reconstruction information of the second sample information; adjust the parameters of the initialized SE-VAE based on the second loss function to obtain a trained SE-VAE.

[0122] In some embodiments of the present disclosure, the communication sample information includes one or more of the following options: evaluation information of the terminal device on the satellite network service, network bandwidth information required by the terminal device, link distance information between the terminal device and the satellite, connection information between the terminal device and the satellite, transmission path information between the terminal device and the satellite, the number of idle channels of the satellite, the signal strength of the satellite, and the power spectrum density information of the satellite.

[0123] In some embodiments of the present disclosure, the sample processing module 720 is further configured to: perform standardization processing on the communication sample information to obtain the standardized sample information, so as to perform training using the standardized sample information.

[0124] Since the principle of solving the problem in the embodiment of the training device for the satellite switching model is similar to that in the above-mentioned method embodiment, the real-time implementation of the embodiment of the training device for the satellite switching model can refer to the implementation of the above-mentioned method embodiment, and the repeated parts will not be repeated.

[0125] Figure 8 FIG. 2 shows a schematic diagram of the structure of a satellite switching device in an embodiment of the present disclosure. Figure 8 As shown, the device 800 includes: an information acquisition module 810 and a satellite switching module 820.

[0126] The information acquisition module 810 is configured to: acquire the communication information of the terminal device to be switched. The satellite switching module 820 is configured to: transmit the communication information of the terminal device to be switched to the satellite switching model to obtain the satellite switching decision result of the device to be switched, wherein the satellite switching model is obtained by training according to the training method of the satellite switching model of the above embodiment; and execute the satellite switching decision result of the device to be switched.

[0127] In some embodiments of the present disclosure, the satellite switching decision result of the device to be switched includes one or more of the following options: identification information of the target satellite, switching time information of the target satellite, switching parameter information corresponding to the target satellite, and service quality estimation information corresponding to the target satellite; wherein, the target satellite is the satellite to be accessed by the device to be switched by performing a switching operation.

[0128] Since the principle of solving the problem in the satellite switching device embodiment is similar to that in the above method embodiment, the implementation of the satellite switching device embodiment can refer to the implementation of the above method embodiment, and the repeated parts will not be repeated.

[0129] Fig. 9 FIG. 1 shows a structural block diagram of an electronic device in an embodiment of the present disclosure. It should be noted that: Fig. 9 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0130] like Fig. 9 As shown, the electronic device 900 includes a central processing unit (CPU) 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage part 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the electronic device 900 are also stored. The CPU 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0131] The following components are connected to the I / O interface 905: an input section 906 including a keyboard, a mouse, etc.; an output section 907 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, a modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the I / O interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 910 as needed, so that a computer program read therefrom is installed into the storage section 908 as needed.

[0132] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part 909, and / or installed from a removable medium 911. When the computer program is executed by a central processing unit (CPU) 901, the above-mentioned functions defined in the system of the present disclosure are executed.

[0133] It should be noted that the computer-readable medium shown in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, terminal device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, a terminal device or a device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, a terminal device or a device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0134] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0135] The units involved in the embodiments described in the present disclosure may be implemented by software or by hardware. The units described may also be provided in a processor, for example, may be described as: a processor including a sample acquisition module, a sample processing module, a first training module, a second training module and a model generation module. The names of these modules do not, in some cases, constitute limitations on the modules themselves, for example, the sample acquisition module may also be described as a "module for acquiring communication sample information".

[0136] As another aspect, the present disclosure further provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or may exist independently without being assembled into the electronic device. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed by an electronic device, the electronic device implements the method described in the following embodiments. For example, the electronic device may implement the following Figure 2 The steps shown.

[0137] According to one aspect of the present disclosure, a computer program product or a computer program is provided, the computer program product or the computer program including computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in various optional implementations of the above-mentioned embodiments.

[0138] It should be understood that any number of elements in the drawings of the present disclosure is for illustration rather than limitation, and any naming is only for distinction rather than having any limiting meaning.

[0139] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0140] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A satellite switching model training method, characterized in that: The method comprises: Acquire communication sample information; the communication sample information is used to indicate parameter information involved in the communication between the terminal device and the satellite; dividing the communication sample information into first sample information and second sample information; Training a variational autoencoder according to the first sample information to obtain a trained variational autoencoder; According to the second sample information and the trained variational autoencoder, a self-expressive variational autoencoder is trained to obtain a trained self-expressive variational autoencoder; According to the trained self-expressive variational autoencoder, a satellite switching model is obtained.

