Wireless link failure prediction method, apparatus, terminal, network device and medium

Abstract

The application provides a wireless link failure prediction method and device, a terminal, a network device and a medium, and relates to the technical field of wireless communication. The wireless link failure prediction method is applied to a terminal, and the method comprises the following steps: sending first information to a first network device, and / or sending second information to the first network device. The first information comprises information related to wireless link failure (RLF) prediction. The second information comprises information related to a first artificial intelligence (AI) model and / or first AI function monitoring. The application defines the process of sending information related to RLF prediction, information related to an AI model and / or information related to AI function monitoring from a terminal to a network device, so as to realize the assistance of the network in switching or configuration optimization by the network device using the predicted RLF result.

Description

Technical Field

[0001] This invention relates to the field of wireless technology, and in particular to a method, apparatus, terminal, network device, and medium for predicting wireless link failures. Background Technology

[0002] When in connected mode, the User Equipment (UE) needs to perform Radio Link Monitoring (RLM) based on the synchronization signal block (SSB), Channel-State Information Reference Signal (CSI-RS), and network-configured thresholds (Qin, Qout). If the UE detects any of the following: physical layer downlink asynchrony, random access failure, or Radio Link Control (RLC), it considers it a Radio Link Failure (RLF).

[0003] Related technologies provide RLF prediction based on Artificial Intelligence (AI), but there is currently no specific procedure to clarify how to send and receive AI-generated RLF prediction information, or to use the predicted RLF results to assist the network in switching or configuration optimization to improve switching performance. Summary of the Invention

[0004] This invention provides a wireless link failure prediction method, apparatus, terminal, network device, and medium to solve the problem in the prior art that there is no specific process for clearly defining the transmission and reception of RLF prediction information, so as to realize the use of predicted RLF results to assist the network in switching or configuration optimization.

[0005] To address the aforementioned technical problems, the embodiments of the present invention provide the following technical solutions:

[0006] In a first aspect, embodiments of the present invention provide a wireless link failure prediction method, applied to a terminal, the method comprising:

[0007] Send first information to the first network device, and / or send second information to the first network device;

[0008] The first information includes information related to radio link failure (RLF) prediction;

[0009] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0010] Optionally, the method further includes:

[0011] The first configuration information sent by the first network device is configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction.

[0012] Optionally, the method further includes:

[0013] The first information is obtained based on the first AI model and / or the first AI function.

[0014] Optionally, the first configuration information includes at least one of the following:

[0015] Measurement configuration information;

[0016] Observation window for RLF prediction;

[0017] The forecast window for RLF forecasting;

[0018] The forecast period for RLF forecasts;

[0019] RLF occurrence probability threshold;

[0020] The first piece of information to be reported;

[0021] The reporting conditions for the first information;

[0022] RRM predicts configuration information.

[0023] Optionally, the reporting conditions for the first information include at least one of the following:

[0024] The difference between the predicted time of RLF occurrence and the current time is less than the preset time difference;

[0025] The terminal receives a first preset number of consecutive link asynchrony indications;

[0026] Meet the reporting cycle;

[0027] RLF was predicted to occur;

[0028] The predicted probability of RLF occurrence is higher than the RLF occurrence probability threshold;

[0029] The reporting time meets the configuration.

[0030] Optionally, the method further includes:

[0031] Based on the first AI model and / or the first AI function, RRM prediction information is obtained;

[0032] The first information is obtained based on the RRM prediction information.

[0033] Optionally, the method further includes:

[0034] Send the first instruction information to the first network device;

[0035] Wherein, the first indication information is used to indicate at least one of the following:

[0036] The terminal is instructed to support RLF prediction based on AI models and / or AI functions;

[0037] The terminal is instructed to support RRM prediction based on AI models and / or AI functions;

[0038] Indicates the AI ​​models and / or AI functions supported by the terminal;

[0039] The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal;

[0040] Input information indicating the AI ​​models and / or AI functions supported by the terminal;

[0041] Output information indicating the AI ​​models and / or AI functions supported by the terminal.

[0042] Optionally, the method further includes:

[0043] Receive the second indication information sent by the first network device;

[0044] The second indication information is used to indicate at least one of the following:

[0045] The terminal is instructed to activate the AI ​​model and / or AI function;

[0046] Instruct the terminal to activate the AI ​​model and / or AI function;

[0047] Instruct the terminal to use the AI ​​model and / or AI function for RLF prediction;

[0048] The terminal is instructed to access the first cell and perform RLF prediction;

[0049] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality;

[0050] The terminal is instructed to activate the AI ​​model and / or AI function based on cell signal quality; the terminal is instructed to activate the AI ​​model and / or AI function based on link asynchrony indication; the AI ​​model and / or AI function is used for RLF prediction;

[0051] AI models and / or AI functions are used for RRM prediction.

[0052] Optionally, the second indication information includes at least one of the following:

[0053] Radio Resource Control (RRC) signaling;

[0054] Media intervention control layer control cell MAC CE;

[0055] Downlink Control Information (DCI);

[0056] System Information Block (SIB).

[0057] Optionally, the first information includes at least one of the following:

[0058] The start status of the first timer;

[0059] The runtime of the first timer;

[0060] Remaining duration of the first timer;

[0061] The number of link synchronization indications received when an RLF is predicted;

[0062] Methods of prediction;

[0063] Predicting whether RLF will occur or not;

[0064] Predict the timing of RLF occurrence;

[0065] Prediction window length;

[0066] Predict the probability of RLF occurring within the window;

[0067] Predict the cell quality when RLF occurs.

[0068] Optionally, the method further includes:

[0069] The system receives second configuration information sent by the first network device, wherein the second configuration information is configuration information related to the second information.

[0070] Optionally, the second configuration information includes at least one of the following:

[0071] The start time of monitoring the first AI model and / or the first AI function;

[0072] The end time of monitoring the first AI model and / or the first AI function;

[0073] Monitoring cycle of the first AI model and / or the first AI function;

[0074] Monitoring termination indication information for the first AI model and / or the first AI function;

[0075] The reporting conditions for the second information.

