Intelligent mobility management and service continuity guarantee method for ntn

Abstract

The application provides an NTN-oriented intelligent mobility management and service continuity guarantee method, which is executed by an intelligent mobility management engine on the network side, and comprises the following steps: based on real-time position information of a user equipment and a network node, a prediction algorithm is used to predict the position in a future set time, and a dynamic network relationship graph is constructed based on the prediction result; a graph neural network is used to analyze the dynamic network relationship graph, so that a network reconfiguration decision is obtained; based on the network reconfiguration decision, a network device is indirectly guided to trigger a handover process conforming to the 3GPP standard, so as to execute handover and core network element migration. By constructing the dynamic network relationship graph, the network reconfiguration decision is obtained, the decision guides the standard 3GPP process to trigger and execute in the direction of global optimization through the way of preconfiguring network resources and optimizing strategies. Therefore, the purposes of seamless handover, reduced signaling overhead, globalized decision and significantly enhanced service continuity are achieved.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, and in particular relates to a method for intelligent mobility management and service continuity assurance for NTN. Background Technology

[0002] With the development of 6G communication technology, building wide-area coverage non-terrestrial networks using non-terrestrial platforms such as low-Earth orbit satellites has become an important direction for 5G-A and 6G. However, the high-speed movement of low-Earth orbit satellites relative to the ground causes their coverage area (beam) to move rapidly on the ground, resulting in continuous changes in the wireless link between ground user equipment (UE) and the satellite.

[0003] Existing technologies primarily employ mobility management schemes similar to those used in terrestrial networks. This involves the UE measuring the signal quality of the serving satellite and neighboring satellites, and triggering a handover process when specific handover conditions are met. This approach has significant drawbacks:

[0004] 1) Response delay: Due to the large propagation delay of the satellite-to-ground link, the entire process from measurement and reporting to network decision-making takes a long time, which can easily lead to handover failure or service interruption in high-speed satellite motion scenarios.

[0005] 2) High signaling overhead: Frequent measurement reports and handover executions generate a large amount of signaling on the satellite link, wasting valuable air interface resources.

[0006] 3) Ping-pong effect: At the edge of multiple satellite coverage, due to rapid signal fluctuations, the UE is prone to frequent and unnecessary switching between multiple satellites, which increases signaling overhead and reduces user experience.

[0007] 4) Lack of global optimization: Decisions based on local instantaneous measurements cannot predict future changes in network state, making it difficult to make long-term optimal switching decisions and easily leading to the "ping-pong effect".

[0008] 5) Lack of core network mobility management: When satellites not only provide access functions but also carry core network user plane functions, existing technologies lack an effective management mechanism for the migration of UPF service nodes as the satellite moves, making it difficult to guarantee end-to-end service continuity. Summary of the Invention

[0009] To address the aforementioned problems in the existing technology, this invention provides a method for intelligent mobility management and service continuity assurance for NTN.

[0010] This disclosure provides a method for intelligent mobility management and service continuity assurance for NTN, executed by the network-side intelligent mobility management engine, and the method includes:

[0011] Based on the real-time location information of user equipment and network nodes, a prediction algorithm is used to predict the location within a set time in the future, and a dynamic network relationship graph is constructed based on the prediction results.

[0012] The dynamic network relationship graph is analyzed using a graph neural network to obtain network reconfiguration decisions;

[0013] Based on the network reconfiguration decision, network devices are indirectly guided to trigger a handover process compliant with 3GPP standards to perform handover and core network element migration.

[0014] Optionally, the dynamic network relationship graph includes:

[0015] The network node set V, the set of connections between nodes E, and the set of weights of the connecting edges W, where the weights in the weight set W are multi-dimensional weighting factors that include transmission delay and node load.

[0016] Optionally, the prediction algorithm used to predict the location within a set time period in the future is used to predict and track the location trajectory of network participants, including user equipment, satellite base stations, satellite core networks, satellite routing, DN networks, ground stations, and ground core networks. The prediction algorithm includes the Kalman filter algorithm.

[0017] The network node set V of the dynamic network relationship graph includes user equipment, serving and neighboring satellite base stations, access and mobility management functions, session management functions, user plane functions, and data network nodes; the connection relationship set E between nodes represents the logical or physical connection between nodes; the weight set W includes at least the transmission delay between nodes and the node load, and the weight set W also includes one or more of the following: link distance, link bandwidth, and historical stability.

[0018] Optionally, the network reconfiguration decision includes at least a predictive optimal access link sequence and / or a globally optimized core network element selection strategy.

[0019] Optionally, the predictive optimal access link sequence includes:

[0020] By using graph neural networks to learn the dynamic laws of network topology, we can predict the sequence of access links that maximizes the multi-objective optimization function within a future time window.

[0021] The multi-objective optimization function is used to jointly maximize link continuity and service quality, and minimize switching costs.

[0022] Optionally, the globally optimized core network element selection strategy includes:

[0023] The user-facing feature selection strategy is based on a comprehensive scoring function to select the optimal user-facing feature. The comprehensive scoring function is as follows:

[0024] Score(UPF) = α*(1-Norm(Delay)) + β*(1-Norm(Load)) + γ*Norm(Stability) + δ*(1-Norm(DN Path Cost)); where α is the delay factor weight, β is the load factor weight, γ is the stability factor weight, δ is the data network path factor weight, the Norm function maps each original value to the interval [0,1], and Score(UPF) is the comprehensive scoring function.

