A ground-non-ground fusion network high-speed terminal seamless handover method and system based on trajectory prediction and resource reservation

Abstract

The application discloses a ground-non-ground fusion network high-speed terminal seamless switching method and system based on trajectory prediction and resource reservation, and belongs to the technical field of space-ground integrated communication. The core of the method is: through bidirectional motion prediction of the terminal and the NTN node, the switching area is accurately judged; the terminal speed, network delay and other factors are combined to adaptively calculate the dynamic switching window; the AI-enabled link scoring model is used to intelligently select the optimal target link; and the resource is cooperatively pre-reserved and uplink and downlink synchronization is completed before the window. The application has the beneficial effects that: the failure problem of traditional RSRP judgment in the NTN dynamic environment is overcome, the switching success rate and prediction accuracy in the high-speed mobile scene are significantly improved, the service interruption time and signaling overhead are greatly reduced, the continuity and reliability of communication are effectively guaranteed, and the application is suitable for intelligent networked vehicles, unmanned aerial vehicles and other high-dynamic service scenarios in 6G networks.

Description

Technical Field

[0001] This invention belongs to the field of converged communication technology of terrestrial network (TN) and non-terrestrial network (NTN), and relates to a high-speed seamless handover method for TN-NTN converged networks based on trajectory prediction and resource pre-reservation. Background Technology

[0002] As 5G evolves into 6G, NTN is gradually becoming an important supplement to global communication infrastructure. Through Low Earth Orbit (LEO), Medium Earth Orbit (MEO), Geostationary Earth Orbit (GEO), and High Altitude Platform Station (HAPS) systems, NTN can provide communication connectivity to areas not covered by traditional cellular networks, such as oceans, mountains, deserts, and disaster areas. The TN-NTN converged network architecture is expected to achieve integrated air-space-ground wide-area access, supporting future large-scale IoT, intelligent vehicle-mounted systems, emergency communications, and other application scenarios.

[0003] However, during the process of high-speed mobile terminals (such as high-speed trains, vehicle-mounted terminals, and aircraft) accessing the TN-NTN converged network, the high-speed movement characteristics of platforms such as LEO satellites cause their coverage areas to change rapidly, and terminals frequently cross the boundaries between terrestrial and aerial networks, resulting in a large number of handover (HO) events. This frequent handover across heterogeneous systems presents the following challenges:

[0004] 1) Traditional handover decision mechanism fails: Current cellular systems mainly make handover decisions based on Reference Signal Received Power (RSRP) or Signal to Interference plus Noise Ratio (SINR). However, in NTN scenarios, due to factors such as wide beam coverage, poor signal stability, and dynamic changes in propagation delay, RSRP fluctuations and distortions are significant, which can easily cause the handover timing to be advanced or delayed, resulting in dropped calls or the Ping-Pong Effect.

[0005] 2) Lack of handover modeling for converged scenarios of "high speed-heterogeneous-dynamic": Although existing research has begun to focus on mobility management under NTN, it is mostly concentrated on scenarios such as static or low-speed terminals accessing LEO constellations and high-altitude platforms. There is still little research on handover modeling and strategy design under the coupling conditions of three factors: "high-speed terminal + dynamic handover boundary + heterogeneous protocol differences", and there is a lack of systematic framework and mechanism.

[0006] 3) Insufficient resource preparation leads to a high handover failure rate: The current TN or NTN side lacks resource reservation and uplink / downlink synchronization mechanisms during the handover process, resulting in delayed response from the target network or failure to establish links in a timely manner. Connection failures are particularly likely to occur during high-speed handover windows, affecting the continuity of critical services.

[0007] 4) Significant issues with terminal load and energy consumption: Frequent signal measurement, location update and handover request reporting significantly increase terminal-side energy consumption, posing a challenge to low-power Internet of Things (IoT) devices; at the same time, when terminal clusters are moving at high speeds (such as convoys or fleets), signaling storms will be triggered, increasing the network-side scheduling load.