[0076] Optionally, the reporting conditions for the second information include at least one of the following:

[0077] The monitoring end time configured for the first network device must be met;

[0078] The timer configured on the first network device stops or times out;

[0079] The occurrence of RLF was predicted, and the RLF actually occurred;

[0080] The occurrence of RLF was predicted, but RLF did not actually occur;

[0081] The prediction was that the RLF would not occur, but the RLF actually did occur.

[0082] Optionally, the second information includes at least one of the following:

[0083] The actual time when RLF occurs;

[0084] The cell identifier where the terminal is located;

[0085] After the first timer expires, the number of link synchronization indications received by the terminal;

[0086] Community quality when RLF actually occurs;

[0087] The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information;

[0088] Cell quality prior to the actual time of RLF occurrence;

[0089] The start status of the first timer;

[0090] The runtime of the first timer;

[0091] RLF (Real-Frequency) indication information that did not actually occur;

[0092] The quality of the cell where the terminal is located;

[0093] First Key Performance Indicator (KPI) information.

[0094] Optionally, the first KPI information includes at least one of the following:

[0095] The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality;

[0096] The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence;

[0097] The time difference between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0098] The time difference between the end of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0099] The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring;

[0100] Evaluation metrics for the first AI model and / or the first AI function.

[0101] Optionally, the method further includes:

[0102] The second information sent to the second network device;

[0103] The second network device sends the second information to the first network device;

[0104] The second network device is the network device corresponding to the cell selected by the terminal for reconstruction after the RLF actually occurs.

[0105] Optionally, the method further includes:

[0106] The terminal receives a third indication message sent by the first network device, the third indication message being used to instruct the terminal to perform a first operation;

[0107] The first operation includes at least one of the following:

[0108] Switch to the second AI model and / or the second AI function;

[0109] Activate the second AI model and / or the second AI function;

[0110] To activate the first AI model and / or the first AI function;

[0111] Revert to non-AI operation.

[0112] Optionally, the method further includes:

[0113] The first operation is performed based on the third instruction information and the first KPI information.

[0114] Optionally, the third indication information is used for at least one of the following:

[0115] The terminal is instructed to access the first cell and execute the first operation;

[0116] The terminal is instructed to perform the first operation based on the cell signal quality;

[0117] The terminal is instructed to perform the first operation according to the link asynchrony indication;

[0118] The terminal is instructed to perform the first operation based on the KPI information;

[0119] KPI thresholds.

[0120] Optionally, the method further includes:

[0121] Receive the fourth indication information sent by the first network device;

[0122] The fourth indication information is used to indicate at least one of the following:

[0123] Instruct the terminal to switch to the second AI model and / or the second AI function;

[0124] The terminal is instructed to activate the second AI model and / or the second AI function;

[0125] Instruct the terminal to activate the first AI model and / or the first AI function;

[0126] Instruct the terminal to revert to non-AI operation.

[0127] Secondly, embodiments of the present invention also provide a wireless link failure prediction method, applied to a first network device, the method comprising:

[0128] The receiving terminal sends a first message, and / or the receiving terminal sends a second message;

[0129] The first information includes information related to radio link failure (RLF) prediction;

[0130] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0131] Optionally, the method further includes:

[0132] Send first configuration information to the terminal, the first configuration information being configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction.

[0133] Optionally, the first configuration information includes at least one of the following:

[0134] Measurement configuration information;

[0135] Observation window for RLF prediction;

[0136] The forecast window for RLF forecasting;

[0137] The forecast period for RLF forecasts;

[0138] RLF occurrence probability threshold;

[0139] The first piece of information to be reported;

[0140] The reporting conditions for the first information;

[0141] RRM predicts configuration information.

[0142] Optionally, the reporting conditions for the first information include at least one of the following:

[0143] The difference between the predicted time of RLF occurrence and the current time is less than the preset time difference;

[0144] The terminal receives a first preset number of consecutive link asynchrony indications;

[0145] Meet the reporting cycle;

[0146] RLF was predicted to occur;

[0147] The predicted probability of RLF occurrence is higher than the RLF occurrence probability threshold;

[0148] The reporting time meets the configuration.

[0149] Optionally, the method further includes:

[0150] Receive the first indication information sent by the terminal;

[0151] Wherein, the first indication information is used to indicate at least one of the following:

[0152] The terminal is instructed to support RLF prediction based on AI models and / or AI functions;

[0153] The terminal is instructed to support RRM prediction based on AI models and / or AI functions;

[0154] Indicates the AI ​​models and / or AI functions supported by the terminal;

[0155] The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal;

[0156] Input information indicating the AI ​​models and / or AI functions supported by the terminal;

[0157] Output information indicating the AI ​​models and / or AI functions supported by the terminal.

[0158] Optionally, the method further includes:

[0159] Send a second instruction message to the terminal;

[0160] The second indication information is used to indicate at least one of the following:

[0161] The terminal is instructed to activate the AI ​​model and / or AI function;

[0162] Instruct the terminal to activate the AI ​​model and / or AI function;

[0163] Instruct the terminal to use the AI ​​model and / or AI function for RLF prediction;

[0164] The terminal is instructed to access the first cell and perform RLF prediction;

[0165] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality;

[0166] The terminal is instructed to activate the AI ​​model and / or AI function based on cell signal quality; the terminal is instructed to activate the AI ​​model and / or AI function based on link asynchrony indication; the AI ​​model and / or AI function is used for RLF prediction;

[0167] AI models and / or AI functions are used for RRM prediction.

[0168] Optionally, the second indication information includes at least one of the following:

[0169] RRC signaling;

[0170] MAC CE;

[0171] DCI;

[0172] SIB.

