Method and apparatus for accessing a device in a network system

Abstract

The application discloses a network system device access method and device, the method comprises the following steps: according to the first operating parameter value of each satellite, the second operating parameter value of the device and the historical connection information, the connection state of the device and each satellite at the target time is predicted, and the connection state prediction result is obtained; according to the connection state prediction result, at least two satellites are screened, the candidate satellite is obtained, and the candidate satellite and the ground base station are used as the candidate network set; according to the network attribute information of each candidate network in the candidate network set, the connection matching degree of the device and each candidate network is determined; according to the connection matching degree, the candidate network set is screened, and the target network is obtained; so that the device accesses the target network at the target time. The application improves the reliability of the ground device access network selection, and avoids multi-user competition.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a method and apparatus for accessing devices in a network system. Background Technology

[0002] With the evolution from 5G-Advanced to 6G, non-terrestrial networks (NTNs) have become a key component of the 3GPP standard. Low Earth Orbit (LEO) satellites, due to their low latency and wide coverage, have become the core carrier of space-terrestrial converged networks. Current technological development presents three major trends: 1) Intelligent handover decision-making: DQN-based handover algorithms reduce handover rates through five-dimensional state modeling of received signal strength (RSS), available bit rate (ABR), relative speed, latency, and network overhead; 2) Multi-objective joint optimization: Constructing a revenue function based on received signal strength, service time, channel utilization, and relay overhead to balance load and handover failure rate; 3) Adaptation to highly dynamic environments: Utilizing spatiotemporal attention mechanisms to predict access traffic and solve the problem of unclear periodic characteristics caused by satellite motion.

[0003] However, existing technologies still have problems: insufficient capture of spatiotemporal dynamics: LEO satellites reach a relative ground speed of 7 km / s, and the time-varying characteristics of the satellite-to-ground link make traditional periodic models fail, leading to misjudgment of switching timing; and in the scenario of a giant constellation of tens of thousands of satellites, competition for access by multiple users triggers signaling storms. Summary of the Invention

[0004] This application provides a method and apparatus for device access in a network system, which can improve the reliability of terrestrial devices in selecting network access and avoid multi-user contention.

[0005] On one hand, this application provides a method for device access in a network system, the network system including a ground base station, at least two satellites, and devices, the method including: The system acquires the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and historical connection information; the historical connection information represents the network accessed by the device at a historical moment and its historical network attribute information. Based on the first operating parameter value, the second operating parameter value, and the historical connection information of each satellite, the connection status between the device and each satellite at the target time is predicted to obtain the connection status prediction result. Based on the connection status prediction results, the at least two satellites are screened to obtain candidate satellites, and the candidate satellites and the ground base station are used as a candidate network set. Based on the network attribute information of each candidate network in the candidate network set, the connection matching degree between the device and each candidate network is determined; Based on the connection matching degree between the device and each candidate network, the candidate network set is filtered to obtain the target network; so that the device can access the target network at the target time.

[0006] In one exemplary embodiment, the network system includes at least two devices; before acquiring the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and historical connection information, the method further includes: Obtain the current satellite coordinates in the geocentric coordinate system, the preset coverage parameter value, and the current device coordinates of the target device in the geocentric coordinate system at the current time; the target device is any one of the at least two devices. Based on the current device coordinates and the current satellite coordinates of each satellite, determine the current coverage parameter values ​​between the target device and each satellite; The current coverage parameter value and the preset coverage parameter value are compared to obtain the coverage parameter comparison result between the target device and each satellite; Based on the coverage parameter comparison results, a current dynamic graph is constructed; the current dynamic graph includes at least two satellite nodes and at least two edges corresponding to each satellite node; each satellite node includes the current satellite coordinate value and preset coverage parameter value of each satellite; each edge represents the connection status between each satellite and the device; The acquisition of the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and historical connection information includes: Based on the current dynamic map, the current satellite coordinates and preset coverage parameter values ​​of each satellite are used as the first operating parameter values; Obtain the second operating parameter value of the target device and the historical connection information.

[0007] In one exemplary embodiment, determining the current coverage parameter value between the target device and each satellite based on the current device coordinate value and the current satellite coordinate value of each satellite includes: Based on the current device coordinates and the current satellite coordinates of the target satellite, determine the current distance between the target device and the target satellite; the target satellite is any one of the at least two satellites. Based on the rotation and translation parameters required for the target satellite to transform from the geocentric coordinate system to the target satellite coordinate system corresponding to the target satellite, coordinate transformation parameters are constructed. Based on the coordinate transformation parameters and the current device coordinates, determine the actual coordinates of the target device in the target satellite coordinate system; Calculate the angle between the actual coordinate value and the preset reference plane to obtain the current elevation angle; The actual coordinate values ​​are projected onto the preset reference plane to obtain reference coordinate values, and the angle between the reference coordinate values ​​and the preset reference axis is calculated to obtain the current azimuth angle. The current distance value, the current elevation angle, and the current azimuth angle are used as the current coverage parameter values.

[0008] In one exemplary embodiment, before obtaining the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and the historical connection information, the method further includes: Based on the channel between the target device and the target satellite, construct a channel fading value calculation strategy corresponding to the channel; Obtain satellite attribute information corresponding to the target satellite, environmental attribute information corresponding to the target device, channel attribute information corresponding to the channel, historical satellite coordinates of the target satellite at the historical time, historical device coordinates of the target device at the historical time, and historical access networks; Based on the channel fading value calculation strategy, the satellite attribute information, the environmental attribute information, the channel attribute information, and the current distance value, the channel fading value corresponding to the channel is determined; The historical connection information is determined based on the channel fading value, the historical satellite coordinates, the historical device coordinates, and the historical access network.

[0009] In one exemplary embodiment, determining the historical connection information based on the channel fading value, the historical satellite coordinate value, the historical device coordinate value, and the historical access network includes: Based on the current dynamic map, networks that are connected to the target device are selected from the at least two satellites to obtain the current access network; The current device coordinates of the target device, the channel fading value, and the current access network are used as the first state information. The historical satellite coordinates and channel fading values ​​of each satellite are used as the second state information. The historical device coordinates of the target device, the channel fading value, and the historical access network are used as third state information. The historical connection information is determined based on the first status information, the second status information, and the third status information.

[0010] In one exemplary embodiment, determining the historical connection information based on the first state information, the second state information, and the third state information includes: The first state information and the second state information are input into the spatial attention weight calculation network to perform spatial attention weight calculation processing to obtain the spatial attention weight; The second state information and the third state information are input into the time attention weight calculation network for time attention weight calculation to obtain the time attention weight. The target weight is obtained by calculating the product of the spatial attention weight and the temporal attention weight. The historical connection information is determined based on the target weight, the second state information, and the third state information.

[0011] In one exemplary embodiment, the step of predicting the connection status between the device and each satellite at a target time based on the first operating parameter value, the second operating parameter value, and the historical connection information of each satellite, to obtain a connection status prediction result, includes: The target time is encoded to obtain a target time vector; The first operating parameter value, the second operating parameter value, the historical connection information, and the target time vector of each satellite are concatenated to obtain the information to be predicted. The information to be predicted is input into the connection state prediction network for connection state prediction processing to obtain the connection state prediction result; the connection state prediction result represents the connection probability between the device and each satellite at the target time.

[0012] In one exemplary embodiment, the step of filtering the at least two satellites based on the connectivity state prediction results to obtain candidate satellites, and then using the candidate satellites and the ground base station as a candidate network set, includes: Based on the connection status prediction results, networks with a connection probability greater than a preset connection probability threshold among the at least two satellites are selected to obtain the filtered satellites; If the currently accessed network is not among the filtered satellites, the currently accessed network and the filtered satellites are used as candidate satellites. The candidate satellites and the ground base stations are used as the candidate network set.

[0013] In one exemplary embodiment, determining the connection matching degree between the device and each candidate network based on the network attribute information of each candidate network in the candidate network set includes: The network attribute information of each candidate network in the candidate network set is input into the connection matching degree calculation network to perform connection matching degree calculation processing, so as to obtain the connection matching degree between the device and each candidate network. After filtering the candidate network set according to the connection matching degree between the device and each candidate network to obtain the target network, the method further includes: Obtain the actual connection matching degree after the device connects to the target network; Based on the actual connection matching degree, train the connection matching degree calculation network to obtain the connection matching degree calculation network after training. The connection matching degree calculation network is updated to the connection matching degree calculation network after training is completed.

[0014] On the other hand, an access device for a device in a network system is provided, the network system including a ground base station, at least two satellites, and devices, the device comprising: The acquisition module is used to acquire the first operating parameter value of each satellite at the current time, the second operating parameter value of the device, and historical connection information; the historical connection information represents the network accessed by the device at a historical time and the historical network attribute information. The connection status prediction module is used to predict the connection status between the device and each satellite at a target time based on the first operating parameter value, the second operating parameter value and the historical connection information of each satellite, and to obtain the connection status prediction result. The candidate network set determination module is used to filter the at least two satellites based on the connection state prediction results to obtain candidate satellites, and to use the candidate satellites and the ground base station as a candidate network set; The connection matching degree calculation module is used to determine the connection matching degree between the device and each candidate network based on the network attribute information of each candidate network in the candidate network set. The target network determination module is used to filter the candidate network set according to the connection matching degree between the device and each candidate network to obtain the target network, so that the device can access the target network at the target time.