[0008] In summary, current TN-NTN converged networks lack a seamless handover method that can adapt to high-speed mobile scenarios, combine trajectory prediction mechanisms with resource pre-configuration strategies, dynamically adjust handover windows, and effectively avoid handover failures. Therefore, a novel strategy suitable for highly dynamic handover scenarios in TN-NTN converged architectures is urgently needed. This strategy should combine terminal motion state prediction to identify potential handover windows in advance, perform cross-domain resource pre-configuration, and ensure coordinated establishment of uplink and downlink connections. This would reduce handover failure rates and signaling load, and improve the connection stability and communication continuity of high-speed terminals. Summary of the Invention

[0009] This invention proposes a seamless handover method for TN-NTN converged networks suitable for high-speed mobile terminals, aiming to address issues such as connection interruptions, handover failures, increased signaling load, and excessive energy consumption caused by frequent handovers in integrated space-ground communication environments. The method integrates multiple mechanisms, including trajectory prediction, handover window estimation, candidate link scoring, and resource pre-reservation. It can dynamically sense the terminal's movement trend in high-speed scenarios, pre-completing target link preparation and uplink / downlink synchronization, thereby achieving a seamless handover process with low latency and low interruption rate. The technical solution of this invention mainly includes the following steps:

[0010] A seamless handover method for high-speed terminals in a terrestrial-non-terrestrial fusion network based on trajectory prediction and resource reservation includes the following steps: acquiring real-time motion state data of the terminal and combining it with ephemeris information from the non-terrestrial network to predict the terminal's motion trajectory within a future time window; dynamically estimating the handover time window based on the predicted trajectory and the coverage boundaries of the terrestrial and non-terrestrial networks; scoring candidate links based on their communication state parameters using a scoring function or neural network model and selecting the optimal target link; initiating a resource reservation and access preparation process to the target link, including identity authentication, resource allocation, synchronization signal configuration, and data bearer pre-establishment; performing seamless handover when the handover triggering conditions are met, and activating a backup link or avoidance strategy when the handover fails.

[0011] Furthermore, the motion state data includes position, velocity, acceleration, and heading angle, acquired by at least one sensor selected from GNSS, IMU, or wheel speedometer.

[0012] Furthermore, the trajectory prediction is implemented using an extended Kalman filter (EKF) or a long short-term memory network (LSTM) model.

[0013] Furthermore, the estimation formula for the switching time window is as follows: T switch To switch time windows, d min To predict the distance between the trajectory and the nearest point on the boundary of the target network, The current driving speed of the terminal. For control plane response delay, Considering the physical synchronization time overhead, the upper and lower bounds of the switching window can be derived as follows: ;

[0014] Among them, T start Indicates the start time of the window switching; T end To indicate the end time of the window switching, This represents the switching time margin, used to buffer uncertainties caused by changes in link state and control flow delays. The entire timeline is divided into three phases: "observation period," "preparation period," and "switching window period." The control module follows... and Dynamically triggered behavior.

[0015] Furthermore, the communication status parameters of the candidate link include at least three of the following: signal-to-noise ratio, load ratio, data rate change rate, historical handover success rate, and number of handover failures.

[0016] Furthermore, the scoring is implemented through a lightweight neural network model, which takes a link feature vector as input and outputs a link score value for link priority ranking.

[0017] Furthermore, the neural network model is trained using historical switching data, the loss function includes a mean squared error term and an L2 regularization term, the model parameters are less than 100KB, and online updates of edge nodes are supported.

[0018] Furthermore, the resource pre-reservation and access preparation process includes:

[0019] Pre-distribute user identity and security key information to target network nodes;

[0020] Reserve radio resource blocks (RBs) and scheduling time slots;

[0021] Pre-configured synchronization signal block (SSB) and channel state information reference signal (CSI-RS);

[0022] Pre-establish data bearer paths for the Packet Data Convergence Protocol (PDCP) and Radio Link Control (RLC) layers.

[0023] Furthermore, the switching triggering condition includes at least one of the following: signal quality threshold, geographical location range, or remaining delay determination.

[0024] A seamless switching system according to the present invention includes:

[0025] The trajectory prediction module is used to predict the future movement trajectory of the terminal;

[0026] The switching window estimation module is used to dynamically calculate the switching time window;

[0027] The link scoring and decision module is used to evaluate and select the optimal link;

[0028] The resource pre-configuration module is used to initiate resource reservation and access preparation to the target network;

[0029] The switching execution control module is used to trigger and execute the switching process, and to start a backup mechanism in case of failure.

[0030] The system is deployed at least once in a terminal, edge node, or core network.