[0173] Optionally, the first information includes at least one of the following:

[0174] The start status of the first timer;

[0175] The runtime of the first timer;

[0176] Remaining duration of the first timer;

[0177] The number of link synchronization indications received when an RLF is predicted;

[0178] Methods of prediction;

[0179] Predicting whether RLF will occur or not;

[0180] Predict the timing of RLF occurrence;

[0181] Prediction window length;

[0182] Predict the probability of RLF occurring within the window;

[0183] Predict the cell quality when RLF occurs.

[0184] Optionally, the method further includes:

[0185] Send second configuration information to the terminal, the second configuration information being configuration information related to the second information.

[0186] Optionally, the second configuration information includes at least one of the following:

[0187] The start time of monitoring the first AI model and / or the first AI function;

[0188] The end time of monitoring the first AI model and / or the first AI function;

[0189] Monitoring cycle of the first AI model and / or the first AI function;

[0190] Monitoring termination indication information for the first AI model and / or the first AI function;

[0191] The reporting conditions for the second information.

[0192] Optionally, the reporting conditions for the second information include at least one of the following:

[0193] The monitoring end time configured for the first network device must be met;

[0194] The timer configured on the first network device stops or times out;

[0195] The occurrence of RLF was predicted, and the RLF actually occurred;

[0196] The occurrence of RLF was predicted, but RLF did not actually occur;

[0197] The prediction was that the RLF would not occur, but the RLF actually did occur.

[0198] Optionally, the second information includes at least one of the following:

[0199] The actual time when RLF occurs;

[0200] The cell identifier where the terminal is located;

[0201] After the first timer expires, the number of link synchronization indications received by the terminal;

[0202] Community quality when RLF actually occurs;

[0203] The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information;

[0204] Cell quality prior to the actual time of RLF occurrence;

[0205] The start status of the first timer;

[0206] The runtime of the first timer;

[0207] RLF (Real-Frequency) indication information that did not actually occur;

[0208] The quality of the cell where the terminal is located;

[0209] First Key Performance Indicator (KPI) information.

[0210] Optionally, the first KPI information includes at least one of the following:

[0211] The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality;

[0212] The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence;

[0213] The time difference between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0214] The time difference between the end of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0215] The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring;

[0216] Evaluation metrics for the first AI model and / or the first AI function.

[0217] Optionally, the method further includes:

[0218] Receive the second information sent by the second network device;

[0219] The first information is sent by the terminal to the second network device;

[0220] The second network device is the network device corresponding to the cell selected by the terminal after the RLF actually occurs.

[0221] Optionally, the method further includes:

[0222] Send a third instruction message to the terminal, the third instruction message being used to instruct the terminal to perform a first operation;

[0223] The first operation includes at least one of the following:

[0224] Switch to the second AI model and / or the second AI function;

[0225] Activate the second AI model and / or the second AI function;

[0226] To activate the first AI model and / or the first AI function;

[0227] Revert to non-AI operation.

[0228] Optionally, the third indication information is used for at least one of the following:

[0229] The terminal is instructed to access the first cell and execute the first operation;

[0230] The terminal is instructed to perform the first operation based on the cell signal quality;

[0231] The terminal is instructed to perform the first operation according to the link asynchrony indication;

[0232] The terminal is instructed to perform the first operation based on the KPI information;

[0233] KPI thresholds.

[0234] Optionally, the method further includes:

[0235] Send a fourth instruction message to the terminal;

[0236] The fourth indication information is used to indicate at least one of the following:

[0237] Instruct the terminal to switch to the second AI model and / or the second AI function;

[0238] The terminal is instructed to activate the second AI model and / or the second AI function;

[0239] Instruct the terminal to activate the first AI model and / or the first AI function;

[0240] Instruct the terminal to revert to non-AI operation.

[0241] Thirdly, embodiments of the present invention also provide a wireless link failure prediction device, the device comprising:

[0242] The first sending module is used to send first information to the first network device, and / or send second information to the first network device;

[0243] The first information includes information related to radio link failure (RLF) prediction;

[0244] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0245] Fourthly, embodiments of the present invention also provide a wireless link failure prediction device, the device comprising:

[0246] The first receiving module is used to receive first information sent by the terminal, and / or to receive second information sent by the terminal;

[0247] The first information includes information related to radio link failure (RLF) prediction;

[0248] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0249] Fifthly, embodiments of the present invention also provide a terminal, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the wireless link failure prediction method as described in any one of the first aspects.

[0250] In a sixth aspect, embodiments of the present invention also provide a network device, including: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the wireless link failure prediction method as described in any one of the second aspects.

[0251] In a seventh aspect, embodiments of the present invention also provide a readable storage medium storing a program that, when executed by a processor, implements the steps of the wireless link failure prediction method as described in any one of the first aspects, or, when executed, implements the steps of the wireless link failure prediction method as described in any one of the second aspects.

[0252] Eighthly, embodiments of the present invention also provide a computer program product, including computer instructions, which, when executed by a processor, implement the steps in the wireless link failure prediction method as described in any one of the first aspects, or, when executed, implement the steps in the wireless link failure prediction method as described in any one of the second aspects.

[0253] The beneficial effects of this invention are:

[0254] The wireless link failure prediction method provided by this invention involves a terminal sending first information and / or second information to a first network device. The first information includes information related to RLF prediction, and the second information includes information related to a first AI model and / or first AI function monitoring. This clarifies the process by which the terminal sends RLF prediction-related information, AI model, and / or AI function monitoring-related information to the network device, enabling the network device to utilize the predicted RLF results to assist in network switching or configuration optimization. Attached Figure Description

[0255] Figure 1 A flowchart illustrating the wireless link failure prediction method for terminals provided in this embodiment of the invention;

[0256] Figure 2 This diagram illustrates the lifecycle of the AI ​​model / AI function provided in this embodiment of the invention.

[0257] Figure 3 This is a flowchart illustrating the terminal capability reporting provided in an embodiment of the present invention;

[0258] Figure 4 This is a flowchart illustrating the wireless link monitoring process provided in an embodiment of the present invention.