[0015] On the other hand, this application provides a method for accessing devices in a network system, the network system including a ground base station, at least two satellites, and at least two devices, the method comprising: Obtain the current satellite coordinates in the geocentric coordinate system, the preset coverage parameter values, and the current device coordinates of the target device in the geocentric coordinate system at the current moment; the target device is any one of the at least two devices. Based on the current device coordinates and the current satellite coordinates of the target satellite, determine the current distance between the target device and the target satellite; the target satellite is any one of the at least two satellites. Based on the rotation and translation parameters required for the target satellite to transform from the geocentric coordinate system to the target satellite coordinate system corresponding to the target satellite, coordinate transformation parameters are constructed. Based on the coordinate transformation parameters and the current device coordinates, determine the actual coordinates of the target device in the target satellite coordinate system; Calculate the angle between the actual coordinate value and the preset reference plane to obtain the current elevation angle; The actual coordinate values ​​are projected onto the preset reference plane to obtain reference coordinate values, and the angle between the reference coordinate values ​​and the preset reference axis is calculated to obtain the current azimuth angle. Use the current distance value, the current elevation angle, and the current azimuth angle as the current coverage parameter values; The current coverage parameter value and the preset coverage parameter value are compared to obtain the coverage parameter comparison result between the target device and each satellite; Based on the coverage parameter comparison results, a current dynamic graph is constructed; the current dynamic graph includes at least two satellite nodes and at least two edges corresponding to each satellite node; each satellite node includes the current satellite coordinate value and preset coverage parameter value of each satellite; each edge represents the connection status between each satellite and the device; Based on the current dynamic map, the current satellite coordinates and preset coverage parameter values ​​of each satellite are used as the first operating parameter values; Based on the channel between the target device and the target satellite, construct a channel fading value calculation strategy corresponding to the channel; Obtain satellite attribute information corresponding to the target satellite, environmental attribute information corresponding to the target device, channel attribute information corresponding to the channel, historical satellite coordinates of the target satellite at the historical time, historical device coordinates of the target device at the historical time, and historical access networks; Based on the channel fading value calculation strategy, the satellite attribute information, the environmental attribute information, the channel attribute information, and the current distance value, the channel fading value corresponding to the channel is determined; Based on the current dynamic map, networks that are connected to the target device are selected from the at least two satellites to obtain the current access network; The current device coordinates of the target device, the channel fading value, and the current access network are used as the first state information. The historical satellite coordinates and channel fading values ​​of each satellite are used as the second state information. The historical device coordinates of the target device, the channel fading value, and the historical access network are used as third state information. The first state information and the second state information are input into the spatial attention weight calculation network to perform spatial attention weight calculation processing to obtain the spatial attention weight; The second state information and the third state information are input into the time attention weight calculation network for time attention weight calculation to obtain the time attention weight. The target weight is obtained by calculating the product of the spatial attention weight and the temporal attention weight. Based on the target weight, the second state information, and the third state information, historical connection information is determined; the historical connection information represents the network accessed by the device at a historical time and the historical network attribute information. Obtain the second operating parameter value of the target device; The target time is encoded to obtain the target time vector; The first operating parameter value, the second operating parameter value, the historical connection information, and the target time vector of each satellite are concatenated to obtain the information to be predicted. The information to be predicted is input into the connection state prediction network for connection state prediction processing to obtain the connection state prediction result; the connection state prediction result represents the connection probability between the device and each satellite at the target time. Based on the connection status prediction results, networks with a connection probability greater than a preset connection probability threshold among the at least two satellites are selected to obtain the filtered satellites; If the currently accessed network is not among the filtered satellites, the currently accessed network and the filtered satellites are used as candidate satellites. The candidate satellites and the ground base stations are used as a candidate network set; The network attribute information of each candidate network in the candidate network set is input into the connection matching degree calculation network to perform connection matching degree calculation processing, so as to obtain the connection matching degree between the device and each candidate network. Based on the connection matching degree between the device and each candidate network, the candidate network set is filtered to obtain the target network, so that the device can access the target network at the target time. Obtain the actual connection matching degree after the device connects to the target network; Based on the actual connection matching degree, train the connection matching degree calculation network to obtain the connection matching degree calculation network after training. The connection matching degree calculation network is updated to the connection matching degree calculation network after training is completed.

[0016] On the other hand, an electronic device is provided, the device including a processor and a memory, the memory storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded by the processor and executed as described above regarding the device access method in a network system.

[0017] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction or at least one program is stored therein, the at least one instruction or at least one program being loaded and executed by a processor to implement the access method of a device in a network system as described above.

[0018] On the other hand, a computer program product or computer program is provided, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, executes the computer instructions, and causes the computer device to perform the access method for a device in a network system as described above.

[0019] This application provides a method and apparatus for device access in a network system, which has the following technical effects: This application obtains a first operating parameter value for each satellite at the current time, a second operating parameter value for the device, and historical connection information; the historical connection information characterizes the network accessed by the device at a historical time and historical network attribute information; based on the first operating parameter value, the second operating parameter value, and the historical connection information of each satellite, a connection state prediction process is performed on the connection state between the device and each satellite at a target time to obtain a connection state prediction result; based on the connection state prediction result, at least two satellites are screened to obtain candidate satellites, and the candidate satellites and the ground base station are used as a candidate network set; based on the network attribute information of each candidate network in the candidate network set, the connection matching degree between the device and each candidate network is determined; based on the connection matching degree between the device and each candidate network, the candidate network set is screened to obtain a target network; so that the device can access the target network at the target time. In a multi-user competitive network system, the connection status between the device and each satellite at the target time is predicted by combining the first operating parameter value of each satellite at the current time, the second operating parameter value of the device at the current time, and historical connection information. Furthermore, the connection matching degree between the device and each candidate network is calculated by combining the network attribute information of each candidate network. By combining the connection probability and the network's own attribute information, the target network is selected, which improves the reliability of the ground equipment's network access selection and avoids multi-user competition. Attached Figure Description

[0020] To more clearly illustrate the technical solutions and advantages in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating a device access method in a network system provided in the embodiments of this specification; Figure 2 This is a flowchart illustrating a smart access method for user equipment in a network system, as provided in the embodiments of this specification. Figure 3 This is a flowchart illustrating a method for dynamic modeling of spatiotemporal graphs in a network system, as provided in the embodiments of this specification. Figure 4 This is a flowchart illustrating an intelligent decision-making method in a network system provided in the embodiments of this specification; Figure 5This is a flowchart illustrating a federated learning optimization method in a network system provided in the embodiments of this specification; Figure 6 This is a schematic diagram of the structure of an access device for a device in a network system provided in the embodiments of this specification; Figure 7 This is a schematic diagram of the server structure for a device access method in a network system provided in the embodiments of this specification. Detailed Implementation

[0022] The technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0024] The following describes a device access method in a network system according to this application. Figure 1 This is a flowchart illustrating a device access method in a network system according to an embodiment of this specification. This specification provides the operational steps described in the embodiments or flowchart, but based on conventional or non-inventive methods, more or fewer operational steps may be included. The order of steps listed in the embodiments is merely one possible execution order among many and does not represent the only possible execution order. In actual system or server product execution, the methods shown in the embodiments or drawings can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment). Specifically, as shown in the embodiments or drawings... Figure 1 As shown, the network system includes a ground base station, at least two satellites, and equipment, and the method includes: S101: Obtain the first operating parameter value of each satellite at the current time, the second operating parameter value of the device, and historical connection information; the historical connection information represents the access network and historical network attribute information of the device at a historical time.

[0025] In the embodiments of this specification, the network system can be a space-ground integrated network, including terrestrial base stations, at least two low-Earth orbit satellites, and equipment. The terrestrial base station is a fixed radio station on the Earth's surface, serving as a core component of the cellular network and providing wireless network access to devices within a limited geographical area. The devices include users' mobile phones, tablets, etc., and in practice, multiple user devices exist. In actual use, assuming the user is in a busy area, they can preferentially connect to the terrestrial base station because the signal is strongest, the network speed is fastest, and the latency is lowest, providing a far better experience than connecting to a distant satellite. If the user is in an area with poor signal and no terrestrial base station signal, they will connect to the low-Earth orbit satellite. If the user is on a high-speed train passing through remote mountainous areas, the terrestrial base station coverage may be unstable and intermittent. When the terrestrial signal weakens, the system can automatically and seamlessly switch to the satellite network to ensure uninterrupted video conferencing or online gaming. Once in an area with good urban base station coverage, the system can switch back to the more economical and efficient terrestrial base station network. Therefore, in a network system composed of rapidly moving low-Earth orbit satellites and ground base stations, to enable multiple devices to intelligently and automatically select the optimal network connection point (a specific satellite or ground base station) and maintain high-quality network service, for each device, the following steps are taken: First operating parameter values ​​for each satellite at the current time t, second operating parameter values ​​for the device, and historical connection information. The first operating parameter values ​​include the satellite's three-dimensional position coordinates and coverage parameters at the current time. The second operating parameter values ​​include the device's three-dimensional coordinates, speed and direction of movement (which can be used to predict which satellite's coverage area it will enter in the future), service demand characteristics (current application type (e.g., web browsing, high-definition video streaming; different services have different bandwidth and latency requirements; historical traffic patterns or quality of service requirements), throughput requirements, etc.). Historical connection information represents the network accessed by the device at a historical time and its historical network attributes (e.g., signal strength, location, load, etc.). By acquiring these three pieces of information, the probability of the device's connection status with each satellite at a target time can be predicted to determine the satellites the device might connect to at that target time.

[0026] S103: Based on the first operating parameter value, the second operating parameter value, and the historical connection information of each satellite, perform prediction processing on the connection status between the device and each satellite at the target time to obtain the connection status prediction result.

[0027] In the embodiments of this specification, a high-dimensional input vector is obtained based on the first operating parameter value, the second operating parameter value, and historical connection information of each satellite. The connection state prediction network is then used to predict the connection state between the device and each satellite at the target time to obtain the connection state prediction result.

[0028] S105: Based on the connection prediction results, the at least two satellites are screened to obtain candidate satellites, and the candidate satellites and the ground base station are used as a candidate network set.