[0025] Optionally, the globally optimized core network element selection strategy includes a control plane anchoring strategy, which includes:

[0026] Prioritize anchoring to access and mobility management functions that can simultaneously serve the current source base station and the future target base station sequence within the predicted time window, in order to trigger the Xn handover process;

[0027] If no access and mobility management function triggers the Xn handover process, the N2 handover process is triggered in advance, and the access and mobility management function that can provide the user equipment with the most persistent service in the subsequent time is selected as the target.

[0028] Optionally, the step of indirectly guiding network devices to trigger a 3GPP-compliant handover process based on the network reconfiguration decision includes:

[0029] By pre-configuring network resources and / or optimizing radio resource management policies, network devices can be indirectly guided to trigger a handover process compliant with 3GPP standards.

[0030] Optionally, the indirect guidance includes:

[0031] The network-side intelligent mobility management engine sends the decision results to the source base station and / or the core network control plane;

[0032] According to the decision, the source base station and / or core network control plane modify the neighbor cell measurement configuration of the user equipment through radio resource control signaling, so as to prompt the user equipment to report the measurement report that meets the handover conditions in advance.

[0033] Upon receiving the measurement report, the source base station and / or core network control plane triggers and executes the corresponding handover procedure according to the 3GPP standard.

[0034] Optionally, the switching process includes at least one of the following scenarios:

[0035] Scenario A: For Xn handover involving only a change in the base station, the source base station directly interacts with the target base station through the Xn interface after receiving the measurement report;

[0036] Scenario B: N2 handover of base station and user plane functions is triggered by the source base station, completed through the cooperation of the source access and mobility management function and the target access and mobility management function, and the reselection and migration of user plane functions is controlled by the session management function.

[0037] Scenario C: Inter-AMF N2 handover where both the base station and access and mobility management functions change is triggered by the source base station and completed through the N14 interface in cooperation between the source access and mobility management functions and the target access and mobility management functions.

[0038] The intelligent mobility management and service continuity assurance method for NTN provided by this invention obtains network reconfiguration decisions by constructing a dynamic network relationship graph. This decision guides standard 3GPP processes towards global optimization through pre-configured network resources and optimization strategies, thereby achieving seamless handover, reduced signaling overhead, global decision-making, and significantly enhanced service continuity. Attached Figure Description

[0039] The above and other objects, features and advantages of this disclosure will become more apparent from the accompanying drawings, in which like reference numerals generally denote like parts.

[0040] Figure 1 A schematic block diagram of an NTN system consisting of a low Earth orbit satellite constellation provided in an embodiment of this disclosure. Detailed Implementation

[0041] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0042] It should be understood that the following specific examples illustrate the implementation of this disclosure, and those skilled in the art can easily understand other advantages and effects of this disclosure from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. This disclosure can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this disclosure. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0043] It should be noted that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this disclosure, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice method. Furthermore, this device and / or practice method can be implemented using other structures and / or functionalities besides one or more of the aspects set forth herein.

[0044] It should also be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this disclosure. The illustrations only show the components related to this disclosure and are not drawn according to the number, shape and size of the components in actual implementation. In actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0045] Furthermore, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the described aspects can be practiced without these specific details.

[0046] Xn refers to the interface that connects two 5G base stations;

[0047] N2 refers to the interface between the 5G radio access network and the core network control plane;

[0048] Inter-AMF N2 refers to a handover process in which the serving base station and core network control plane elements of the user equipment change, and this change process is completed through N2 interface signaling.

[0049] This embodiment is used for methods and systems for mobility management, core network element selection, and service continuity assurance in non-terrestrial networks (NTNs), such as spaceborne core networks. It solves the problem that existing passive mobility management solutions cannot meet the stringent service continuity requirements of NTNs in highly dynamic environments.

[0050] Based on the above background, the technical problem to be solved in this embodiment is: how to overcome the passivity, lag and limitations of existing NTN mobility management schemes.

[0051] Specifically, this embodiment aims to solve the following problems:

[0052] (1) How to proactively predict handover events to overcome the lag problem caused by large time delays between satellite and ground;

[0053] (2) How to reduce unnecessary measurement reports and handover signaling, and reduce resource consumption during transition;

[0054] (3) How to avoid switching between ping-pong games and improve user experience;

[0055] (4) How to optimize the selection of core network elements (especially UPF) globally and ensure end-to-end service continuity in scenarios where both satellite and onboard network elements are in motion.

[0056] This embodiment discloses a method for intelligent mobility management and service continuity assurance for NTN, which is executed by the intelligent mobility management engine on the network side. The method includes:

[0057] Based on the real-time location information of user equipment and network nodes, a prediction algorithm is used to predict the location within a set time in the future, and a dynamic network relationship graph is constructed based on the prediction results.

[0058] The dynamic network relationship graph is analyzed using a graph neural network to obtain network reconfiguration decisions;

[0059] Based on the network reconfiguration decision, network devices are indirectly guided to trigger a handover process compliant with 3GPP standards to perform handover and core network element migration.