[0031] The aforementioned modules are primarily implemented through software, operating as independent functional units or integrated into the communication protocol stack. Their operating environment can be a terminal operating system (such as Android or Linux), an edge computing platform (MEC), or a cloud-based / virtualized environment of the core network, relying on general-purpose or dedicated processors to execute algorithms. This invention is not limited to a specific hardware form.

[0032] The beneficial effects of this invention are:

[0033] The seamless handover method for high-speed terminals in TN-NTN converged networks proposed in this invention has significant advantages such as high handover success rate, low service interruption risk, and high resource scheduling efficiency. Through a converged trajectory prediction and dynamic handover window estimation mechanism, accurate decision-making can be completed before the terminal enters the handover zone, significantly improving the accuracy and predictability of handover timing and effectively avoiding erroneous handovers caused by delays or fluctuations in traditional RSRP decisions. The introduced resource pre-reservation and uplink / downlink synchronization mechanism can complete communication resource preparation and parameter alignment in the target network, shortening the response latency during handover and enhancing the real-time performance and robustness of handover in high-speed, rapidly changing NTN network environments such as LEO. Simultaneously, the link scoring mechanism based on a multi-factor weighted strategy enables the terminal to intelligently select the optimal target link, improving the system's resource scheduling capability under load awareness. Compared to traditional handover methods that rely on frequent measurements and signaling reporting, this invention reduces the burden and energy consumption on the terminal side, making it particularly suitable for power-sensitive high-speed terminal devices. Each functional module has good decoupling and interface compatibility, and can be flexibly deployed on the terminal side, edge nodes or access network. It has strong engineering feasibility and system expansion capabilities, and is suitable for a variety of converged communication application scenarios such as future 6G NTN systems, low-altitude airspace communication, and intelligent connected vehicle terminals. Attached Figure Description

[0034] Figure 1 System overall architecture diagram;

[0035] Figure 2 Schematic diagram of trajectory prediction and switching area overlay;

[0036] Figure 3 Switch between window and timeline segmented chart;

[0037] Figure 4 Link scoring module structure diagram. Detailed Implementation

[0038] The method of the present invention includes the following steps:

[0039] 1) Trajectory perception and prediction: The terminal acquires its current position, velocity vector, acceleration and heading angle in real time, and performs dynamic modeling by combining historical motion trajectories; at the same time, the system scheduling side acquires the ephemeris orbit information, beam coverage and angle change rate of the NTN network for the current and future period; based on the relative motion relationship between the terminal and the satellite, trajectory prediction models such as Kalman filtering or Long Short-Term Memory Network (LSTM) are used to generate the spatial trajectory estimation sequence of the terminal within the future time window Δt.

[0040] 2) Dynamic estimation of handover window: Based on the predicted trajectory and the coverage boundaries of TN and NTN (including geographical boundaries and beam overlap areas), the potential handover zone where the terminal will enter the heterogeneous network is calculated; the time width of the handover zone is adaptively adjusted according to parameters such as prediction error, safety margin, terminal rate and signal propagation delay to determine the dynamic handover window.

[0041] 3) Candidate Link Screening and Scoring: Based on the expected location of the terminal within the handover window, coverage beam number, available resource pool, estimated channel quality (such as path loss, SINR), and current system load, a candidate link set is constructed. A multi-dimensional weighted scoring function is used to calculate the handover priority of each candidate link, and the link with the highest overall score is selected as the target network link. The optimization strategy supports adaptive weight updates, dynamically adjusting the scoring factor weights based on service type (such as URLLC / eMBB) and terminal type (such as high-reliability, low-latency vehicle-mounted devices, low-power, wide-coverage IoT devices, etc.).

[0042] 4) Target link resource pre-reservation and uplink / downlink synchronization: After the target link is determined, the system negotiates the resource pre-reservation request with the target network (TN or NTN) in advance, reserving spectrum resources, scheduling time slices and access identifiers; at the same time, in order to improve the handover success rate, the system initiates processes such as link authentication preparation, uplink / downlink time synchronization and transmission parameter calibration to ensure that the target network is in a "ready state" when the terminal enters the coverage area.

[0043] 5) Seamless handover execution and confirmation mechanism: When the handover window trigger conditions are met (such as entering a preset geographical area, signal quality threshold switching, delay remaining judgment, etc.), the terminal executes the handover instruction according to the predetermined process; the handover process does not rely on RSRP threshold judgment, does not require repeated authentication or re-initialization of the communication protocol stack, and reduces signaling interaction and system load; after the handover is completed, the system confirms the access status. If the target link fails, a fast reselection and backup link handover mechanism is triggered.