[0029] In the embodiments of this specification, a preset connection probability threshold is set in advance. At least two satellites are screened based on the connection prediction results to obtain candidate satellites, thereby filtering out satellites with unstable connections and low connection probabilities, narrowing the candidate list from all satellites to high-probability satellites, and avoiding wasting computing resources on satellites with low connection probabilities. After the candidate satellites are selected, ground base stations are always an important alternative, especially in areas with good coverage such as urban areas. Therefore, the candidate satellites and ground base stations are used as a candidate network set for joint evaluation.

[0030] S107: Determine the connection matching degree between the device and each candidate network based on the network attribute information of each candidate network in the candidate network set.

[0031] In the embodiments of this specification, network attribute information of each candidate network in the candidate network set is obtained, including Received Signal Strength (RSSI), Available Bit Rate (ABR), Delay, Satellite Load (Load_j), and Relative Speed ​​(Δv_rel). With the goal of maximizing the reward (maximizing connection quality, minimizing handover penalty, and maximizing load balancing), the reward value between the device and each candidate network, i.e., the connection matching degree, is calculated.

[0032] S109: Based on the connection matching degree between the device and each candidate network, the candidate network set is filtered to obtain the target network; so that the device can access the target network at the target time.

[0033] In the embodiments of this specification, the candidate network set is filtered according to the connection matching degree between the device and each candidate network to obtain the target network, so that the device can access the target network at the target time, avoiding multiple users competing for access. This provides the user device with a high-speed and stable connection in the network system, avoids unnecessary frequent switching, and prevents brief service interruptions during switching. In addition, users can connect to a network with a lighter load, avoiding network congestion.

[0034] In this embodiment of the specification, the network system includes at least two devices; before acquiring the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and historical connection information, the method further includes: Obtain the current satellite coordinates in the geocentric coordinate system, the preset coverage parameter value, and the current device coordinates of the target device in the geocentric coordinate system at the current time; the target device is any one of the at least two devices. In this embodiment of the specification, the network system includes at least two devices; the geocentric coordinate system has the Earth's center of mass as its origin, the Z-axis pointing to the North Pole, the X-axis pointing to the intersection of the Prime Meridian and the Equator, and the Y-axis forming a right-handed coordinate system; the system acquires the current satellite coordinates, preset coverage parameter values, and the current device coordinates of the target device in the geocentric coordinate system at the current moment; the target device is any one of the at least two devices; wherein, the current satellite coordinates are three-dimensional position coordinates. j is the satellite's number, and t represents the current time. Let x be the x-axis coordinate of satellite j. Let j be the y-axis coordinate of satellite j. Here is the z-axis coordinate of satellite j; preset coverage parameter values ​​include... , Let be the minimum radius of the coverage area of ​​satellite j on the Earth's surface. Let $j$ be the maximum radius of the coverage area of ​​satellite $j$ on the Earth's surface. Let be the angle between the projection of the beam center axis of satellite j onto the ground and the direction of true north, i.e., the azimuth angle. The elevation angle is the angle between the line of sight from the ground user equipment to satellite j and the local horizontal plane; the current equipment coordinates are also three-dimensional position coordinates.

[0035] Based on the current device coordinates and the current satellite coordinates of each satellite, determine the current coverage parameter values ​​between the target device and each satellite; In the embodiments of this specification, the current distance, elevation angle and azimuth angle between the target device and each satellite are calculated based on the current device coordinates and the current satellite coordinates of each satellite, thereby determining the current coverage parameter values ​​between the target device and each satellite.

[0036] The current coverage parameter value and the preset coverage parameter value are compared to obtain the coverage parameter comparison result between the target device and each satellite; In the embodiments of this specification, the current coverage parameter value and the preset coverage parameter value are compared and processed to obtain the coverage parameter comparison result between the target device and each satellite, and to determine whether the target device is within the coverage area. This realizes the conversion of the physical relationship between the device and each satellite into a binary relationship (1 indicates that the user device is within the satellite coverage area, and 0 indicates that the user device is outside the satellite coverage area).

[0037] Based on the coverage parameter comparison results, a current dynamic graph is constructed; the current dynamic graph includes at least two satellite nodes and at least two edges corresponding to each satellite node; each satellite node includes the current satellite coordinate value and preset coverage parameter value of each satellite; each edge represents the connection status between each satellite and the device; In the embodiments of this specification, spatiotemporal information is integrated based on the coverage parameter comparison results. That is, the current satellite coordinates and preset coverage parameter values ​​of each satellite are used to construct satellite nodes, and the connection status between each satellite and the device is established as edges to form a dynamic graph. The dynamic graph = (satellite node set, user connection set); the satellite node set is all satellite nodes, and the user connection set is the connection status between the satellite corresponding to each satellite node and all devices. In addition, each satellite node in the dynamic graph includes not only the coordinates and preset coverage parameters of each satellite at the current moment, but also the coordinates and coverage parameters of each satellite at historical moments. The user connection set includes not only the connection status between each satellite and each device at the current moment, but also the connection status between each satellite and each device at historical moments.

[0038] Secondly, since the dynamic graph includes the satellite's position, coverage parameters, and connection status between the satellite and the equipment at multiple points in time, a distance change threshold is preset. After obtaining the new position coordinates of the satellites at the next processing time, for each satellite, calculate its displacement distance since the last update:

[0039] in, This represents the displacement distance of satellite j since the last update; for example, it can be the Euclidean distance. For satellite j at time 3D coordinate vector; For satellite j at time 3D coordinate vector; Calculated Distance change threshold Comparison, assumption Less than or equal to This indicates that the position of satellite j has not changed significantly, its coverage area and its connection with the device have not changed, therefore the global update of satellite j and its related edges is skipped; assuming Greater than This indicates that satellite j has moved far enough and its coverage area may have changed significantly. At this point, a local update is triggered, the coverage parameter value of satellite j is calculated, and the connection status between satellite j and all devices is re-verified. Nodes and edges in the dynamic graph that are not related to satellite j remain unchanged. The satellite nodes and their related edges at this moment are updated to generate a new dynamic graph.

[0040] In this embodiment of the specification, obtaining the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and historical connection information includes: Based on the current dynamic map, the current satellite coordinates and preset coverage parameter values ​​of each satellite are used as the first operating parameter values; In the embodiments of this specification, based on the current dynamic graph, the current satellite coordinate value of each satellite corresponding to each satellite node in the current dynamic graph and the preset coverage parameter value are used as the first operating parameter value.

[0041] Obtain the second operating parameter value of the target device and the historical connection information.

[0042] In the embodiments described in this specification, the second operating parameter value and historical connection information of the target device are obtained to calculate the connection probability between the device and each satellite at the target time. Based on the satellite coordinate values, preset coverage parameter values, and the device coordinate values ​​of the target device, a current dynamic graph is constructed. This abstracts each satellite as a node containing its location and coverage area, and determines the connection state between the device and the satellite as edges based on geometric calculations. The physical relationship between the satellite and the device is digitally mapped, providing a temporal context for predicting future connection states for the spatiotemporal attention mechanism. In addition, the dynamic graph is updated incrementally, avoiding updating all satellites at every time step, saving the bandwidth required for signaling and synchronization, and providing a basis for the device's real-time intelligent switching decision.

[0043] In this embodiment of the specification, determining the current coverage parameter value between the target device and each satellite based on the current device coordinate value and the current satellite coordinate value of each satellite includes: Based on the current device coordinates and the current satellite coordinates of the target satellite, determine the current distance between the target device and the target satellite; the target satellite is any one of the at least two satellites. In the embodiments of this specification, the preset coverage parameter values ​​include the preset coverage radius, preset coverage azimuth, and preset coverage elevation angle of each satellite; the target satellite is any one of at least two satellites; based on the current device coordinates and the current satellite coordinates of the target satellite, the current distance between the target device and the target satellite is calculated, thereby determining whether the target device is within the effective communication range of the target satellite.

[0044] Based on the rotation and translation parameters required for the target satellite to transform from the geocentric coordinate system to the target satellite coordinate system corresponding to the target satellite, coordinate transformation parameters are constructed. In the embodiments of this specification, coordinate transformation parameters, namely the transformation matrix, are constructed based on the rotation and translation parameters required to transform the target position from the geocentric coordinate system to the target satellite coordinate system corresponding to the target satellite. The target satellite coordinate system takes the centroid of the target satellite as its origin, and the coordinate system is determined by the satellite attitude, reflecting the actual direction of the antenna beam.

[0045] Based on the coordinate transformation parameters and the current device coordinates, determine the actual coordinates of the target device in the target satellite coordinate system; In the embodiments described in this specification, the target device is transformed to the target satellite coordinate system according to the coordinate transformation parameters and the current device coordinate value to obtain the actual coordinate value.

[0046] Calculate the angle between the actual coordinate value and the preset reference plane to obtain the current elevation angle; In the embodiments of this specification, the preset reference plane is the XY plane in the target satellite coordinate system. The angle between the actual coordinate value and the preset reference plane is calculated to obtain the current elevation angle.

[0047] The actual coordinate values ​​are projected onto the preset reference plane to obtain reference coordinate values, and the angle between the reference coordinate values ​​and the preset reference axis is calculated to obtain the current azimuth angle. In the embodiments of this specification, a preset reference axis is the X-axis; the actual coordinate values ​​are projected onto the preset reference plane to obtain reference coordinate values, and the angle between the reference coordinate values ​​and the preset reference axis is calculated to obtain the current azimuth angle.

[0048] The current distance value, the current elevation angle, and the current azimuth angle are used as the current coverage parameter values.