[0060] In a specific scenario, an AI-based NTN (Network Transaction Management) intelligent mobility management and business continuity assurance method is presented. By introducing Kalman filtering (location prediction and tracking) and graph neural networks (GNNs) (optimal link prediction), a predictive and global integrated intelligent management system is constructed. The method is triggered by an intelligent model management unit and mainly includes the following steps:

[0061] Step 1: Dynamic Position Tracking and Relationship Graph Construction

[0062] (1) The Kalman filter algorithm is used to predict the trajectory of the real-time GPS location information of the UE and satellites (including UE, onboard gNB, onboard UPF, onboard AMF / SMF and other nodes) and maintain their location tracking table, as well as smoothing processing; the smoothing processing includes filtering out the stars outside the circle range with the UE coordinates as the center and a set radius to form the current service area, such as using Euclidean clustering, radius search and other algorithms.

[0063] By predicting end-to-end locations, an optimal serving satellite link is selected: the gNB with the longest coverage, the AMF with the longest access, and the UPF with the best latency.

[0064] (2) Network relationship graph construction: Based on the aforementioned prediction results, and with the current UE location as the center, a dynamic network relationship graph G(V, E, W) covering its service area is constructed.

[0065] Node (V): Includes all relevant network nodes such as UE, serving and neighboring satellites (as gNB), onboard AMF / SMF, onboard and terrestrial UPF, onboard and terrestrial data networks (DN);

[0066] Edge (E): Represents the potential or actual connection between nodes (such as N2, N3, N9, N6 interfaces, which can be represented as inter-satellite link distance).

[0067] Edge weight (W): is a multi-dimensional weighting factor that includes at least inter-node transmission delay and node load, and can be extended to link distance, link bandwidth, historical stability, etc.

[0068] By transforming highly dynamic physical networks into graph data structures that can be processed and analyzed by computers, a global network view containing spatiotemporal predictive information is provided for subsequent intelligent decision-making.

[0069] Step 2: Intelligent Decision Making and Strategy Formulation Based on GNN:

[0070] The graph neural network (GNN) is used to analyze and learn the dynamic network relationship graph described above, and outputs a forward-looking network reconfiguration decision. Two functions are provided to work together to complete the decision-making process.

[0071] Function 1: Predictive Link Planning and Definition of "Optimal" Strategy:

[0072] GNN learns the dynamic patterns of network topology by aggregating the features of nodes in the graph, and predicts the "optimal access link sequence" within a future time window, such as the next 30 seconds, which can be flexibly adjusted according to the actual situation.

[0073] The "optimal" strategy: The loss function of GNN is designed as a multi-objective optimization function, which aims to jointly maximize the following metrics:

[0074] Link persistence: Select the satellite gNB with the longest possible connection time within the future time window;

[0075] Service Quality: Prioritize gNBs with high signal quality (e.g., RSRP / RSRQ) and the longest predicted service distance, and low load.

[0076] Switching cost: Penalize unnecessary switching and encourage maintaining the current connection;

[0077] Integration with 3GPP procedures: Based on the optimal link sequence predicted in this way, the network side can initiate a handover request to the target gNB in ​​advance before the actual handover occurs, and establish forward transmission tunnels in advance, such as Xn-U tunnels for intra-AMF handover, or N2 / Uu related resources for inter-AMF handover, thereby absorbing the resource establishment delay during the handover process in advance and achieving "seamless" handover.

[0078] Function 2: Globally optimized core network element selection strategy:

[0079] UPF Intelligent Migration Strategy: GNN jointly optimizes the selection of the UPF and the access satellite. Its decision is based on a comprehensive scoring function:

[0080] Score(UPF) = α*(1-Norm(delay)) + β*(1-Norm(load)) + γ*Norm(stability) + δ*(1-Norm(DN path cost));

[0081] Before actual calculation, all parameters must be normalized to ensure consistent dimensions and fair weighting. The Norm() function maps each original value to the interval [0, 1].

[0082] Latency: is the transmission delay from UPF to the target gNB and DN;

[0083] Load: This refers to the CPU / memory / bandwidth utilization of the UPF.

[0084] Path stability: This is based on the predicted duration for which the UPF will serve future UEs. The UPF with the highest overall score is selected.

[0085] α, β, and γ are weighting coefficients, which quantify the relative importance of different optimization objectives in the final decision.

[0086] Specifically:

[0087] α (Alpha): Delay factor weight. It weighs the importance of transmission delay from the UPF to the target gNB and DN.

[0088] β (Beta): Load factor weight. It weighs the importance of the UPF's own (CPU / memory) load.

[0089] γ (Gamma): Stability factor weight. It weighs the importance of the predicted UPF future service duration.

[0090] δ(Delta): Data network path factor weight. It weighs the importance of the path cost from the UPF to the target data network (DN). For example, hop count: the number of routing nodes required between the UPF and the DN. Fewer hops generally indicate a better path.

[0091] Control surface anchoring strategy:

[0092] GNN first determines whether there exists an AMF that can simultaneously serve the current source gNB and the future target gNB sequence within the prediction time window.

[0093] If it exists, anchor to that AMF first. In this case, the subsequent handover process should be completed through Xn Handover as much as possible to avoid N14 (AMF<----->AMF) interface signaling.

[0094] If no such AMF exists (e.g., the UE is about to move out of the current AMF's service area), the N2Handover procedure is triggered in advance based on prediction, and an AMF that can provide the UE with the most persistent service in the subsequent time is selected as the target. The selection principle also considers the AMF's load and connectivity with the target gNB.