[0259] Figure 5 One of the schematic diagrams illustrating the time relationship between the prediction window provided in this embodiment of the invention and the actual time when the RLF occurs;

[0260] Figure 6 The second schematic diagram illustrating the time relationship between the prediction window provided in this embodiment of the invention and the actual time when the RLF occurs;

[0261] Figure 7 The third schematic diagram illustrating the time relationship between the prediction window provided in this embodiment of the invention and the actual time when RLF occurs;

[0262] Figure 8 A flowchart illustrating the wireless link failure prediction method for a first network device provided in an embodiment of the present invention;

[0263] Figure 9 This is one of the structural schematic diagrams of the wireless link failure prediction device provided in an embodiment of the present invention;

[0264] Figure 10 This is the second schematic diagram illustrating the structure of the wireless link failure prediction device provided in this embodiment of the invention.

[0265] Figure 11 This is a schematic diagram of the structure of the terminal provided in an embodiment of the present invention;

[0266] Figure 12This is a schematic diagram illustrating the structure of the network device provided in an embodiment of the present invention. Detailed Implementation

[0267] The technical solutions of this invention can be applied to various communication systems, such as GSM (Global System of Mobile communication), LTE (Long Term Evolution), 5G, and future systems (such as 6G). Optionally, a 5G system or 5G network can also be referred to as a New Radio (NR) system or NR network.

[0268] For example, the communication system used in this embodiment of the invention may include network devices and terminal devices (also referred to as terminals, communication terminals, etc.); the network device may be a device that communicates with the terminal device. The network device can provide communication coverage within a certain area and can communicate with terminals located within that area. Optionally, the network device may be a base station in various communication systems, such as an evolved Node B (eNB) in an LTE system, a gNB in ​​a 5G or NR system, or a base station in a 6G system.

[0269] AI functions and models are two different granularities. Generally, functions are coarser-grained, such as mobility management, RLF prediction, and beamforming, while models are finer-grained. In this case, one function can correspond to multiple models. Of course, if a model has good generalization ability, it can also be applied to multiple functions.

[0270] To make the technical problems, technical solutions, and advantages of this application clearer, a detailed description will be provided below in conjunction with the accompanying drawings and specific embodiments. In the following description, specific details such as particular configurations and components are provided merely to aid in a comprehensive understanding of the embodiments of this application. Therefore, those skilled in the art should understand that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Furthermore, for clarity and brevity, descriptions of known functions and structures have been omitted.

[0271] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0272] In the various embodiments of this application, it should be understood that the sequence number of each process described below does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0273] In the embodiments provided in this application, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0274] In this embodiment of the invention, the term "and / or" describes the relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The character " / " generally indicates that the preceding and following associated objects have an "or" relationship.

[0275] To address the lack of a specific process for transmitting and receiving RLF prediction information in existing technologies, and to enable the use of predicted RLF results to assist in network switching or configuration optimization, this invention provides a wireless link failure prediction method, apparatus, terminal, network device, and medium.

[0276] like Figure 1 As shown, this embodiment of the invention provides a wireless link failure prediction method, applied to a terminal, the method comprising:

[0277] Step 101: Send first information to the first network device, and / or send second information to the first network device;

[0278] The first information includes information related to Radio Link Failure (RLF).

[0279] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0280] It is understood that the RLF-related information is obtained based on the first AI model and / or the first AI function, and is predicted based on the AI ​​model and / or the first AI function.

[0281] It should be noted that, as Figure 2 As shown, the lifecycle management of AI models / AI functions includes processes such as data collection, model training, inference, management, and model storage. The specific processes are as follows: Figure 2As shown, the process includes sending training data, monitoring data, inference data, performance feedback / retraining requests, inference output, management instructions, trained / updated models, model transfer / delivery requests, and model transfer / delivery. In this invention, the information generated by inference can be simply understood as prediction information.

[0282] The RLF prediction-related information includes RLF prediction information, which indicates whether a wireless link failure and / or a wireless link failure prediction situation and / or a wireless link failure prediction result has been predicted.

[0283] Information related to the monitoring of the first artificial intelligence (AI) model and / or the first artificial intelligence (AI) function can be understood as information obtained by monitoring the AI ​​model and / or the AI ​​function.

[0284] This step clarifies the process by which the terminal sends RLF prediction-related information, AI model, and / or AI function monitoring-related information to the network device, so that the network device can use the predicted RLF results to assist the network in switching or configuration optimization.

[0285] It should be noted that the prediction method for RLF prediction based on AI models / AI functions includes direct prediction and indirect prediction. Direct prediction means that the AI ​​model / AI function outputs the RLF prediction result (including the probability of RLF occurrence, etc.). Indirect prediction means that the AI ​​model / AI function outputs the Radio Resource Management (RRM) prediction result, such as the predicted Signal to Interference plus Noise Ratio (SINR) value, Reference Signal Receiving Power (RSRP) value, and Reference Signal Receiving Quality (RSRQ) value. The UE derives whether an RLF will occur and the predicted time of the RLF based on the RRM prediction result, thus obtaining the RLF prediction result.

[0286] In an optional embodiment, before sending the first information to the first network device and / or sending the second information to the first network device, the method further includes:

[0287] Receive first configuration information sent by the first network device, wherein the first configuration information is configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction;

[0288] It is understandable that this first configuration information is the configuration sent by the network device to the terminal, including measurement configuration, reporting information, and other content.

[0289] The terminal performs RLF prediction and / or RRM prediction based on the first configuration information.

[0290] Optionally, the first configuration information includes at least one of the following:

[0291] Measurement configuration information, including traditional RRM configuration information and RLM configuration information, such as the measured signal and the parameters mentioned above, such as T310, N310, N311, Qin, Qout, etc.