[0049] In the embodiments of this specification, the current distance, current elevation angle, and current azimuth angle are used as the current coverage parameter values. In subsequent comparisons, the current distance value is compared with a preset coverage radius value, the current azimuth angle is compared with a preset coverage azimuth angle, and the current elevation angle is compared with a preset elevation angle to obtain the coverage parameter comparison results. By transforming the device's coordinates from the geocentric coordinate system to the target satellite coordinate system, the actual pointing of the antenna beam is accurately reflected based on the coordinate system transformation. The calculation of the distance, elevation angle, and azimuth angle between the device and the satellite improves the accuracy of the beam coverage judgment results.

[0050] In this embodiment of the specification, before obtaining the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and the historical connection information, the method further includes: Based on the channel between the target device and the target satellite, construct a channel fading value calculation strategy corresponding to the channel; In the embodiments described in this specification, a channel fading value calculation strategy is constructed based on the channel between the target device and the target satellite. Specifically, this strategy is a shadowed Ricean fading channel model. It describes the propagation process of the signal from satellite j to device i, and its calculation formula is:

[0051] in, This represents the combined channel gain from satellite j to device i at time t, i.e., the channel fading value, which includes the amplitude attenuation and phase change of the signal. This is a large-scale fading component that describes the slow decay of the signal on a macroscopic scale. It is mainly affected by distance and large obstacles. Since the distance changes continuously due to the rapid movement of the satellite, this component must be calculated. This is a small-scale fading component that describes the rapid random fluctuations of a signal at a microscopic scale. It is mainly caused by multipath effects. Since the instantaneous distance remains constant, the signal strength will also change drastically due to the interference of the reflection path, affecting the instantaneous connection quality. Therefore, this component must be calculated.

[0052] Obtain satellite attribute information corresponding to the target satellite, environmental attribute information corresponding to the target device, channel attribute information corresponding to the channel, historical satellite coordinates of the target satellite at the historical time, historical device coordinates of the target device at the historical time, and historical access networks; In the embodiments of this specification, satellite attribute information corresponding to the target satellite, environmental attribute information corresponding to the target device, channel attribute information corresponding to the channel, historical coordinate values ​​of the target satellite at historical times, historical device coordinate values ​​of the target device at historical times, and historical access networks are obtained. The satellite attribute information includes the satellite's position vector, transmitter parameters (transmit power, transmit antenna gain, antenna polarization, beamforming weight), and satellite system parameters (operating frequency, signal bandwidth); the environmental attribute information includes the device type (handheld terminal / vehicle terminal / fixed station), usage scenario, and movement speed; the channel attribute information includes propagation path parameters (transmission distance between satellite and device, elevation angle), large-scale fading parameters (shadow fading standard deviation, shadow fading correlation distance, path loss index), and small-scale fading parameters (multipath component number, delay spread, Doppler spread, Rice K-factor), etc.

[0053] Based on the channel fading value calculation strategy, the satellite attribute information, the environmental attribute information, the channel attribute information, and the current distance value, the channel fading value corresponding to the channel is determined; In the embodiments described in this specification, satellite attribute information, environmental attribute information, channel attribute information, and the current distance value are substituted into the channel fading value calculation strategy to calculate the large-scale fading component. and small-scale fading components Then calculate the product of the two to obtain the channel fading value. .

[0054] The historical connection information is determined based on the channel fading value, the historical satellite coordinates, the historical device coordinates, and the historical access network.

[0055] In the embodiments of this specification, by integrating channel fading values, historical satellite coordinates, historical device coordinates, and historical access networks, the historical state vectors of the satellites (second state information), the historical state vectors of the devices (third state information), and the current state vectors of the devices (first state information) are determined. Based on a spatiotemporal attention mechanism, joint spatiotemporal weights are calculated to further obtain historical connection information. The channel fading value is calculated, and by separating large-scale path loss from small-scale multipath fading, the spatiotemporal attenuation characteristics of the satellite signal are accurately quantified, providing a real physical basis for connection quality prediction. The channel fading value is incorporated as a key feature into the state vector, and then historical information is weighted and integrated through a spatiotemporal attention mechanism, enabling the system to accurately predict fading trends and improving the accuracy of connection state prediction.

[0056] In this embodiment of the specification, determining the historical connection information based on the channel fading value, the historical satellite coordinate value, the historical device coordinate value, and the historical access network includes: Based on the current dynamic map, networks that are connected to the target device are selected from the at least two satellites to obtain the current access network; In the embodiments of this specification, when the target device is currently connected to a satellite, the connection status between each satellite and each device can be determined based on the current dynamic graph and the user connection set in the dynamic graph. The network currently connected to the target device can be selected from at least two satellites, and the network can be represented as an identifier ID.

[0057] The current device coordinates of the target device, the channel fading value, and the current access network are used as the first state information. In the embodiments described in this specification, the current device coordinates of the target device, the channel fading value of the channel between the target device and the target satellite, and the current access network are used as the current state vector of the target device, i.e., the first state information. ; In addition, the first state information may also include the device's current direction of movement, the device's service requirements (service type, data mode, etc.), and the device's current network environment (such as a list of visible satellites).

[0058] The historical satellite coordinates and channel fading values ​​of each satellite are used as the second state information. In the embodiments described in this specification, for each satellite, the historical satellite coordinates and channel fading values ​​are used as the historical state vector of each satellite, i.e., the second state information. Specifically, the channel fading value here refers to the overall channel statistics of each satellite, rather than the channel response of a specific device. It includes the arithmetic mean of the signal-to-noise ratio, channel quality distribution, and load level of all devices served by the satellite at historical times. In addition, the second status information may also include the satellite's operating speed, coverage parameter values, and network load status (number of historical connected devices, resource utilization rate, etc.) at historical moments.

[0059] The historical device coordinates of the target device, the channel fading value, and the historical access network are used as third state information. In the embodiments of this specification, the historical device coordinates of the target device, the channel fading value of the channel between the target device and the satellite at a historical time, and the historical access network are used as the historical state vector of the target device at a historical time, i.e., the third state information. ; In addition, the third state information can also include the target device's movement direction and service requirements at historical moments.

[0060] The historical connection information is determined based on the first status information, the second status information, and the third status information.

[0061] In the embodiments of this specification, based on the first state information, the second state information, and the third state information, and combined with a spatiotemporal attention mechanism, spatial attention weights and temporal attention weights are calculated, and further combined to obtain the target weight, generating historical connection information. By generating historical connection information based on the historical state information of satellites (second state information), the current state information of the target device (first state information), and the historical state information of the target device (third state information), historical connection information is generated. This achieves dynamic evaluation of the historical correlation of different satellites through spatial attention, captures the user's own connection pattern changes through temporal attention, and merges the two into a joint weight, improving the accuracy of connection state prediction and contributing to stable, low-latency intelligent network switching.

[0062] In this embodiment of the specification, determining the historical connection information based on the first state information, the second state information, and the third state information includes: The first state information and the second state information are input into the spatial attention weight calculation network to perform spatial attention weight calculation processing to obtain the spatial attention weight; In the embodiments described in this specification, at a certain historical moment τ, the device may be covered by multiple satellites. A spatial attention mechanism is needed to dynamically assess the device's current needs and determine which satellite's state at a past moment is more valuable. For example, a satellite with a strong signal at that time and whose trajectory matches the device's future path would receive higher weight. Specifically, the first state information and the second state information are input into the spatial attention weight calculation network for spatial attention weight calculation, and the spatial attention weight is obtained. The calculation formula is as follows:

[0063] in, The spatial attention weights for satellite x; This is a linear transformation function used to transform the device's first state information. Converted into a query vector, representing the device's current connection requirements and network status; This is a linear transformation function used to transform the satellite's second state information. Converted into a key vector, it represents the supply capacity and service status of satellite x at historical time τ; To obtain the similarity score between the query vector and the key vector, the higher the similarity score, the better the state of satellite x at that moment matches the current needs of the device. The dimension of the key vector; Divide the similarity score Scaling is performed to stabilize the gradient, and the Softmax function is used for normalization to obtain the spatial attention weight of each satellite x with respect to device i at the current time t, which represents the importance distribution of different satellite historical states from the current decision perspective.

[0064] The second state information and the third state information are input into the time attention weight calculation network for time attention weight calculation to obtain the time attention weight. In the embodiments described in this specification, since the importance of historical information does not decrease linearly over time, the time attention mechanism can adaptively learn to determine how long ago the user's state was, for the current time t. It is the most valuable reference.

[0065] Specifically, the second and third state information are input into the temporal attention weight calculation network for temporal attention weight calculation, and the calculation formula is as follows:

[0066] in, Time-based attention weights; is a trainable weight matrix; For activation functions; The device's first status information and the device's third status information By concatenating the features along the feature dimension, a longer joint feature vector is obtained. Using trainable weight matrices Perform a linear transformation on this joint vector and input the result of the linear transformation into the activation function. In (such as Sigmoid), the output is a value between 0 and 1, which is the time attention weight.

[0067] The target weight is obtained by calculating the product of the spatial attention weight and the temporal attention weight. In the embodiments of this specification, the spatial attention weight and the temporal attention weight are multiplied element by element to obtain the joint spatiotemporal weight for each satellite x at historical time τ, i.e., the target weight. This determines the overall importance of the state information of satellite x at historical time τ for device i. The target weight considers whether satellite x is important at historical time from a spatial dimension and whether historical time is important to the present from a temporal dimension. The formula for calculating the target weight is as follows:

[0068] in, The target weight; Spatial attention weights; For time-based attention weights.

[0069] The historical connection information is determined based on the target weight, the second state information, and the third state information.

[0070] In the embodiments described in this specification, the second state information and the third state information are weighted using target weights, and the weighted information is summed to obtain historical connection information. By dynamically quantifying the information value of different satellites at different historical moments through spatiotemporal attention weights, massive time-series data is refined into a concise context vector. The spatiotemporal patterns most relevant to the current decision are extracted from massive historical data, accurately predicting channel changes and load trends, thereby improving the connection stability of the device and the network resource utilization rate.