[0095] Step 3: Switching triggered by the network side:

[0096] Under strict 3GPP specifications, the AMF itself will not initiate a handover procedure "unnecessarily" using "prediction-based policy guidance" and "pre-configuration" mechanisms. The handover procedure always originates from the Radio Access Network (RAN), i.e., the source gNB.

[0097] Therefore, this embodiment incorporates a "network-side guidance, RAN-side decision-making" mechanism to fully comply with 3GPP specifications. This "indirect triggering" mechanism ensures complete compatibility with 3GPP specifications.

[0098] The following is a detailed analysis of compliance for the three scenarios:

[0099] Detailed description of handover execution in accordance with 3GPP specifications;

[0100] The intelligent engine does not replace standard network elements, but rather influences the triggering and execution of standard processes through pre-configuration and policy guidance, thereby achieving intelligent decision-making.

[0101] Core principle: The intelligent engine "guides" the UE and gNB to generate measurement reports and handover decisions that conform to their predictive decisions by "indirectly triggering" the radio resource management (RRM) policy of the control source gNB.

[0102] Detailed compliance requirements for each scenario:

[0103] (1) Scenario A: Only gNB changes (Xn switching):

[0104] Pre-configuration: The intelligent engine will notify the source AMF of the decision (target gNB-B), and the intelligent engine or OAM will forward the instruction to the source gNB-A. The base station can then modify the neighbor cell measurement configuration for the UE, for example, by increasing the RSRP offset value for the target gNB-B at a specific time point, or by other means such as forcing a handover directly through the RRC reconfiguration message.

[0105] Standard trigger: The source gNB-A executes this policy by modifying the UE's measurement configuration via RRC signaling. After the UE performs the measurement, it reports the measurement earlier because the target cell signal "looks better".

[0106] Standard Implementation: Upon receiving this report, the source gNB-A decides to initiate a handover based on the standard handover algorithm. Since it knows through the N2 interface that the target gNB-B is served by the same AMF, it initiates the Xn-based Handover procedure in accordance with the 3GPP 23.502 standard.

[0107] The role of the intelligent engine is to proactively optimize RRM parameters, while subsequent processes are entirely executed by gNB and AMF in accordance with standard specifications.

[0108] (2) Scenario B: gNB and UPF change, AMF anchoring (N2 switching):

[0109] Pre-configuration and preparation: The intelligent engine will notify the source AMF and the service SMF of the decision (target gNB-B, target UPF-X). The SMF can pre-establish an N4 session with the target UPF-X. At the same time, the source gNB-A is guided to modify the measurement configuration, such as increasing the RSRP bias of gNB-B.

[0110] Standard Trigger: After the UE reports the measurement report, the source gNB-A decides to initiate a handover. Because the target gNB-B is not within its Xn connection range (or based on other policies), it sends a "Handover Required" message (N2 interface) to the source AMF.

[0111] Standard execution: The source AMF initiates the N2 Handover process. During the path switching phase, the SMF, based on the pre-received decisions and prepared resources, naturally triggers the "SMF-initiated UPF Re-selection" process to complete the UPF migration.

[0112] The gNB triggers N2 handover, and the SMF controls UPF reselection; both are standard procedures. The intelligent engine provides optimized target selection and resource pre-configuration.

[0113] (3) Scenario C: Both gNB and AMF are changed (Inter-AMF N2 switching):

[0114] Pre-configuration and preparation: The intelligent engine notifies source AMF-1 of the decision (target gNB-B, target AMF-2, target UPF-Y). Source gNB-A is then guided to modify its measurement configuration.

[0115] Standard trigger: The UE reports a measurement report, the source gNB-A decides to switch, and sends "HandoverRequired" to the source AMF-1.

[0116] Standard execution: Source AMF-1 selects target AMF-2 based on the pre-received decision and sends a "Handover Request" message (N14 interface) to the former. Subsequent procedures fully follow the Inter-AMFN2 Handover procedure defined in 3GPP 23.502, with the two AMFs cooperating to complete the context transfer and establish connections to the target gNB and UPF.

[0117] In the standard process, the source AMF-1 is already responsible for selecting the target AMF; this embodiment simply provides it with the optimal selection.

[0118] In summary, this embodiment is compatible with the 3GPP core network specifications during the handover execution phase. The main reasons are:

[0119] An intelligent prediction and decision-making layer (AI) is introduced to make network reconfiguration decisions, which sits on top of the standard network elements.

[0120] This intelligent layer "guides" standard 3GPP procedures (Xn handover, N2 handover, Inter-AMF handover, and SMF-triggered UPF reselection) to be triggered and executed in a globally optimal direction by pre-configuring network resources (such as pre-establishing tunnels) and optimizing RRM strategies (such as adjusting RSRP offset).

[0121] All specific signaling interactions are still strictly completed by standard network elements such as gNB, AMF, and SMF in accordance with 3GPP specifications.

[0122] In this embodiment, the GNN can be any one of the following or a combination of graph convolutional networks (GCN), graph attention networks (GAT), spatiotemporal graph neural networks (ST-GNN).

[0123] The search for the "optimal path" can be accomplished not only by Genetic Neural Networks (GNNs), but also by introducing or combining traditional heuristic search algorithms, such as the A* algorithm, Dijkstra's algorithm, and genetic algorithms, to find the optimal or near-optimal path on a graph G(V, E, W) that satisfies constraints such as time delay and load. GNNs can be used to predict edge weights, and then traditional algorithms can be used for path search.