[0292] The observation window for performing RLF prediction is, for example, 100ms;

[0293] The prediction window for performing RLF prediction, such as 100ms;

[0294] The prediction period for performing RLF predictions is, for example, 40ms (predicting once every 40ms, using the measurement results of the past 100ms to predict the results of the next 100ms);

[0295] The probability threshold for RLF occurrence, such as 50%, can be used by the terminal to determine whether the RLF will occur, and / or to calculate the key performance indicator (KPI) for monitoring AI models and / or AI functions. For example, if the terminal predicts that the probability of RLF occurrence is 80%, which is higher than the RLF occurrence probability threshold of 50%, it is considered that RLF will occur.

[0296] The first information reported, that is, the first information reported by the network device configuration terminal, includes what specific content;

[0297] The reporting conditions for the first information can be understood as the network configuring the reporting conditions for the first information. When the reporting conditions are met, the terminal reports the first information to the network device.

[0298] RRM prediction configuration information is used in the indirect prediction process, whereby the terminal obtains RRM prediction information based on AI functions or models, and then derives RLF prediction information. The RRM prediction configuration information may include at least one of the following: measurement configuration information, the observation window for RRM prediction, the prediction window for RRM prediction, the prediction period for RRM prediction, filtering configuration (e.g., layer 1 filtering parameters, layer 3 filtering parameters), measurement reduction rate (e.g., 50%, 30%), model output (e.g., layer 1 beam quality, layer 3 cell quality), and cell quality information (e.g., RSRP, RSRQ, SINR).

[0299] This first configuration information can be used to implicitly instruct the UE to perform RLF prediction based on the AI ​​model and / or AI function. The UE receives the first configuration information from the network device for inference configuration of the AI ​​model and / or AI function.

[0300] Among them, the observation window for RLF prediction, the prediction window for RLF prediction, and the prediction period for RLF prediction are prediction-related configurations.

[0301] It should be noted that the RLF-related parameters include Qin, Qout, T310, N310, N311, and T311. The radio link monitoring process is as follows: Figure 4As shown, if N310 consecutive out-of-sync indications (i.e., link asynchrony indications, or signal quality (CQI) below the threshold Qout) are received, the UE starts T310; if N311 consecutive in-sync indications (i.e., link synchronization indications, or signal quality above the threshold Qin) are received during the operation of T310, T310 stops; if N311 consecutive in-sync indications are not received, T310 times out and is considered to have occurred RLF, the UE performs RRC reconstruction procedure and starts T311.

[0302] The reporting conditions for the first information include at least one of the following:

[0303] The difference between the predicted time of RLF occurrence and the current time is less than the preset time difference, that is, the difference between the predicted RLF occurrence time and the current time is less than a threshold, such as 100ms (if the predicted time is too long, the accuracy may be lower).

[0304] The terminal receives a first preset number of consecutive link asynchrony indications, wherein the first preset number is the maximum count value of N310, that is, timer T310 starts (T310 starts, which means that the wireless link is not good and the probability of RLF is higher. The network can be configured that the UE will only report when it predicts RLF after T310 starts).

[0305] If the reporting period is met, for example, if the reporting period is 40ms, the report is submitted once every 40ms. If a prediction window (e.g., 100ms) is also configured, then the prediction result for the next 100ms is submitted every 40ms.

[0306] If an RLF is predicted to occur, it should be reported; if an RLF is not predicted, it does not need to be reported.

[0307] If the predicted probability of an RLF occurrence is higher than the RLF occurrence probability threshold, then the report is submitted; otherwise, no report is submitted.

[0308] The reporting time is configured, which may include a reporting start time and a reporting end time. That is, the terminal starts reporting the first information at the reporting start time and stops reporting the first information at the reporting end time.

[0309] In an optional embodiment, the method further includes:

[0310] The first information is obtained based on the first AI model and / or the first AI function.

[0311] In this optional embodiment, the terminal performs data collection, model training, inference, and management processes based on the first AI model and / or the first AI function to predict information related to RLF prediction.

[0312] The first information is obtained based on the first AI model and / or the first AI function, including:

[0313] Based on the first AI model and / or the first AI function, RLF prediction is performed according to the first configuration information to obtain the first information.

[0314] That is, the terminal, based on the first AI model and / or the first AI function, performs data collection, model training, inference, management and other processes according to the first configuration information to predict information related to RLF prediction.

[0315] In another alternative embodiment, the method further includes:

[0316] Based on the first AI model and / or the first AI function, RRM prediction information is obtained;

[0317] The first information is obtained based on the RRM prediction information.

[0318] In this optional embodiment, the terminal performs indirect prediction (RRM prediction) based on the first AI model and / or the first AI function to obtain RRM prediction information (such as SINR value), the terminal derives RLF prediction-related information based on the RRM prediction information, and the terminal sends the RLF prediction-related information and the RRM prediction information to the first network device.

[0319] In an optional embodiment, before sending the first information to the first network device and / or sending the second information to the first network device, the method further includes:

[0320] Send the first instruction information to the first network device;

[0321] Wherein, the first indication information is used to indicate at least one of the following:

[0322] The terminal is instructed to support RLF prediction based on AI models and / or AI functions;

[0323] The terminal is instructed to support RRM prediction based on AI models and / or AI functions;

[0324] Indicates the AI ​​models and / or AI functions supported by the terminal;

[0325] The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal;

[0326] Input information indicating the AI ​​models and / or AI functions supported by the terminal;

[0327] Output information indicating the AI ​​models and / or AI functions supported by the terminal.

[0328] Before the terminal receives the first configuration information sent by the first network device, the terminal reports its capabilities to the first network device. In one optional embodiment, an indication (i.e., first indication information) is added to the UE capability reporting information. This indication can indicate that the UE supports RLF prediction based on AI models and / or AI functions, or it can indicate whether the output information of the AI ​​models and / or AI functions supported by the UE is RLF prediction information (i.e., the UE supports direct prediction based on AI) or RRM prediction information (i.e., the UE supports indirect prediction based on AI). In another optional embodiment, the indication can also indicate the AI ​​models and / or AI functions supported by the UE, such as AI model identifiers (model ID or index) and / or AI function identifiers, as well as the accuracy information of the AI ​​models and / or AI functions supported by the UE (i.e., model accuracy and / or function accuracy), and the input and output information of the AI ​​models and / or AI functions supported by the UE (i.e., whether the AI ​​model and / or AI function is associated with direct prediction or indirect prediction). In this other optional embodiment, the indication can also be reported in the UE auxiliary information.