[0071] In this embodiment of the specification, the step of predicting the connection status between the device and each satellite at a target time based on the first operating parameter value, the second operating parameter value, and the historical connection information of each satellite, to obtain a connection status prediction result, includes: The target time is encoded to obtain a target time vector; In the embodiments described in this specification, the satellite network exhibits strong periodicity, and the service mode of the equipment also exhibits periodicity. Since traditional timestamps are linearly increasing, it is necessary to encode the target time to obtain a target time vector, thereby transforming linearly passing time into periodic location encoding. The target time vector can be represented as... ,in, To represent a future moment starting from the current moment; T is usually set as the orbital period of a low-Earth orbit satellite; when When the value approaches an integer multiple of the orbital period T, the encoded value cyclically recurs, indicating that the satellite will return to a similar position, thus accurately predicting the periodic recurrence of coverage relationships.

[0072] The first operating parameter value, the second operating parameter value, the historical connection information, and the target time vector of each satellite are concatenated to obtain the information to be predicted. ; In the embodiments of this specification, the first operating parameter value, the second operating parameter value, historical connection information and the target time vector of each satellite are spliced ​​together to construct a comprehensive feature vector and obtain the information to be predicted.

[0073] The information to be predicted is input into the connection state prediction network for connection state prediction processing to obtain the connection state prediction result; the connection state prediction result characterizes the connection probability between the device and each satellite at the target time.

[0074] In the embodiments described in this specification, a pre-set trained neural network weight matrix W is provided, the dimension of which is (output dimension, total input feature dimension). A linear transformation is performed on the information to be predicted using the neural network weight matrix to obtain a linear output. The algorithm learns the complex nonlinear contribution weights of different features to the connection probability. Then, the linear output is input into a connection state prediction network for connection state prediction processing. This network can be a sigmoid function, mapping the linear output to the (0,1) interval, representing the connection probability value between the device and each satellite at the target time, i.e., the connection state prediction result. By concatenating the first operating parameter value of each satellite, user characteristics (second operating parameter value), historical connection information, and the target time vector, it integrates channel attenuation trends, device mobility modes, changes in service demands, and orbital periodicity. Based on the information to be predicted, connection state prediction processing is performed, achieving intelligent evaluation that integrates historical patterns and future predictions, thus improving the connection stability and resource utilization of the network system.

[0075] In this embodiment of the specification, the step of filtering the at least two satellites based on the connection status prediction results to obtain candidate satellites, and using the candidate satellites and the ground base station as a candidate network set, includes: Based on the connection status prediction results, networks with a connection probability greater than a preset connection probability threshold among the at least two satellites are selected to obtain the filtered satellites; In the embodiments of this specification, a preset connection probability threshold is set, which can be 0.7 for example. Based on the connection status prediction results, at least two satellites are selected to form networks with a connection probability greater than the preset connection probability threshold. Satellites with too low a connection probability are filtered out, and high-probability satellites are retained to form the selected satellites, thereby reducing the complexity of subsequent decision-making.

[0076] If the currently accessed network is not among the filtered satellites, the currently accessed network and the filtered satellites are used as candidate satellites. In the embodiments of this specification, in order to avoid unnecessary switching, if the currently accessed network is not found among the satellites after screening, the currently accessed network should also be included in the options as a candidate for the device to maintain operation, thus obtaining candidate satellites.

[0077] The candidate satellites and the ground base stations are used as the candidate network set.

[0078] In the embodiments described in this specification, terrestrial base stations are always considered important alternatives, especially in areas with good coverage such as urban areas. Therefore, terrestrial base stations are also considered, and candidate satellites and terrestrial base stations are included in the candidate network set. Pre-screening based on connection status prediction results yields filtered satellites, enabling the rapid elimination of satellites with poor connection stability. Including the current access network and terrestrial base stations in the candidate network set ensures the completeness of subsequent connection matching degree calculations, avoiding unnecessary handover oscillations while retaining the important option of switching to a terrestrial base station network, thus contributing to achieving globally optimal decision-making within the network system.

[0079] In this embodiment of the specification, determining the connection matching degree between the device and each candidate network based on the network attribute information of each candidate network in the candidate network set includes: The network attribute information of each candidate network in the candidate network set is input into the connection matching degree calculation network to perform connection matching degree calculation processing, so as to obtain the connection matching degree between the device and each candidate network. In the embodiments of this specification, each device in the network system corresponds to an intelligent agent, which has a defined action space, including three action options (stay / switch to a satellite / switch to a ground base station) to enable it to automatically make the optimal access choice based on the complex network environment; Specifically, the network attribute information of each candidate network in the candidate network set is a 5-dimensional state vector, which can be represented as follows: ,in, The Received Signal Strength Indicator (RSSI) measures the strength of the current connection signal; a higher value is better. Available bit rate represents the theoretically achievable data rate under current channel conditions; End-to-end latency includes propagation latency and processing latency; The payload of satellite j reflects the resource scarcity of satellite j; This represents the relative speed between device i and satellite j, which affects channel stability and connection hold-up time. As can be seen, this 5-dimensional vector quantifies connection quality, network status, and connection sustainability.

[0080] The network system pre-designs a connection matching degree calculation network, which is a policy network trained based on a multi-objective reward function. Specifically, during the training phase, the initial policy network interacts extensively with the network system in a simulated environment, tries different actions based on the current state, and then evaluates the merits of the actions based on the comprehensive reward value calculated by the multi-objective reward function based on the environmental feedback. Using the evaluation results, the network parameters of the initial policy network are adjusted so that the adjusted policy network increases the probability of actions that can obtain high comprehensive rewards and reduces the probability of actions that lead to low rewards or penalties. Through continuous iterative training, the policy network is obtained. Specifically, this multi-objective reward function guides the agent to learn and optimize multiple objectives, including three objectives: maximizing throughput, minimizing handover penalty, and maximizing load balancing. Maximizing throughput encourages the agent to provide high-speed connections to ensure basic performance. Minimizing handover penalty suppresses unnecessary frequent handovers, as each handover is accompanied by signaling overhead and brief interruptions, thus improving stability. Maximizing load balancing encourages the agent to select networks with lighter loads, avoiding overloading some satellites while leaving others idle, thereby improving the overall capacity and efficiency of the entire network system. Specifically, the multi-objective reward function can be:

[0081] in, The immediate reward obtained by the agent after performing an action at time t; This is a weighting factor used to adjust the importance of throughput in the total reward; A measure of the throughput or rate of the current connection; This is a weighting coefficient used to adjust the importance of the switching penalty in the total reward; This is a switching indicator function. If the action is a switch, the value is 1; otherwise, it is 0. It is used to apply a switching penalty. This is a weighting factor used to adjust the importance of load balancing in the total reward; The standard deviation of the load for all satellites is given; the smaller the standard deviation, the more balanced the load. In actual use, the network attribute information (5-dimensional state vector) of each candidate network is concatenated to obtain a high-dimensional global state vector, which is then input into the policy network to obtain the probability distribution of each action output by the policy network, i.e. the connection matching degree between the device and each candidate network, while keeping the sum of the connection matching degrees of the current access network, switching to any candidate network, and switching to the ground base station at 1.

[0082] In this embodiment of the specification, after filtering the candidate network set according to the connection matching degree between the device and each candidate network to obtain the target network, the method further includes: Obtain the actual connection matching degree after the device connects to the target network; In the embodiments of this specification, after the device accesses the target network, it will calculate and return an immediate reward based on the multi-target reward function, namely the actual connection matching degree; for example, +5 indicates that the handover is successful and of good quality, and -3 indicates that an unnecessary handover is triggered.

[0083] Based on the actual connection matching degree, train the connection matching degree calculation network to obtain the connection matching degree calculation network after training. In the embodiments described in this specification, the device stores the state before the switch (network condition), the action performed (switching to the target network), the actual connection matching degree obtained after performing the action, and the new state after the switch as an empirical data in a buffer. The device periodically extracts a batch of data from the buffer, calculates the advantage function, and for example, analyzes whether the long-term benefit of performing the action in the state before the switch is higher or lower than the average benefit at that time. It also calculates the gradient, which indicates in which direction the parameters of the policy network should be fine-tuned to make the probability of the optimal action appearing in the future higher and the probability of the undesirable action appearing lower. Then, using the calculated gradient, the network parameters of the policy network are updated to obtain the connection matching degree calculation network after training.

[0084] The connection matching degree calculation network is updated to the connection matching degree calculation network after training is completed.

[0085] In the embodiments of this specification, the connection matching degree calculation network is updated to the connection matching degree calculation network after training. Additionally, each device's corresponding agent can update its own policy network locally and then upload its network parameters to the central server. The central server aggregates all network parameters, for example, by averaging them to obtain updated network parameters, and then distributes these updated network parameters to each agent, updating their network parameters accordingly, thus obtaining the updated connection matching degree calculation network. Distributed real-time intelligent decision-making is achieved through MA-PPO, ensuring that devices select the optimal access point in the network system. Combined with incremental federated learning, while protecting user data privacy, global experience is aggregated to continuously optimize the policy model, achieving highly stable, low-latency seamless satellite-to-ground roaming and improving overall network resource utilization and long-term performance.

[0086] In one exemplary implementation, such as Figure 2 As shown, Figure 2A flowchart illustrating a smart access method for user equipment in a network system, provided in an embodiment of this specification, includes: S201: Satellite communication network environment.

[0087] In the embodiments described in this specification, the satellite communication environment includes low-Earth orbit satellites, ground base stations, and user terminals.