[0124] The loss function of GNN can be designed as a multi-objective weighted sum, combined with cross-entropy loss to optimize node selection of AMF / UPF / gNB, and mean squared error loss to accurately predict handover timing, thereby ensuring the global optimality of mobility management decisions.

[0125] In conjunction with measurement reports, the prediction results of GNN can be used as a gate, triggering the UE to report detailed neighbor cell measurements only when the prediction confidence is low or anomalies occur, thereby significantly reducing the signaling overhead of regular measurement reports.

[0126] In another application scenario, taking an NTN consisting of a low Earth orbit (LEO) satellite constellation as an example, some of the satellites are equipped with UPF functionality.

[0127] The method is implemented on the network side and includes:

[0128] (1) Initial prerequisites:

[0129] like Figure 1 As shown, the Intelligent Mobility Management Engine is deployed as a network function. It interfaces with all AMF / SMF and UPF, autonomous management systems, and subscribes to the following data:

[0130] Obtain the UE's GPS location data from the LMF or the UE's location reported by the NG interface;

[0131] Obtain ephemeris data (GPS location data) for all satellites from autonomous management.

[0132] Load status (CPU, memory, service specifications) periodically reported by all AMF and UPF.

[0133] Indirectly obtain base station load information from gNB NGAP (via AMF);

[0134] Real-time status and transmission latency of all network links (traditional detection mechanism).

[0135] (2) Data collection and Kalman filter prediction:

[0136] The engine maintains an independent Kalman filter for each UE and satellite (it can also maintain the filter for a certain range of UE representatives through clustering, which will further reduce the amount of computation).

[0137] Taking a satellite as an example, its Kalman filtering process is as follows:

[0138] State vector: x k =[lon k ,lat k ,alt k ,v lon,k ,vlat,k ,v alt,k ] T That is, the longitude, latitude, altitude and their rate of change (velocity) at time k, with each velocity initialized to 0.

[0139] State transition equation (process model): x k =Ax k-1 +Bw k-1 ;

[0140] This state equation infers the current state based on the state and control variables from the previous moment. k-1 It is noise that follows a Gaussian distribution, also known as process noise, and is the noise used to predict the process. It corresponds to x. k The noise in each component is Gaussian white noise w with an expectation of 0 and a covariance of Q. k-1 ~N(0,Q), where Q is the excitation noise. A is the state transition matrix, and B is the control input matrix.

[0141] For satellites, it is built based on their accurate orbital dynamics model;

[0142] For satellite and ground-based UEs, this can be simplified to a uniform motion model (e.g., A= ).

[0143] Observation equation: z k =Hx k +v k .

[0144] z k Here are the GPS observations (longitude, latitude, altitude) at time k. H is the observation matrix, where H = ( v k It is observation noise, which follows a high-speed distribution, v k ~N(0, R), where R is the measurement noise.

[0145] Recursive estimation, the Kalman filter algorithm consists of two steps: prediction and update;

[0146] Prediction (Time Update): Estimate the state at the current time (time k) based on the posterior estimate of the previous time step (time k-1), thus obtaining the prior estimate at time k;

[0147] Update (Measurement Update): Use the measurement value at the current time to correct the estimate in the prediction stage and obtain the posterior estimate at the current time.

[0148] The filter operates recursively through two steps: "prediction" and "update," outputting the optimal state estimate for multiple future time points (e.g., the next 30 seconds). That is, predicting the trajectory.

[0149] (3) Network graph construction and GNN intelligent decision-making:

[0150] Graph Construction: Based on the predicted UE and satellite locations, a dynamic network relationship graph G(V, E, W) centered on the UE is constructed.

[0151] Node V includes the UE, the currently serving satellite gNB-A, the predicted target satellites gNB-B and gNB-C, AMF-1 serving gNB-A, AMF-2 serving gNB-B, and multiple candidate UPFs (UPF-X is on a satellite, UPF-Y is on the ground).

[0152] Edge E: (gNB-A,AMF-1), (gNB-B,AMF-1): N2 interface connection, multiple base stations are jointly managed by one large AMF.

[0153] (gNB-A,AMF-1),(gNB-B,AMF-2): N2 interface connection, different base stations in different core networks under special scenarios or boundary conditions.

[0154] (AMF-1,UPF-X), (AMF-1,UPF-Y): Same core network, different service UPFs; (AMF-1,UPF-X), (AMF-2,UPF-Y): Different core networks, different service UPFs; (UPF-X,DN-1), (UPF-X,DN-2): Same UPF, but the DN position has changed; (UPF-X,DN-1), (UPF-Y,DN-2): Different UPFs, different DN positions have changed; (Star-A, Star-B): Inter-satellite links.

[0155] Edge weight W: For a wireless link edge (such as gNB-UE), the weight = load factor + predicted path loss.

[0156] For wired / logical link edges (such as gNB-AMF, UPF-DN), the weight = latency + link distance + node load, etc.

[0157] GNN Decision: The GNN model processes graph G.

[0158] In most cases where the UE does not cross AMFs, the GNN identifies that the UE is within the same AMF-1 service area, and the AMF can cover the dynamic mobility of gNB-A and gNB-B within a certain range. Therefore, it outputs a decision:

[0159] Trigger type: Xn toggle;

[0160] Triggering timing: 10 seconds in advance (time reserved for Xn signaling interaction);

[0161] Target: gNB-B, UPF-Y (UPF-Y was chosen because it has the lowest real-time load and the shortest latency to other DNs on the ground or on the satellite).