[0329] The following is combined Figure 3 This explains the process for reporting terminal capabilities:

[0330] The network device sends a UE capability reporting request to the terminal (UE), and the terminal reports its UE capabilities to the network device (carrying whether the terminal supports RLF prediction based on AI models and / or AI functions). The terminal can also report UE auxiliary information to the network device (carrying information such as the AI ​​models and / or AI functions supported by the UE).

[0331] The network device configures the terminal to activate / deactivate the UE's AI model and / or AI function. Specifically, in an optional embodiment, before sending first information to the first network device and / or sending second information to the first network device, the method further includes:

[0332] Receive the second indication information sent by the first network device;

[0333] Optionally, the second indication information is used to indicate at least one of the following:

[0334] The terminal is instructed to activate the AI ​​model and / or AI function, that is, to activate the RLF prediction function based on the direct prediction of the AI ​​model and / or AI function (the second indication information can be one indication) or the RRM prediction function based on the indirect prediction (the second indication information can be one indication or two indications).

[0335] The terminal is instructed to activate the RLF prediction function based on the AI ​​model and / or AI function, that is, to activate the RRM prediction function based on the direct prediction of the AI ​​model and / or AI function or the RRM prediction function based on the indirect prediction of the AI ​​model and / or AI function.

[0336] The AI ​​model and / or AI function used by the terminal for RLF prediction are indicated by the model identifier and / or function identifier, that is, the UE is instructed to use which AI model and / or AI function for RLF prediction by means of the model identifier and / or function identifier;

[0337] The terminal is instructed to access the first cell and perform RLF prediction. That is, after the terminal accesses the first cell where the first network device is located, it is instructed to perform RLF prediction based on the AI ​​model and / or AI function (the RLF prediction method includes direct prediction and indirect prediction).

[0338] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality (the RLF prediction includes direct RLF prediction and indirect RLF prediction);

[0339] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality (the RLF prediction includes direct RLF prediction and indirect RLF prediction);

[0340] The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication (the RLF prediction includes direct RLF prediction and indirect RLF prediction);

[0341] The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication (the RLF prediction includes direct RLF prediction and indirect RLF prediction);

[0342] AI models and / or AI functions are used for RLF prediction, i.e., AI models and / or AI functions are used for direct prediction;

[0343] AI models and / or AI functions are used for RRM prediction, i.e., AI models and / or AI functions are used for indirect prediction.

[0344] In one optional implementation, the network configures the UE to automatically activate / deactivate based on an AI model and / or AI function to perform RLF prediction via event triggering. Specifically, the network device configures the UE's triggering events to meet the conditions for automatic UE activation / deactivation based on the AI ​​model and / or AI function to perform RLF prediction. The triggering events include at least one of the following: based on cell signal quality, link asynchrony indication, T310 activation, KPI threshold, etc. For example, the signal quality of the serving cell is lower than threshold A (first quality threshold), the signal quality of the neighboring cell is higher than threshold B (second quality threshold), the signal quality of the neighboring cell is higher than the signal quality of the serving cell and the signal quality difference is greater than threshold C (third quality threshold), the KPI monitored by the AI ​​function or model is lower than the KPI threshold, the link asynchrony indication is higher than the threshold, etc.

[0345] Optionally, the second indication information includes at least one of the following:

[0346] Radio Resource Control (RRC) signaling;

[0347] Media Access Control Layer Control Element (MACCE);

[0348] Downlink Control Information (DCI);

[0349] System Information Block (SIB).

[0350] In one optional implementation, the network configures the terminal to activate the AI ​​model and / or AI function. For each UE, the terminal is configured to activate / deactivate RLF prediction based on the AI ​​model and / or AI function through explicit indication. Specifically, the UE is instructed to activate / deactivate the AI-based RLF prediction function (one indication) through RRC signaling, MAC CE, DCI, etc., or to activate the AI-based RLF indirect prediction (one or two indications), or to specifically indicate which model / function the UE uses (model identifier / function identifier).

[0351] In another optional implementation, the network configures the terminal to activate the AI ​​model and / or AI function. For each cell, the terminal is configured to activate / deactivate RLF prediction based on the AI ​​model and / or AI function by means of explicit indication. Specifically, the field carried in the SIB message indicates that the UE in the cell is allowed to perform AI-based RLF prediction. When the UE receives the SIB message containing the field or indicating permission, it activates / deactivates the AI-based RLF prediction.

[0352] Optionally, the first information includes at least one of the following:

[0353] The start status of the first timer, i.e. whether the first timer is started, wherein the first timer is started when the terminal receives a first preset number of consecutive out-of-sync indications, and the first timer is stopped when the terminal receives a second preset number of in-sync indications after it is started. The first preset number can be N310 (N310 can be a counter, and the first preset number is the maximum count value of the counter), the second preset number can be N311 (N311 can be a counter, and the second preset number is the maximum count value of the counter), and the first timer can be T310.

[0354] The runtime of the first timer, that is, the running time of T310 if T310 is started;

[0355] The remaining duration of the first timer, that is, the remaining time of T310 if T310 is started;

[0356] The number of link synchronization indications received when an RLF is predicted, i.e., the count value of N311 when an RLF is predicted if T310 is started;

[0357] The methods of forecasting include direct forecasting and indirect forecasting.

[0358] Predict whether RLF will occur or not, or predict whether RLF will occur within the prediction window;

[0359] If an RLF occurs, predict the time when the RLF will occur, including the specific time or time period (which can be a period of time less than or equal to the prediction window length).

[0360] Prediction window length: If the network device does not configure a prediction window for the UE, the UE can report the prediction window.