[0088] S202: Satellite Ephemeris / Location User Measurement Report.

[0089] In the embodiments described in this specification, accurate satellite ephemeris data is obtained from the satellite operator, and the real-time three-dimensional position of each satellite at any given time is calculated using an orbital mechanics model. In addition, the user equipment locates itself using GPS / BeiDou and obtains a channel measurement report by measuring downlink reference signals, thereby obtaining the status data of the satellites and the user equipment.

[0090] S203: Spatiotemporal graph dynamic modeling and incremental update mechanism.

[0091] In the embodiments described in this specification, each satellite is modeled as a graph node, whose attributes include the satellite's location, coverage area, etc. Based on the geometric calculation between the location of the user equipment and the location of the satellite, it is determined whether the user equipment and the satellite are connected, thereby establishing the edge between the user equipment and the satellite to represent the connection relationship between the satellite and the user equipment; Meanwhile, in order to reduce computational overhead, a displacement threshold can be set, and the position, coverage parameters, and connection relationship with user equipment associated with the satellite will only be recalculated when the satellite position changes beyond the displacement threshold.

[0092] S204: Spatiotemporal attention encoder.

[0093] In the embodiments of this specification, based on the satellite positions and coverage areas in the dynamic graph and the user characteristics of the user equipment, and combined with the spatial attention mechanism, it is analyzed which satellite states are more important to the current decision at a certain historical moment; combined with the temporal attention mechanism, it is analyzed which historical states are most valuable for predicting the future; finally, a weighted feature vector is output, which integrates the most critical spatiotemporal patterns extracted from history, and is used to predict the future connection probability.

[0094] S205: MA-PPO agent.

[0095] In the embodiments of this specification, candidate satellites are screened based on the predicted connection probability. Combined with ground base stations, the network attribute information of the candidate satellites and ground base stations is used as the input of the policy network (MA-PPO agent). The output is a probability distribution of all possible actions (maintain the current network / switch to a certain satellite / switch to a ground base station). The action with the highest probability is selected as the final decision instruction. This realizes the transformation of complex, multi-objective network optimization problems into real-time decision-making, and obtains the comprehensive optimal access selection under the current situation.

[0096] S206: Execute the switching instruction.

[0097] In the embodiments described in this specification, a handover command is generated based on the MA-PPO decision; the user equipment executes the link handover process according to the handover command, releases the connection with the currently connected network, and establishes a new connection with the target node (satellite or ground base station), thereby realizing the transformation of intelligent decision into actual action and changing the network connection status.

[0098] S207: Performance data feedback.

[0099] In the embodiments of this specification, after the handover is performed, the performance indicators of the new link are continuously monitored, including RSSP, throughput, handover interruption time, load changes of the new service node, etc., and these raw indicators are input into the multi-objective reward function defined in a priority manner to calculate the instant reward value, thereby quantifying the overall effect of this decision.

[0100] S208: Federated Learning Optimizer.

[0101] In the embodiments of this specification, each user equipment uses its own generated experience data, including multiple sets of data, each set of data including the current access network, handover action, post-handover reward value, and post-handover network, to locally calculate the update of policy network parameters. The user equipment only uploads the encrypted model parameter update to the central server. The server securely aggregates the parameter updates from a large number of devices, generates an updated global model, and distributes the updated global model to all user equipment to update their ontology policies.

[0102] In this embodiment, raw data such as satellite ephemeris, user location, and channel measurement reports are continuously collected. A spatiotemporal graph is dynamically modeled, with high-speed moving satellites designated as nodes and user connections as edges, constructing the spatiotemporal graph. An incremental update mechanism ensures efficient synchronization between the model and satellite movement. Based on a spatiotemporal attention encoder, the system dynamically learns the importance weights of different historical moments and different satellite nodes for the current decision, accurately predicting future connection quality. A multi-agent near-end policy optimization (MA-PPO) decision network outputs the probability distribution of each agent's action (hold / switch to a satellite / switch to a ground base station) and selects the optimal command for execution. After the command is executed, the system calculates the reward value based on a multi-objective reward function (comprehensively considering signal quality, handover cost, and load balancing) as a quantitative evaluation of the decision's effectiveness. This empirical data is optimized through incremental federated learning, automatically balancing competitive indicators such as signal strength, handover overhead, and network load, avoiding localized congestion caused by a single objective. The federated learning framework enables the system to continuously improve network adaptability by utilizing global data while protecting user privacy.

[0103] In one exemplary implementation, such as Figure 3 As shown, Figure 3 A flowchart illustrating a spatiotemporal graph dynamic modeling method in a network system, provided in an embodiment of this specification, includes: S301: Graph node.

[0104] In the embodiments of this specification, the position and coverage parameters of each satellite (such as satellite 1 and satellite 2) are calculated, and each satellite becomes a graph node, whose attribute vector includes the three-dimensional coordinates and coverage area of ​​each satellite.

[0105] S302: Connection status.

[0106] In the embodiments described in this specification, the user equipment calculates and judges the distance, elevation angle and azimuth angle between itself and the satellite based on its own location information, and determines the connection status between the user equipment and each satellite.

[0107] S303: Spatiotemporal Evolution Diagram.

[0108] In the embodiments of this specification, each satellite and the connection status between each satellite and the device are combined to form a spatiotemporal evolution diagram.

[0109] S304: Future Connectivity Forecast.

[0110] In the embodiments of this specification, satellite characteristics are obtained based on the spatiotemporal evolution map, and the connection probability between the user equipment and each satellite at a future time is predicted by combining the required prediction time, user characteristic information, and historical connection information.

[0111] In this embodiment, satellites and user equipment serve as dual data sources. Satellites provide real-time position and coverage parameters to establish graph nodes, while user equipment calculates its connection status with each satellite using distance, elevation, and azimuth angles, establishing graph edges. By combining nodes and edges, a spatiotemporal evolution graph is constructed to capture the patterns of satellite motion and connection changes. This allows the system to predict connection trends based on historical patterns (rather than just instantaneous states), providing crucial information for switching decisions. Based on the spatiotemporal evolution graph, predictions of future connections are output. Furthermore, an incremental update mechanism is used for the spatiotemporal evolution graph, reducing computational overhead.

[0112] In one exemplary implementation, such as Figure 4 As shown, Figure 4 A flowchart illustrating an intelligent decision-making method in a network system provided in this specification includes: S401: Input: Environment state s_t.

[0113] In the embodiments described in this specification, the environmental state s_t of all candidate satellites and ground base stations will be included in the input policy network.

[0114] S402: MA-PPO policy network.

[0115] In the embodiments described in this specification, the MA-PPO policy network processes the environmental state and outputs the probability distribution of all possible actions (maintaining the currently connected network, switching to a satellite, or switching to a ground base station).

[0116] S403: Generate switching action a_t.

[0117] In the embodiments described in this specification, the switching action a_t is generated based on the probability distribution, selecting the one with the highest probability.

[0118] S404: Perform the action.

[0119] In the embodiments described in this specification, action a_t is executed to change the access network of the user equipment.

[0120] S405: Observe new states and rewards.

[0121] In the embodiments described in this specification, the user equipment accesses a new network and calculates a reward value based on the new network and a predefined multi-objective reward function.

[0122] S406: Model Update and Optimization.

[0123] In the embodiments of this specification, the collected empirical data (s_t, a_t, r_t, s_{t+1}) are used, where s_t is the network accessed by the user equipment before the handover, a_t is the handover action, r_t is the reward value calculated by the multi-objective reward function after the handover access network, and s_{t+1} is the network accessed by the user equipment after the handover action, to update and optimize the model, use the PPO algorithm to calculate the gradient, and adjust the policy network parameters so that it is more inclined to choose actions that can obtain high long-term rewards in the future.

[0124] In this embodiment, the environmental state vector encapsulates all network information perceived by the user device. Combined with the MA-PPO policy network, a switching action is generated, and the optimal choice is output in probabilistic form. The system then executes the action, changing the network connection. After the network switch, the environment provides feedback observation of the new state and reward, and the collected empirical data is used to update and optimize the model. The PPO algorithm adjusts the policy network parameters, making it more inclined to choose high-reward actions in similar states in the future, achieving multi-objective optimization. The updated model is then immediately used in the next round of decision-making, repeating the process to improve network performance and resource utilization efficiency.

[0125] In one exemplary implementation, such as Figure 5 As shown, Figure 5 A flowchart illustrating a federated learning optimization method in a network system, provided as an embodiment of this specification, includes: S501: Terminal node.

[0126] In the embodiments described in this specification, each terminal node (user device, such as mobile phone, vehicle terminal, IoT device) has its own local dataset and a copy of the global model, so that the data is always kept on the user side, ensuring the distribution of data and the privacy of users.

[0127] S502: Local data training.

[0128] In the embodiments described in this specification, the user equipment uses its own local, private historical data (e.g., the historical signal quality and handover records of connections with different satellites recorded by the device) to train and optimize the global model downloaded from the server, generating an updated local model. This ensures that the original sensitive data (such as precise location and usage habits) does not leave the user equipment. At the same time, local training allows the model to adapt to the network environment and behavioral patterns of each user equipment.

[0129] S503: Upload model parameters.

[0130] In the embodiments described in this specification, the user device does not upload the original data. Instead, it encrypts the updated values ​​(or gradients) of the trained model parameters and then uploads them to the central server, further strengthening privacy and security. The server can only see how the parameters are adjusted, but not the original data, reducing the risk of data leakage.

[0131] S504: Global Aggregation Server.

[0132] In the embodiments described in this specification, encrypted model parameter updates are received from a large number of terminal devices. The server provides a platform for parameter aggregation and model synchronization, ensuring that distributed learning can be uniformly integrated.

[0133] S505: Aggregate to generate a global model.