[0162] In this cross-AMF scenario, the GNN identifies that the UE is moving from the service area of ​​AMF-1 to the service area of ​​AMF-2, and there is no AMF that can simultaneously cover both gNB-A and gNB-B. Therefore, it outputs a decision:

[0163] Trigger type: N2 switching;

[0164] Triggering timing: 10 seconds in advance (to allow time for N2 signaling interaction);

[0165] Target: gNB-B, AMF-2, UPF-Y (UPF-Y was chosen because it has the lowest real-time load and the shortest latency to the ground DN).

[0166] (4) Handover Execution: Detailed description of cross-AMF (N2 handover) scenario:

[0167] The network side triggers the UE handover process through location policies or radio resource management (RRM) policies, rather than directly through the core network.

[0168] Pre-configuration and resource reservation: Before the predicted handover time (T-10 seconds), the intelligent engine sends the decision (UE-ID, target gNB-B, target AMF-2) to the source gNB-A. Gnb-1 uses a preset strategy (modifying the dBM to trigger the RRC reconfiguration message) mapping table: "At time T-5 seconds, modify the neighbor cell measurement configuration for UE-ID, and offset the RSRP value of the target gNB-B (e.g., increase it by 3-5dB)."

[0169] B. Indirect triggering of measurement reporting:

[0170] At time T-5, the source gNB-A sends a modified measurement configuration to the UE via an RRC reconfiguration message, artificially increasing the signal quality offset of gNB-B.

[0171] The UE will meet the triggering conditions of the A3 event (neighbor cell quality is better than the serving cell by an offset) earlier when it performs measurements according to this configuration.

[0172] The UE then sends a measurement report to the network, indicating that the signal quality of gNB-B is better than that of the current serving cell.

[0173] C. Standard N2 switchover process initiated:

[0174] Upon receiving this measurement report, the source gNB-A decides to initiate a handover according to the 3GPP standard procedure. Since it knows that the target gNB-B is served by another AMF (AMF-2), it sends a "Handover Required" message (N2 interface) to the source AMF-1.

[0175] Based on the anticipated decision, the source AMF-1 selects the target AMF-2 and sends a "Handover Request" message (N14 interface) to the former, carrying the UE context.

[0176] Target AMF-2 interacts with target gNB-B, allocates resources, and collaborates with SMF to determine the use of target UPF-Y. Then, it returns a "Handover Request Acknowledge" to source gNB-A via source AMF-1.

[0177] The subsequent process is carried out according to the standard: the source gNB-A sends a "Handover Command" to the UE, the UE synchronizes to the target gNB-B, completes random access, and completes path handover on the target side.

[0178] D. The advantages of the solution in this scenario or the expected results:

[0179] Forward-looking: By making decisions 10 seconds in advance and pre-configuring gNBs, sufficient processing time is given for N2 / N14 signaling, overcoming the large satellite-to-ground latency.

[0180] Resource optimization: The selection of UPF-Y is based on a global load and latency view, ensuring user plane performance.

[0181] Seamless compatibility: The entire process is ultimately triggered by a standard UE measurement report, fully compliant with 3GPP specifications. The core network (intelligent engine) achieves intelligent guidance by indirectly influencing the radio access network (RRM policy), avoiding non-standard interfaces.

[0182] (5) Switching Execution: Details of scenarios without crossing AMF (Xn switching), provided that the Xn interface is involved:

[0183] In this scenario, the intelligent decision-making module predicts that the UE's best serving satellite is about to change, but the source satellite (gNB-A) and the target satellite (gNB-B) are managed by the same AMF (AMF-1). The goal is to perform the most efficient Xn handover.

[0184] A. Pre-configuration and resource reservation:

[0185] The intelligent engine sends the decision (UE-ID, target gNB-B) to the source gNB-A before the predicted switching time (e.g., T-5 seconds).

[0186] B. Indirect triggering and rapid execution:

[0187] Similar to the cross-AMF scenario, at time T-2, the source gNB-A modifies the measurement configuration for the UE via an RRC reconfiguration message, appropriately increasing the RSRP measurement offset of the target gNB-B.

[0188] The UE quickly met the A3 event conditions and sent a measurement report to the source gNB-A.

[0189] Since the source gNB-A was aware of the handover and its resources were ready, it no longer needed to interact with the AMF via N2 signaling. Instead, it directly sent a "Handover Request" message to the target gNB-B through the Xn interface.

[0190] The target gNB-B uses the pre-allocated resources to immediately reply "HandoverCommand" to the UE through the source gNB-A.

[0191] C. Complete the switch:

[0192] After receiving the "Handover Command", the UE uses the C-RNTI pre-assigned by the target gNB-B to directly perform random access and complete uplink synchronization.

[0193] After successfully accessing the target gNB-B, the UE sends a "Handover Confirm" message.

[0194] The target gNB-B then sends a "Path Switch Request" to AMF-1, requesting that the user plane path be updated from "gNB-A->UPF" to "gNB-B->UPF".

[0195] AMF-1 coordinates the SMF and UPF to complete the path handover. Simultaneously, the target gNB-B notifies the source gNB-A to release the UE context, completing the entire handover process.