2. The method according to claim 1, characterized in that The step of training a variational autoencoder according to the first sample information to obtain a trained variational autoencoder includes: Performing encoding processing on the first sample information based on the variational autoencoder to obtain reconstruction information of the first sample information; constructing a first loss function according to the first sample information and the reconstruction information of the first sample information; The parameters of the variational autoencoder are adjusted based on the first loss function to obtain the trained variational autoencoder.

3. The method according to claim 1, characterized in that The step of training a self-expressive variational autoencoder according to the second sample information and the trained variational autoencoder to obtain the trained self-expressive variational autoencoder includes: Initializing the self-expressive variational autoencoder using the trained variational autoencoder to obtain an initialized self-expressive variational autoencoder; The initialized self-expressive variational autoencoder is trained according to the second sample information to obtain the trained self-expressive variational autoencoder.

4. The method according to claim 3, characterized in that The step of training the initialized self-expressive variational autoencoder according to the second sample information to obtain the trained self-expressive variational autoencoder includes: Performing encoding processing on the second sample information based on the initialized self-expressive variational autoencoder to obtain reconstruction information of the second sample information; constructing a second loss function according to the second sample information and the reconstruction information of the second sample information; The parameters of the initialized self-expressive variational autoencoder are adjusted based on the second loss function to obtain the trained self-expressive variational autoencoder.

5. The method according to any one of claims 1 to 4, characterized in that: The communication sample information includes one or more of the following options: The terminal device’s evaluation information on satellite network services, the network bandwidth information required by the terminal device, the link distance information between the terminal device and the satellite, the connection information between the terminal device and the satellite, the transmission path information between the terminal device and the satellite, the number of idle channels of the satellite, the signal strength of the satellite and the power spectrum density information of the satellite.

6. The method according to claim 5, characterized in that After acquiring the communication sample information, the method further includes: The communication sample information is standardized to obtain standardized sample information, so as to perform training using the standardized sample information.

7. A satellite switching method, characterized in that: The method comprises: Obtaining communication information of the terminal device to be switched; Transmitting the communication information of the terminal device to be switched to the satellite switching model to obtain a satellite switching decision result of the device to be switched; the satellite switching model is trained according to the satellite switching model training method according to any one of claims 1 to 6; Execute the satellite switching decision result of the device to be switched.

8. The method according to claim 7, characterized in that The satellite switching decision result of the device to be switched includes one or more of the following options: Identification information of the target satellite, switching time information of the target satellite, switching parameter information corresponding to the target satellite and service quality estimation information corresponding to the target satellite; the target satellite is the satellite to be accessed by the device to be switched through the switching operation.

9. A training device for a satellite switching model, characterized in that: The device comprises: A sample acquisition module is configured to acquire communication sample information; the communication sample information is used to indicate parameter information involved in the communication between the terminal device and the satellite; A sample processing module, configured to divide the communication sample information into first sample information and second sample information; A first training module is configured to train a variational autoencoder according to the first sample information to obtain a trained variational autoencoder; A second training module is configured to train a self-expressive variational autoencoder according to the second sample information and the trained variational autoencoder to obtain a trained self-expressive variational autoencoder; The model generation model is configured to obtain a satellite switching model according to the trained self-expressive variational autoencoder.

10. A satellite switching device, characterized in that: The device comprises: An information acquisition module, configured to acquire communication information of a terminal device to be switched; A satellite switching module is configured to transmit the communication information of the terminal device to be switched to a satellite switching model to obtain a satellite switching decision result of the device to be switched; the satellite switching model is trained according to the training method of the satellite switching model according to any one of claims 1 to 6; and execute the satellite switching decision result of the device to be switched.

11. An electronic device, characterized in that: include: one or more processors; A storage device configured to store one or more programs, when the one or more programs are executed by the one or more processors, enables the one or more processors to implement the training method of the satellite switching model as described in any one of claims 1 to 6, or implement the satellite switching method as described in claim 7 or 8.

12. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the training method of the satellite switching model as described in any one of claims 1 to 6 is implemented, or the satellite switching method as described in claim 7 or 8 is implemented.

13. A computer program product comprising: A computer program or instruction, characterized in that when the computer program or instruction is executed by a processor, it implements the training method of the satellite switching model as described in any one of claims 1 to 6, or implements the satellite switching method as described in claim 7 or 8.

Images