[0134] In the embodiments of this specification, since the data and field of view of a single device are limited, by aggregating the local experience of thousands of devices, the new model can learn more general and comprehensive network rules, thereby improving the robustness of the model. Specifically, the server uses a federated averaging algorithm to perform a weighted average of all received local model parameter updates, merging them into a new and improved global model.

[0135] S506: Distribute the global model.

[0136] In the embodiments described in this specification, the server distributes the aggregated and upgraded global model to all terminal nodes to ensure that the decision-making intelligence of the entire network is improved synchronously, thus avoiding differences in model performance between devices.

[0137] In addition, the terminal node uses the newly distributed global model as a starting point to continue combining the newly generated local data and begin the next round of local training.

[0138] In this embodiment, each terminal node holds local private data. Each terminal uses its local data to train the latest downloaded global model locally, generating a local model update that is more suitable for its own environment. The encrypted model parameters are then uploaded to the global aggregation server, avoiding the risk of privacy leakage. The server securely aggregates the parameters uploaded by a massive number of terminals, generates an updated global model, and distributes it to all terminal nodes, replacing the old model. The terminals then retrain based on the new model and fresh data, starting the next round of optimization, which improves the resource utilization of the network system and the connection stability of users.

[0139] This manual also provides access devices for network systems, such as... Figure 6 As shown, the network system includes a ground base station, at least two satellites, and equipment, the device including: The acquisition module 601 is used to acquire the first operating parameter value of each satellite at the current time, the second operating parameter value of the device, and historical connection information; the historical connection information represents the access network and historical network attribute information of the device at a historical time. The connection status prediction module 602 is used to predict the connection status between the device and each satellite at a target time based on the first operating parameter value, the second operating parameter value and the historical connection information of each satellite, and to obtain the connection status prediction result. The candidate network set determination module 603 is used to filter the at least two satellites according to the connection state prediction results to obtain candidate satellites, and to use the candidate satellites and the ground base station as a candidate network set; The connection matching degree calculation module 604 is used to determine the connection matching degree between the device and each candidate network based on the network attribute information of each candidate network in the candidate network set; The target network determination module 605 is used to filter the candidate network set according to the connection matching degree between the device and each candidate network to obtain the target network, so that the device can access the target network at the target time.

[0140] In some embodiments, the apparatus further includes: The current time acquisition module is used to acquire the current satellite coordinates, preset coverage parameter values, and current device coordinates of the target device in the geocentric coordinate system for each satellite at the current time; the target device is any one of the at least two devices. The current coverage parameter value determination module is used to determine the current coverage parameter value between the target device and each satellite based on the current device coordinate value and the current satellite coordinate value of each satellite; The coverage parameter comparison result determination module is used to compare the current coverage parameter value and the preset coverage parameter value to obtain the coverage parameter comparison result between the target device and each satellite. The current dynamic graph construction module is used to construct a current dynamic graph based on the coverage parameter comparison results; the current dynamic graph includes at least two satellite nodes and at least two edges corresponding to each satellite node; each satellite node includes the current satellite coordinate value and preset coverage parameter value of each satellite; each edge represents the connection status between each satellite and the device; In some embodiments, the acquisition module further includes: The first determining submodule is used to take the current satellite coordinates of each satellite and the preset coverage parameter value as the first operating parameter value according to the current dynamic map. The second determining submodule is used to obtain the second operating parameter value of the target device and the historical connection information.

[0141] In some embodiments, the current coverage parameter value determination module further includes: The current distance value determination submodule is used to determine the current distance value between the target device and the target satellite based on the current device coordinates and the current satellite coordinates of the target satellite; the target satellite is any one of the at least two satellites; The coordinate transformation parameter construction submodule is used to construct coordinate transformation parameters based on the rotation and translation parameters required for the target satellite to transform from the geocentric coordinate system to the target satellite coordinate system corresponding to the target satellite. The actual coordinate value determination submodule is used to determine the actual coordinate value of the target device in the target satellite coordinate system based on the coordinate transformation parameters and the current device coordinate value; The current elevation angle determination submodule is used to calculate the angle between the actual coordinate value and the preset reference plane to obtain the current elevation angle; When the forward azimuth angle determination submodule is used, it is used to project the actual coordinate value onto the preset reference plane to obtain the reference coordinate value, and calculate the angle between the reference coordinate value and the preset reference axis to obtain the current azimuth angle; The current coverage parameter value determination submodule is used to use the current distance value, the current elevation angle, and the current azimuth angle as the current coverage parameter value.

[0142] In some embodiments, the apparatus further includes: The channel fading value calculation strategy construction module is used to construct the channel fading value calculation strategy corresponding to the channel based on the channel between the target device and the target satellite. The information acquisition module is used to acquire satellite attribute information corresponding to the target satellite, environmental attribute information corresponding to the target device, channel attribute information corresponding to the channel, historical satellite coordinates of the target satellite at the historical time, historical device coordinates of the target device at the historical time, and historical access networks. The channel fading value determination module is used to determine the channel fading value corresponding to the channel based on the channel fading value calculation strategy, the satellite attribute information, the environmental attribute information, the channel attribute information, and the current distance value. The historical connection information determination module is used to determine the historical connection information based on the channel fading value, the historical satellite coordinate value, the historical device coordinate value, and the historical access network.

[0143] In some embodiments, the historical connection information determination module further includes: The current access network determination submodule is used to filter the networks that are connected to the target device from the at least two satellites according to the current dynamic map, so as to obtain the current access network; The first state information determination submodule is used to use the current device coordinates of the target device, the channel fading value, and the current access network as the first state information. The second state information determination submodule is used to use the historical satellite coordinates of each satellite and the channel fading value as the second state information. The third state information determination submodule is used to use the historical device coordinates of the target device, the channel fading value, and the historical access network as the third state information. The historical connection information determination submodule is used to determine the historical connection information based on the first status information, the second status information, and the third status information.

[0144] In some embodiments, the historical connection information determination submodule further includes: The spatial attention weight determination unit is used to input the first state information and the second state information into the spatial attention weight calculation network to perform spatial attention weight calculation processing to obtain spatial attention weights. The time attention weight determination unit is used to input the second state information and the third state information into the time attention weight calculation network to perform time attention weight calculation processing to obtain the time attention weight; The target weight determination unit is used to calculate the product between the spatial attention weight and the temporal attention weight to obtain the target weight; The historical connection information determination unit is used to determine the historical connection information based on the target weight, the second state information, and the third state information.

[0145] In some embodiments, the connection state prediction module further includes: The target time vector determination submodule is used to encode the target time to obtain the target time vector. The submodule for determining the information to be predicted is used to concatenate the first operating parameter value, the second operating parameter value, the historical connection information, and the target time vector of each satellite to obtain the information to be predicted. The connection state prediction result determination submodule is used to input the information to be predicted into the connection state prediction network for connection state prediction processing to obtain the connection state prediction result; the connection state prediction result characterizes the connection probability between the device and each satellite at the target time.

[0146] In some embodiments, the candidate network set determination module further includes: The filtered satellite determination submodule is used to filter the networks with a connection probability greater than a preset connection probability threshold among the at least two satellites based on the connection status prediction results, and obtain the filtered satellites. The candidate satellite determination submodule is used to select the current access network and the selected satellites as candidate satellites if the current access network is not among the selected satellites. The candidate network set determination submodule is used to select the candidate satellites and the ground base stations as the candidate network set.

[0147] In some embodiments, the connection matching degree calculation module further includes: The connection matching degree calculation submodule is used to input the network attribute information of each candidate network in the candidate network set into the connection matching degree calculation network for connection matching degree calculation processing, so as to obtain the connection matching degree between the device and each candidate network. In some embodiments, the apparatus further includes: The actual connection matching degree acquisition module is used to acquire the actual connection matching degree of the device after it accesses the target network; The training module is used to train the connection matching degree calculation network based on the actual connection matching degree, and obtain the connection matching degree calculation network after training. An update module is used to update the connection matching degree calculation network to the connection matching degree calculation network after training.

[0148] The apparatus and method embodiments described herein are based on the same inventive concept.

[0149] This specification provides an electronic device including a processor and a memory. The memory stores at least one instruction or at least one program, which is loaded and executed by the processor to implement the device access method in a network system as provided in the above method embodiments.

[0150] Embodiments of this application also provide a computer storage medium, which can be disposed in a terminal to store at least one instruction or at least one program related to implementing a device access method in a network system according to the method embodiments. The at least one instruction or at least one program is loaded and executed by the processor to implement the device access method in the network system provided by the above method embodiments.

[0151] Embodiments of this application also provide a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the device access method in the network system provided in the above-described method embodiments.

[0152] The memory described in the embodiments of this specification can be used to store software programs and modules. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for the functions, etc.; the data storage area may store data created according to the use of the device, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory may also include a memory controller to provide the processor with access to the memory.

[0153] The device access methods in the network system provided in this specification can be executed on mobile terminals, computer terminals, servers, or similar computing devices. Taking running on a server as an example... Figure 7 This is a hardware structure block diagram of a server for a device access method in a network system provided in the embodiments of this specification. For example... Figure 7As shown, the server 700 can vary significantly due to different configurations or performance. It may include one or more Central Processing Units (CPUs) 710 (CPUs 710 may include, but are not limited to, microprocessors (MCUs) or programmable logic devices (FPGAs), a memory 730 for storing data, and one or more storage media 720 (e.g., one or more mass storage devices) for storing application programs 723 or data 722. The memory 730 and storage media 720 may be temporary or persistent storage. The program stored in the storage media 720 may include one or more modules, each module may include a series of instruction operations on the server. Furthermore, the CPU 710 may be configured to communicate with the storage media 720 and execute the series of instruction operations stored in the storage media 720 on the server 700. Server 700 may also include one or more power supplies 760, one or more wired or wireless network interfaces 750, one or more input / output interfaces 740, and / or one or more operating systems 721, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0154] The input / output interface 740 can be used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of server 700. In one example, the input / output interface 740 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the input / output interface 740 may be a radio frequency (RF) module used for wireless communication with the Internet.