[0196] D. The advantages of the solution in this scenario:

[0197] Extremely low handover latency: Since the signaling interaction between the N2 and N14 interfaces is eliminated and the target side resources are pre-allocated, the entire handover process (from the UE sending the measurement report to the target side sending the Handover Confirm) is compressed to an extremely short time, and the service interruption is barely noticeable.

[0198] Extremely low signaling overhead: The core network control plane (AMF / SMF) only participates lightly in the final path handover phase, and most of the process is completed in the access network through the Xn interface, which greatly saves core network resources and satellite-to-ground signaling resources.

[0199] Achieving a seamless experience: This process perfectly embodies the core idea of ​​"predictive-pre-execution" in this embodiment. By taking action in advance, it minimizes the waiting latency in the standard process, making it very suitable for the high-dynamic environment of NTN and providing a solid continuity guarantee for services such as high-definition voice and real-time gaming.

[0200] (6) Switching Execution: Detailed Explanation of UPF Selection and Migration Scenarios:

[0201] This scenario focuses on the intelligent selection and migration of User Plane Functions (UPFs). This process can occur during handovers across or without AMFs and is crucial for ensuring service continuity. The Generated Neural Network (GNN) determines, based on latency, load, and path stability, that a better UPF service node needs to be selected for the UE.

[0202] UPF migration is controlled by the Session Management Function (SMF) and is performed as a "Session Update" process during the handover procedure.

[0203] A. Intelligent Decision-Making and SMF Notification:

[0204] When making a switching decision (e.g., switching to gNB-B), the GNN module of the intelligent engine simultaneously outputs the UPF selection decision. This decision is based on a comprehensive scoring function:

[0205] Score(UPF) = α*(1-Norm(delay)) + β*(1-Norm(load)) + γ*Norm(stability) + δ*(1-Norm(DN path cost)).

[0206] Assume it calculates that the score of the spaceborne UPF-X is higher than that of the currently used ground-based UPF-Y.

[0207] The engine sends the final decision (target gNB-B, target UPF-X) to the SMF serving the UE via the Nsmf interface.

[0208] B. Session update process initiated by SMF:

[0209] After receiving the decision, the SMF determines that a UPF relocation is required. This process can be triggered in parallel with the N2 handover process, and is tightly coupled.

[0210] SMF first interacts with the target UPF-X, establishes or modifies the process through the N4 interface session, pre-establishes downlink tunnel information for N3 (gNB-UPF) and N9 (if there is an intermediate UPF) tunnel on the target UPF-X, and configures the necessary packet detection and forwarding rules.

[0211] At the same time, SMF interacts with the current UPF-Y to pre-establish a forward transmission tunnel, which is used to forward data packets from UPF-Y to UPF-X during handover to ensure that no data is lost.

[0212] C. Complete the UPF migration during the switchover process:

[0213] When the switching process reaches the path switch stage (i.e., after the target gNB-B sends a "PathSwitch Request" to the AMF), the AMF will forward this request to the SMF.

[0214] SMF is performing a critical user plane switch at this moment:

[0215] It sends an N4 session modification command to the target UPF-X, activating its user plane functionality and enabling it to begin sending downlink data to the target gNB-B.

[0216] It notifies the target gNB-B via AMF to activate the uplink tunnel to the target UPF-X.

[0217] It updates the rules of the current UPF-Y, stops sending data to the source gNB-A, and begins forwarding all received uplink data to the target UPF-X through a forward tunnel.

[0218] Ultimately, the user plane path is smoothly updated from "source gNB-A->UPF-Y->DN" to "target gNB-B->UPF-X->DN".

[0219] D. Subsequent cleanup:

[0220] After a brief forwarding period, once the SMF confirms that all data has been transmitted via the new path, it initiates the process of releasing the old UPF-Y resources.

[0221] E. The advantages of the solution are demonstrated in UPF migration scenarios:

[0222] User plane performance optimization: By dynamically migrating the UPF to nodes with lower latency, lighter load, or more stable paths (such as migrating from a ground-based UPF to a spaceborne UPF), end-to-end latency is directly reduced and data transmission rate is improved.

[0223] Seamless business continuity: Through pre-established tunnels and forward transmission mechanisms, the migration of UPF is transparent to ongoing business sessions and imperceptible to users, achieving true business continuity.

[0224] Efficient utilization of network resources: It enables on-demand and dynamic scheduling of user plane resources, avoids the situation where some UPFs are overloaded while others are idle, and improves the utilization rate of the entire NTN user plane infrastructure.

[0225] This embodiment has the following effects:

[0226] Achieve seamless handover: By proactively predicting and establishing resources in advance, key handover processes are moved forward, effectively overcoming large satellite-to-ground latency and significantly reducing the perception of business interruption.

[0227] Significantly reduced signaling overhead: Intelligent anchoring of core network elements reduces unnecessary N2 / N14 handovers; predictive gating reduces UE measurement reports, saving air interface and core network resources.

[0228] Global optimal decision-making: By using GNN to learn the network graph globally and optimize the multi-objective loss function, the ping-pong effect is fundamentally avoided, and joint optimal management of the access network and core network is achieved.

[0229] Enhanced business continuity: Especially in the context of spaceborne core network, the intelligent migration of UPF and the precise anchoring of AMF ensure the continuity and quality of end-to-end sessions during high-speed movement of satellites and network elements.