[0361] Predict the probability of RLF occurring within the window, such as 90%;

[0362] The predicted cell quality at the time of RLF occurrence can also be used to indicate whether the RLF is predicted directly or indirectly. If the RLF prediction information carries the predicted cell quality at the time of RLF occurrence, it is an indirect prediction; otherwise, it is a direct prediction.

[0363] It should be noted that the start status of the first timer, the runtime of the first timer, the remaining duration of the first timer, and the number of link synchronization indications received when an RLF is predicted are used by network devices to assess the likelihood of an actual RLF occurrence. If the T310 runtime is relatively long or the N311 count is relatively small, the likelihood of an actual RLF occurrence is relatively high.

[0364] In an optional embodiment, before the network device configures the terminal to perform AI model and / or AI function monitoring configuration information, i.e., before sending first information to the first network device, and / or sending second information to the first network device, the method further includes:

[0365] The system receives second configuration information sent by the first network device, wherein the second configuration information is configuration information related to the second information.

[0366] The AI ​​model and / or AI functions can be configured for monitoring using the second configuration information.

[0367] Optionally, the second configuration information includes at least one of the following:

[0368] The monitoring start time of the first AI model and / or the first AI function, i.e., the second configuration information explicitly indicates the monitoring start time T1;

[0369] The monitoring end time of the first AI model and / or the first AI function, i.e. the monitoring end time T2 is clearly indicated by the second configuration information;

[0370] The monitoring period of the first AI model and / or the first AI function, i.e., the monitoring period P explicitly indicated by the second configuration information;

[0371] The monitoring end indication information for the first AI model and / or the first AI function may not be an explicit indication, but rather a network release of monitoring configuration;

[0372] The reporting conditions for the second information are configured so that the second information is reported based on event triggers.

[0373] In one optional implementation, the second configuration information can explicitly indicate the monitoring period P, the monitoring start time T1, and the monitoring end time T2. That is, the UE continuously monitors during the time period {T1, T2} and reports the network monitoring results (i.e., reports the second information) every monitoring period P. In another optional implementation, only the monitoring period P can be configured, and monitoring can begin when the UE receives the configuration and continue until the network device indicates the end. In yet another optional implementation, the monitoring period P is configured, but monitoring is not started immediately; monitoring is started only after the network device gives an explicit indication.

[0374] The reporting conditions for the second information include at least one of the following:

[0375] The monitoring end time configured for the first network device must be met;

[0376] When the timer configured by the first network device stops or times out, that is, when the network device is configured with a timer, the second information is reported to the network device when the timer stops or times out.

[0377] The occurrence of RLF was predicted, and the RLF actually occurred;

[0378] The occurrence of RLF was predicted, but RLF did not actually occur;

[0379] The prediction was that the RLF would not occur, but the RLF actually did occur.

[0380] Among them, the following are triggering events: predicting that an RLF will occur and the RLF will actually occur, predicting that an RLF will occur and the RLF will not actually occur, and predicting that an RLF will not occur and the RLF will actually occur. That is, the occurrence of any one of these triggering events will trigger the reporting.

[0381] Optionally, the second information includes at least one of the following:

[0382] The actual time when RLF occurs;

[0383] The cell ID of the terminal;

[0384] After the first timer expires, the number of link synchronization indications received by the terminal is as follows: the first timer is started when the terminal receives a first preset number of consecutive link desynchronization indications, and the first timer is stopped when the terminal receives a second preset number of link synchronization indications after it is started. The first timer is T310 as described above. After the first timer expires, the number of link synchronization indications received by the terminal is the N311 count value after T310 expires.

[0385] Cell quality at the time when RLF actually occurs, where cell quality can be predicted and / or actually measured;

[0386] The cell quality in the RLF prediction information between the predicted time of RLF occurrence and the actual time of RLF occurrence is the cell quality during the period from the predicted RLF occurrence to the actual RLF occurrence. The cell quality can be predicted and / or actually measured.

[0387] Cell quality before the actual occurrence of RLF, i.e., cell quality for a period of time before the RLF occurs (which can be based on network configuration);

[0388] The startup status of the first timer, i.e., whether T310 is started;

[0389] The runtime of the first timer;

[0390] The indication information that the RLF did not actually occur means that if the RLF did not occur within the predicted time or time period, the UE will report an indication that the RLF did not occur or the prediction was wrong.

[0391] The quality of the cell where the terminal is located;

[0392] The first key performance indicator (KPI) information is the KPI information monitored by the first AI model and / or the first AI function. The first KPI information is calculated by the terminal and reported to the network based on the configuration of the network device.

[0393] It should also be noted that the second information reported by the UE is different when an RLF actually occurs and when an RLF does not actually occur. In an optional implementation, when an RLF actually occurs, the second information includes at least one of the following:

[0394] The actual time when RLF occurs;

[0395] The cell ID of the terminal;

[0396] After the first timer expires, the number of link synchronization indications received by the terminal;

[0397] Community quality when RLF actually occurs;

[0398] The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information;

[0399] Cell quality prior to the actual time of RLF occurrence.

[0400] In an alternative implementation, if the RLF does not occur after the prediction window has expired, i.e., if the RLF does not actually occur, the second information includes at least one of the following:

[0401] The quality of the cell where the terminal is located;

[0402] The start status of the first timer;

[0403] The runtime of the first timer;

[0404] RLF (Real-Frequency) indication information that did not actually occur;

[0405] The quality of the cell where the terminal is located.

[0406] The first KPI information includes at least one of the following:

[0407] The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality can be the average signal quality difference. Taking SINR as an example, the signal quality difference is (average) SINR difference.

[0408] The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence;

[0409] The time difference (early or late) between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0410] The time difference (early or late) between the end of the forecast window for RLF prediction and the actual time when the RLF occurs.