[0155] Those skilled in the art will understand that Figure 7 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, server 700 may also include... Figure 7 The more or fewer components shown, or having the same Figure 7 The different configurations shown.

[0156] As can be seen from the embodiments of the device access method and apparatus in the network system provided in this application, this application obtains the first operating parameter value of each satellite at the current time, the second operating parameter value of the device, and historical connection information; the historical connection information characterizes the access network and historical network attribute information of the device at a historical time; based on the first operating parameter value, the second operating parameter value, and the historical connection information of each satellite, the connection status between the device and each satellite at the target time is predicted to obtain a connection status prediction result; based on the connection status prediction result, at least two satellites are filtered to obtain candidate satellites, and the candidate satellites and the ground base station are used as a candidate network set; based on the network attribute information of each candidate network in the candidate network set, the connection matching degree between the device and each candidate network is determined; based on the connection matching degree between the device and each candidate network, the candidate network set is filtered to obtain a target network; so that the device can access the target network at the target time. In a multi-user competitive network system, a dynamic graph is constructed to provide a temporal context for predicting future connection states for the spatiotemporal attention mechanism. Incremental updates to the dynamic graph avoid updating all satellites at every time step, saving bandwidth required for signaling and synchronization, and providing a foundation for real-time intelligent handover decisions. By dynamically quantifying the information value of different satellites at different historical moments through spatiotemporal attention weights, the spatiotemporal patterns most relevant to the current decision are extracted from massive historical data to generate historical connection information, accurately predicting channel changes and load trends, and improving the connection stability and network resource utilization of the device. A multi-agent near-end policy optimization algorithm is adopted to generate access / handover decisions, optimizing a multi-objective reward function that includes throughput, handover frequency, and load balancing, ensuring that the device selects the optimal access point in the network system. Combined with incremental federated learning, while protecting user data privacy, global experience is aggregated to continuously optimize the policy model, achieving highly stable, low-latency seamless satellite-to-ground roaming and improving handover success rate and system throughput.

[0157] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0158] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0159] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer storage medium, such as a read-only memory, a disk, or an optical disk.

[0160] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for accessing devices in a network system, characterized in that, The network system includes a ground base station, at least two satellites, and equipment; the method includes: The system acquires the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and historical connection information; the historical connection information represents the network accessed by the device at a historical moment and its historical network attribute information. Based on the first operating parameter value, the second operating parameter value, and the historical connection information of each satellite, the connection status between the device and each satellite at the target time is predicted to obtain the connection status prediction result. Based on the connection status prediction results, the at least two satellites are screened to obtain candidate satellites, and the candidate satellites and the ground base station are used as a candidate network set. Based on the network attribute information of each candidate network in the candidate network set, the connection matching degree between the device and each candidate network is determined; Based on the connection matching degree between the device and each candidate network, the candidate network set is filtered to obtain the target network; so that the device can access the target network at the target time.

2. The method according to claim 1, characterized in that, The network system includes at least two devices; before acquiring the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and historical connection information, the method further includes: Obtain the current satellite coordinates in the geocentric coordinate system, the preset coverage parameter value, and the current device coordinates of the target device in the geocentric coordinate system at the current time; the target device is any one of the at least two devices. Based on the current device coordinates and the current satellite coordinates of each satellite, determine the current coverage parameter values ​​between the target device and each satellite; The current coverage parameter value and the preset coverage parameter value are compared to obtain the coverage parameter comparison result between the target device and each satellite; Based on the coverage parameter comparison results, a current dynamic graph is constructed; the current dynamic graph includes at least two satellite nodes and at least two edges corresponding to each satellite node; each satellite node includes the current satellite coordinate value and preset coverage parameter value of each satellite; each edge represents the connection status between each satellite and the device; The acquisition of the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and historical connection information includes: Based on the current dynamic map, the current satellite coordinates and preset coverage parameter values ​​of each satellite are used as the first operating parameter values; Obtain the second operating parameter value of the target device and the historical connection information.

3. The method according to claim 2, characterized in that, Determining the current coverage parameter value between the target device and each satellite based on the current device coordinate value and the current satellite coordinate value of each satellite includes: Based on the current device coordinates and the current satellite coordinates of the target satellite, determine the current distance between the target device and the target satellite; the target satellite is any one of the at least two satellites. Based on the rotation and translation parameters required for the target satellite to transform from the geocentric coordinate system to the target satellite coordinate system corresponding to the target satellite, coordinate transformation parameters are constructed. Based on the coordinate transformation parameters and the current device coordinates, determine the actual coordinates of the target device in the target satellite coordinate system; Calculate the angle between the actual coordinate value and the preset reference plane to obtain the current elevation angle; The actual coordinate values ​​are projected onto the preset reference plane to obtain reference coordinate values, and the angle between the reference coordinate values ​​and the preset reference axis is calculated to obtain the current azimuth angle. The current distance value, the current elevation angle, and the current azimuth angle are used as the current coverage parameter values.

4. The method according to claim 3, characterized in that, Before obtaining the first operating parameter value of each satellite at the current moment, the second operating parameter value of the device, and the historical connection information, the method further includes: Based on the channel between the target device and the target satellite, construct a channel fading value calculation strategy corresponding to the channel; Obtain satellite attribute information corresponding to the target satellite, environmental attribute information corresponding to the target device, channel attribute information corresponding to the channel, historical satellite coordinates of the target satellite at the historical time, historical device coordinates of the target device at the historical time, and historical access networks; Based on the channel fading value calculation strategy, the satellite attribute information, the environmental attribute information, the channel attribute information, and the current distance value, the channel fading value corresponding to the channel is determined; The historical connection information is determined based on the channel fading value, the historical satellite coordinates, the historical device coordinates, and the historical access network.

5. The method according to claim 4, characterized in that, The step of determining the historical connection information based on the channel fading value, the historical satellite coordinate value, the historical device coordinate value, and the historical access network includes: Based on the current dynamic map, networks that are connected to the target device are selected from the at least two satellites to obtain the current access network; The current device coordinates of the target device, the channel fading value, and the current access network are used as the first state information. The historical satellite coordinates and channel fading values ​​of each satellite are used as the second state information. The historical device coordinates of the target device, the channel fading value, and the historical access network are used as third state information. The historical connection information is determined based on the first status information, the second status information, and the third status information.

6. The method according to claim 5, characterized in that, Determining the historical connection information based on the first state information, the second state information, and the third state information includes: The first state information and the second state information are input into the spatial attention weight calculation network to perform spatial attention weight calculation processing to obtain the spatial attention weight; The second state information and the third state information are input into the time attention weight calculation network for time attention weight calculation to obtain the time attention weight. The target weight is obtained by calculating the product of the spatial attention weight and the temporal attention weight. The historical connection information is determined based on the target weight, the second state information, and the third state information.

7. The method according to claim 6, characterized in that, The step of predicting the connection status between the device and each satellite at a target time based on the first operating parameter value, the second operating parameter value, and the historical connection information of each satellite, to obtain a connection status prediction result, includes: The target time is encoded to obtain a target time vector; The first operating parameter value, the second operating parameter value, the historical connection information, and the target time vector of each satellite are concatenated to obtain the information to be predicted. The information to be predicted is input into the connection state prediction network for connection state prediction processing to obtain the connection state prediction result; the connection state prediction result characterizes the connection probability between the device and each satellite at the target time.

8. The method according to claim 7, characterized in that, The step of filtering the at least two satellites based on the connection state prediction results to obtain candidate satellites, and using the candidate satellites and the ground base station as a candidate network set, includes: Based on the connection status prediction results, networks with a connection probability greater than a preset connection probability threshold among the at least two satellites are selected to obtain the filtered satellites; If the currently accessed network is not among the filtered satellites, the currently accessed network and the filtered satellites are used as candidate satellites. The candidate satellites and the ground base stations are used as the candidate network set.

9. The method according to claim 1, characterized in that, The step of determining the connection matching degree between the device and each candidate network based on the network attribute information of each candidate network in the candidate network set includes: The network attribute information of each candidate network in the candidate network set is input into the connection matching degree calculation network to perform connection matching degree calculation processing, so as to obtain the connection matching degree between the device and each candidate network. After filtering the candidate network set according to the connection matching degree between the device and each candidate network to obtain the target network, the method further includes: Obtain the actual connection matching degree after the device connects to the target network; Based on the actual connection matching degree, train the connection matching degree calculation network to obtain the connection matching degree calculation network after training. The connection matching degree calculation network is updated to the connection matching degree calculation network after training is completed.

10. An access device for a device in a network system, characterized in that, The network system includes a ground base station, at least two satellites, and equipment, the device including: The acquisition module is used to acquire the first operating parameter value of each satellite at the current time, the second operating parameter value of the device, and historical connection information; the historical connection information represents the network accessed by the device at a historical time and the historical network attribute information. The connection status prediction module is used to predict the connection status between the device and each satellite at a target time based on the first operating parameter value, the second operating parameter value and the historical connection information of each satellite, and to obtain the connection status prediction result. The candidate network set determination module is used to filter the at least two satellites based on the connection state prediction results to obtain candidate satellites, and to use the candidate satellites and the ground base station as a candidate network set; The connection matching degree calculation module is used to determine the connection matching degree between the device and each candidate network based on the network attribute information of each candidate network in the candidate network set. The target network determination module is used to filter the candidate network set according to the connection matching degree between the device and each candidate network to obtain the target network, so that the device can access the target network at the target time.

Images