[0230] In this document, the use of "or" in a list of items beginning with "at least one" indicates a separate list, such that a list of, for example, "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not imply that the described example is preferred or better than other examples.

[0231] It should also be noted that in the systems and methods of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions to this disclosure.

[0232] Various changes, substitutions, and modifications can be made to the techniques described herein without departing from the teachings defined in this embodiment. Furthermore, the scope of this embodiment is not limited to the specific aspects of the processes, machines, manufacturing processes, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufacturing processes, events, means, methods, or actions that perform substantially the same functions or achieve substantially the same results as the corresponding aspects described herein can be utilized. Therefore, this embodiment includes such processes, machines, manufacturing processes, events, means, methods, or actions within its scope.

[0233] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0234] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A method for intelligent mobility management and service continuity assurance for NTN, characterized in that, This is executed by the network-side intelligent mobility management engine, and the methods include: Based on the real-time location information of user equipment and network nodes, a prediction algorithm is used to predict the location within a set time in the future, and a dynamic network relationship graph is constructed based on the prediction results. The dynamic network relationship graph is analyzed using a graph neural network to obtain network reconfiguration decisions; Based on the network reconfiguration decision, network devices are indirectly guided to trigger a handover process compliant with 3GPP standards to perform handover and core network element migration. The network reconfiguration decision includes at least a predictive optimal access link sequence and / or a globally optimized core network element selection strategy; The predictive optimal access link sequence includes: By using graph neural networks to learn the dynamic laws of network topology, we can predict the sequence of access links that maximizes the multi-objective optimization function within a future time window. The multi-objective optimization function is used to jointly maximize link continuity and service quality, and minimize switching costs; The globally optimized core network element selection strategy includes a control plane anchoring strategy, which includes: Prioritize anchoring to access and mobility management functions that can simultaneously serve the current source base station and the future target base station sequence within the predicted time window, in order to trigger the Xn handover process; If there is no access and mobility management function that triggers the Xn handover process, the N2 handover process will be triggered in advance, and the access and mobility management function that can provide the user equipment with the most persistent service in the subsequent time will be selected as the target. The indirect guidance includes: The network-side intelligent mobility management engine sends the decision results to the source base station and / or the core network control plane; According to the decision, the source base station and / or core network control plane modify the neighbor cell measurement configuration of the user equipment through radio resource control signaling, so as to prompt the user equipment to report the measurement report that meets the handover conditions in advance. Upon receiving the measurement report, the source base station and / or core network control plane triggers and executes the corresponding handover procedure according to the 3GPP standard.

2. The method for intelligent mobility management and service continuity assurance for NTN according to claim 1, characterized in that, The dynamic network relationship diagram includes: The network node set V, the set of connections between nodes E, and the set of weights of the connecting edges W, where the weights in the weight set W are multi-dimensional weighting factors that include transmission delay and node load.

3. The method for intelligent mobility management and service continuity assurance for NTN according to claim 2, characterized in that, The prediction algorithm used in the prediction algorithm for predicting the location within a set time period in the future is used to predict and track the location trajectory of network participants, including user equipment, spaceborne base stations, spaceborne core networks, spaceborne routing, DN networks, ground stations and ground core networks. The prediction algorithm includes the Kalman filter algorithm. The network node set V of the dynamic network relationship graph includes user equipment, serving and neighboring satellite base stations, access and mobility management functions, session management functions, user plane functions, and data network nodes; the connection relationship set E between nodes represents the logical or physical connection between nodes; the weight set W includes at least the transmission delay between nodes and the node load, and the weight set W also includes one or more of the following: link distance, link bandwidth, and historical stability.

4. The method for intelligent mobility management and service continuity assurance for NTN according to claim 1, characterized in that, The globally optimized core network element selection strategy includes: The user-facing feature selection strategy is based on a comprehensive scoring function to select the optimal user-facing feature. The comprehensive scoring function is as follows: Score(UPF) = α*(1-Norm(Delay)) + β*(1-Norm(Load)) + γ*Norm(Stability) + δ*(1-Norm(DN Path Cost)); where α is the delay factor weight, β is the load factor weight, γ is the stability factor weight, δ is the data network path factor weight, the Norm function maps each original value to the interval [0,1], and Score(UPF) is the comprehensive scoring function.

5. The method for intelligent mobility management and service continuity assurance for NTN according to claim 1, characterized in that, The process of indirectly guiding network devices to trigger a 3GPP-compliant handover procedure based on the network reconfiguration decision includes: By pre-configuring network resources and / or optimizing radio resource management policies, network devices can be indirectly guided to trigger a handover process compliant with 3GPP standards.

6. The method for intelligent mobility management and service continuity assurance for NTN according to claim 1, characterized in that, The switching process includes at least one of the following scenarios: Scenario A: For Xn handover involving only a change in the base station, the source base station directly interacts with the target base station through the Xn interface after receiving the measurement report; Scenario B: N2 handover of base station and user plane functions is triggered by the source base station, completed through the cooperation of the source access and mobility management function and the target access and mobility management function, and the reselection and migration of user plane functions is controlled by the session management function. Scenario C: Inter-AMF N2 handover where both the base station and access and mobility management functions change is triggered by the source base station and completed through the N14 interface in cooperation between the source access and mobility management functions and the target access and mobility management functions.

Images