[0044] like Figure 1 As shown, the seamless handover system described in this embodiment mainly includes: terminal equipment, ground base stations (NextGeneration Node B, gNB), non-terrestrial network access nodes (NTN, such as LEO satellites or UAV relays), and a control center deployed in an edge computing platform or core network. The system can adapt to heterogeneous user equipment such as high-speed mobile vehicle terminals, drones, and high-speed rail terminals, and has dynamic trajectory prediction, link quality assessment, and resource coordination capabilities.

[0045] The method includes the following steps:

[0046] Step 1: Terminal trajectory perception and prediction;

[0047] The system acquires real-time motion status data of the terminal and combines it with ephemeris information from non-terrestrial networks to predict the terminal's trajectory within a future time window. The terminal uses multi-source information such as GNSS, IMU, and wheel speedometers for fusion sensing to acquire its own status data, including position, velocity, acceleration, and heading angle, forming a trajectory state vector. ; where s t Let x be the trajectory state vector at time t. t y t Let v be the terminal position at time t. t Let a be the terminal velocity at time t. t Let be the terminal acceleration at time t, and θ. t Let be the terminal heading angle at time t;

[0048] The control center receives this status and makes predictions based on the extended Kalman filter (EKF) to estimate the terminal's future... Location information for each time slot;

[0049] ;

[0050] ;

[0051] in Given the predicted location information of the terminal at time t+k, the output is a set of future trajectory points: ,in The prediction step size is defined by N, the number of prediction steps, and P, the set of predicted path points. Using map services and a network topology database, it is determined when the terminal will enter the TN and NTN coverage boundaries. Figure 2 As shown, the trajectory prediction module, when overlaid with the map boundary, can form a "predicted switching area".

[0052] Step 2: Dynamic estimation of multi-factor switching window

[0053] The handover time window is dynamically estimated based on the predicted trajectory and the coverage boundaries of the terrestrial and non-terrestrial networks.

[0054] The switching window depends not only on the predicted trajectory but also on link jitter margin and network synchronization period. Therefore, the following definition is provided: T switch To switch time windows, d min To predict the distance between the trajectory and the nearest point on the boundary of the target network, The current driving speed of the terminal. For control plane response delay (e.g., 200–300ms). Considering the physical synchronization time overhead (such as receiving the primary synchronization signal PSS / synchronization signal block SSB, etc.), the upper and lower bounds of the switching window can be obtained as follows:

[0055] ;

[0056] Among them, T start Indicates the start time of window switching, T end To indicate the end time of the window switching, This represents the handover margin, used to buffer against uncertainties caused by changes in link state and control flow delays. This parameter typically ranges from 0.5 to 2 seconds and can be dynamically set based on terminal speed, adaptive adjustment strategies, and network feedback. Figure 3 As shown, the entire timeline is divided into three stages: "observation period," "preparation period," and "switching window period." The control module follows... and Dynamically triggered behavior.

[0057] Step 3: Link scoring function and candidate route set generation

[0058] Based on the communication status parameters of candidate links, the candidate links are scored using a scoring function or neural network model, and the optimal target link is selected.

[0059] To ensure communication continuity for high-speed mobile terminals in the TN-NTN converged network, all candidate links must be evaluated for quality and prioritized before handover, the optimal handover target must be selected, and a set of alternative routes must be constructed.

[0060] The overall process for this step is as follows: Figure 4 As shown, it includes three parts: feature extraction module, AI-enabled scoring module, and sorting and fault tolerance management module.

[0061] (1) Feature extraction module;

[0062] First, the system collects all currently available links L i The multidimensional communication state parameters are obtained, and the feature vector fi is constructed as follows: ,in:

[0063] Link signal-to-noise ratio (unit: dB);

[0064] This represents the current load ratio (current RB utilization, ranging from 0 to 1).

[0065] The data rate change rate (Mbps / s) over the past 10 seconds.

[0066] For historical switching success rate;

[0067] This represents the number of times the link failed to switch over in the past minute.

[0068] These features are collected collaboratively by the terminal side, access point (ground base station or NTN node), and core network, and then transmitted to the scoring module after normalization processing.

[0069] (2) Lightweight neural network scoring module;

[0070] To replace the traditional static weighted scoring function, this invention introduces a lightweight neural network scoring model that learns the nonlinear mapping relationship between link features and switching performance. The model structure is as follows:

[0071] Input layer: , Represents the set of real numbers;

[0072] First hidden layer: ,in ReLU is the corrected linear unit;

[0073] Second hidden layer: ,in ;

[0074] Output layer: ,in , b3 is a sigmoid or linear function, and b4 is the output layer bias vector used to adjust the linear transformation result of the output layer, ultimately outputting a score. It is used to rank the links based on their quality.

[0075] To train this scoring module, data from real network simulations or historical handover logs are used as the data source. Labels are generated based on "communication stabilization time after handover" or "number of interruptions." The training method employs periodic online updates from edge nodes, with the terminal downloading a lightweight model copy (<100KB) on demand. The training loss function L is as follows:

[0076] ;

[0077] in This is the score output of the neural network for the i-th candidate link sample; The label for the actual link performance of the i-th sample can be constructed based on the communication stability after switching (such as interruption rate, duration, etc.); n represents the total number of samples participating in training. This is the complete set of trainable parameters for the neural network model, including the weights and biases of each layer. is the regularization coefficient, used to balance the relationship between prediction error and model complexity; its value can be obtained through cross-validation. The loss function consists of two parts: the first term is the mean squared error between the predicted score and the true label, driving the model to learn an accurate scoring strategy; the second term is the L2 regularization term, which suppresses overfitting and improves generalization ability under varying wireless network conditions.

[0078] (3) Link ranking and alternative mechanism;

[0079] After the link scoring is completed, the links with the highest scores are selected by sorting: Where Li is the i-th candidate link. Simultaneously, m backup links are reserved, forming... , where L backup This is a set of alternative links. This link information will be sent in parallel to:

[0080] Step four: Resource pre-configuration module (pre-allocating resources);

[0081] Step 5: Switching the execution module (for fast retry or exception rollback).

[0082] The control center is used for continuous strategy updates and model calibration.

[0083] Step 4: Resource pre-configuration and access preparation;

[0084] Initiate a resource pre-reservation and access preparation process for the target link, including identity authentication, resource allocation, synchronization signal configuration, and data bearer pre-establishment;

[0085] The control center initiates a pre-configuration process to the target link node, including:

[0086] 1) Distribute user identity and key information to complete the authentication context preparation;

[0087] 2) Allocate radio resource blocks (RBs), time slots, and frequency bands;

[0088] 3) Configure synchronization signals (PSS / SSB) and pilot signals;

[0089] 4) Start the data bearer configuration and pre-build the Packet Data Convergence Protocol (PDCP) / Radio Link Control (RLC) path.

[0090] The above sub-processes are executed in order according to logical dependencies, but some operations (such as steps 2 and 3) can be carried out in parallel to shorten the total time.

[0091] Step 5: Switch execution and confirm connection;

[0092] When the handover triggering conditions are met, a seamless handover is performed, and in the event of a handover failure, a backup link or circumvention strategy is enabled.

[0093] When the following switching trigger conditions are met: ,in, This represents the signal-to-noise ratio (SNR) currently measured by the terminal, used to reflect the quality of the existing service link in real time. This represents the preset signal-to-noise ratio (SNR) threshold. When SNR current Drop to below Furthermore, when the time interval is within the set range, the handover process is triggered, and the terminal immediately initiates the process of accessing the target link:

[0094] 1) Directly access the target network using a preset context;

[0095] 2) Establish protocol stack connection (PHY→MAC→RLC→PDCP);

[0096] 3) The control plane completes the AMF or UPF switching;

[0097] 4) Service forwarding resumes after the switchover is complete, with no noticeable interruption.

[0098] If the connection fails, the system will attempt to use an alternative set of links: The number of retries shall not exceed ,in This is the maximum number of retries. If it fails, it will enter a breakpoint buffer transmission or area avoidance strategy (such as aircraft turning to avoid obstacles).

[0099] The timing of the entire TN-NTN seamless handover process can be divided into the following stages:

[0100] 1) Sensing phase: Terminal devices periodically report their position, speed, acceleration and other status data to the control center;

[0101] 2) Prediction phase: Based on the received status data, the control center combines the map database and network coverage information to perform trajectory prediction and determine the time and location of future entry into the TN / NTN boundary;

[0102] 3) Handover window estimation stage: The system dynamically calculates the handover time window by combining terminal speed, network control plane response latency and synchronization time;

[0103] 4) Link scoring and target link selection stage: The control center collects the status characteristics of candidate links in real time, generates scores using an AI scoring model, and selects the primary link and several backup links;

[0104] 5) Resource pre-configuration phase: The system initiates preparatory work such as user context transfer, radio resource reservation, and synchronization signal delivery to the target link node;

[0105] 6) Handover Execution Phase: After the terminal detects the handover trigger condition within the window, it executes the fast access process to complete the protocol stack reconstruction and control plane update;

[0106] 7) Confirmation and Service Recovery Phase: The terminal completes data bearer recovery on the target link. If the handover fails, it enters the alternative route set fallback mechanism or triggers a caching / avoidance strategy.

Claims

1. A seamless handover method for high-speed terminals in ground-non-ground integrated networks based on trajectory prediction and resource pre-reservation, characterized in that, The method comprises the following steps: acquiring real-time motion state data of the terminal, and combining ephemeris information of the non-ground network to predict a motion trajectory of the terminal within a future time window; dynamically estimating a handover time window according to the predicted trajectory and coverage boundaries of the ground network and the non-ground network; scoring candidate links by a neural network model based on communication state parameters of the candidate links, and selecting an optimal target link; initiating resource pre-reservation and access preparation procedures to the target link, including identity authentication, resource allocation, synchronization signal configuration and data bearer pre-establishment; performing seamless handover when a handover trigger condition is met, and enabling a backup link or an avoidance strategy when the handover fails; the motion state data comprises position, speed, acceleration and heading angle, and is acquired by at least one sensor of GNSS, IMU or wheel speed meter; the trajectory prediction is implemented by an extended Kalman filter (EKF) or a long short-term memory (LSTM) network model; The estimation formula of the switching time window is: wherein T switch is the switching time window, d min is the distance between the predicted trajectory and the closest point of the boundary of the target network, is the driving speed of the current terminal, is the control plane response delay, is the physical synchronization time overhead, and thus the upper and lower bounds of the switching window are: ; wherein T start denotes the start time of the switching window; T end denotes the end time of the switching window, δ denotes the switching time margin for buffering the uncertainty caused by the link state change and control procedure delay, the whole timeline is divided into three stages: "observation period", "preparation period", "switching window period", the control module determines the switching time according to the With dynamic triggering behavior.

2. The method of claim 1, wherein, the communication state parameters of the candidate links include at least three of signal-to-noise ratio, load ratio, data rate change rate, historical handover success rate and handover failure times.

3. The method of claim 1 or 2, wherein, The scoring is implemented by a lightweight neural network model, the input of which is a link feature vector, and the output is a link score value, which is used for link priority sorting.

4. The method of claim 3, wherein, The neural network model is trained by historical handover data, the loss function contains a mean square error term and an L2 regularization term, the model parameters are less than 100 KB, and the edge node supports online updating.

5. The method of claim 1, wherein, The resource pre-reservation and access preparation procedures comprise: pre-distributing user identity and security key information to the target network node; pre-reserving radio resource blocks (RBs) and scheduling time slots; pre-configuring synchronization signal blocks (SSBs) and channel state information reference signals (CSI-RSs); pre-establishing data bearer paths of packet data convergence protocol (PDCP) and radio link control (RLC) layers.

6. The method of claim 1, wherein, The handover trigger condition includes at least one of a signal quality threshold, a geographical position interval or a residual delay judgment.

7. A ground-non-ground converged network high-speed terminal seamless handover system based on trajectory prediction and resource pre-reservation, used for implementing the method of any one of claims 1-6, characterized in that, The system comprises: a trajectory prediction module for predicting a future motion trajectory of the terminal; a handover window estimation module for dynamically calculating a handover time window; a link scoring and decision module for evaluating and selecting an optimal link; a resource pre-configuration module for initiating resource pre-reservation and access preparation to a target network; a handover execution control module for triggering and executing a handover procedure, and starting a backup mechanism when the handover fails; the system is deployed in at least one of the terminal, the edge node or the core network.

Images