[0411] One of the time relationships between the prediction window of RLF prediction and the actual time of RLF occurrence (actual RLF occurrence) is as follows (Case A): Figure 5 As shown, the second case (Case B) illustrates the temporal relationship between the prediction window and the actual occurrence of the RLF. Figure 6 As shown, the third case (Case C) illustrates the temporal relationship between the prediction window and the actual occurrence of the RLF. Figure 7 As shown, Figures 5 to 7 As shown, if the occurrence of RLF is predicted, Case B can be disregarded when calculating the time difference, or t2 can be calculated as negative, because Case B's prediction can be considered accurate.

[0412] The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring is the difference between the predicted probability (the probability of an RLF occurring within the prediction window) and the actual probability (the actual probability of an RLF occurring). If the actual probability of an RLF occurring is 100%, and the probability of an RLF occurring within the prediction window (i.e., the predicted probability) is 70%, then the probability difference is 30%. If the actual probability of an RLF not occurring is 0%, and the probability of an RLF occurring within the prediction window (i.e., the predicted probability) is 70%, then the probability difference is 70%.

[0413] Evaluation metrics for the first AI model and / or the first AI function.

[0414] The evaluation metrics are those used in traditional machine learning algorithms, and the evaluation metrics include at least one of the following:

[0415] Precision, recall, and F1 score.

[0416] The calculation table for the evaluation indicators is shown in Table 1 below.

[0417] Table 1

[0418]

[0419] Among them, “Baseline event” represents the baseline event, “Predicted event” represents the predicted event, “Positive” represents a positive sample, “Negative” represents a negative sample, “TP” represents the number of true positives, “FN” represents the number of negative positives, “FP” represents the number of false positives, and “TN” represents the number of false negatives.

[0420] Precision = TP / (TP+FP);

[0421] Recall = TP / (TP+FN);

[0422] F1 score=2*Precision*Recall / (Precision+Recall).

[0423] The output is the predicted probability of RLF occurring, and the statistical method is as follows:

[0424] Method (1): Set a probability threshold (e.g., 70%). If the probability exceeds the threshold, it is considered that the predicted RLF has occurred (TP or FP count is incremented by 1). If the probability is below the threshold, it is considered that the predicted RLF has not occurred (FN or TN count is incremented by 1). For example, if the predicted probability of RLF is 80% and RLF actually occurs, the TP value is incremented by 1; if the predicted probability is 30% and RLF actually occurs, the FN value is incremented by 1.

[0425] Method (2): The count is different for different probabilities, and it is not a simple addition of 1, such as:

[0426] Set a threshold. If the predicted probability exceeds the threshold, it is considered that an RLF has occurred, but the count is increased by the probability value (TP or FP count is increased by 0.X). If the probability is below the threshold, it is considered that an RLF has not occurred, and the count is increased by the probability value (FN or TN count is increased by 0.X). For example, assuming the threshold is 50%, if an RLF actually occurs, the predicted probability is 70%, the TP value is increased by 0.7, and the predicted probability is 40%, the FN value is increased by 0.4. If an RLF does not actually occur, the predicted probability is 70%, the FP value is increased by 0.7, and the predicted probability is 40%, the TN value is increased by 0.4.

[0427] In an alternative embodiment, if an RLF actually occurs, the method further includes:

[0428] The second information sent to the second network device;

[0429] The second network device sends the second information to the first network device;

[0430] The second network device is the network device corresponding to the cell selected by the terminal for reconstruction after the RLF actually occurs.

[0431] In this optional embodiment, after an RLF actually occurs, if the cell selected by the UE for reconstruction is still the original cell, the recorded second information is reported to the source cell; if the UE selects another cell to perform reconstruction, the report is sent to the new cell (the second network device), and then the new cell forwards it to the source cell (the first network device), indicating that this is an AI-based RLF prediction monitoring report (i.e., the second information).

[0432] In one alternative embodiment, the terminal manages the AI ​​model and / or AI functions itself based on the configuration of the network device.

[0433] If the terminal calculates the KPI information itself, the method further includes:

[0434] The terminal receives a third indication message sent by the first network device. The third indication message is used to instruct the terminal to perform a first operation. This configuration instructs the terminal to perform the first operation.

[0435] The first operation includes at least one of the following:

[0436] Switch to the second AI model and / or the second AI function;

[0437] Activate the second AI model and / or the second AI function;

[0438] To activate the first AI model and / or the first AI function;

[0439] Revert to non-AI operation, i.e., do not perform RLF prediction through AI models and / or AI functions.

[0440] The second AI model is different from the first AI model. Specifically, which model can be selected autonomously by the terminal based on the model's performance or configured by the network device. The second AI function is different from the first AI function. Specifically, which function can be selected autonomously by the terminal based on the function's performance or configured by the network device.

[0441] In this optional embodiment, the terminal autonomously calculates KPI information and autonomously switches AI models and / or AI functions, or deactivates AI models and / or AI functions, or rolls back to non-AI operations, and needs to report its own decisions to the network device.

[0442] Furthermore, the method also includes:

[0443] The first operation is performed based on the third instruction information and the first KPI information.

[0444] That is, the terminal triggers the execution of the first operation based on the third instruction information, and decides whether to execute the first operation based on the first KPI information.

[0445] Optionally, the third indication information includes at least one of the following:

[0446] The terminal is instructed to access the first cell and perform the first operation, that is, after the terminal is instructed to access the first cell where the first network device is located, the first operation is triggered.

[0447] The terminal is instructed to perform the first operation based on the cell signal quality. For example, if the cell signal quality is lower than a certain preset signal quality, the first operation is triggered.

[0448] The terminal is instructed to perform the first operation according to the link asynchrony indication, such as T310 activation, that is, the UE receives N310 consecutive out-of-sync indications (i.e., asynchrony indications), the UE activates T310, and triggers the execution of the first operation;

[0449] The terminal is instructed to perform the first operation based on KPI information, for example, based on the relationship between the evaluation indicators of the first AI model and / or the first AI function and the corresponding thresholds.

[0450] KPI thresholds, for example, if the KPI threshold is set to 90%, and the evaluation index of the first AI model and / or the first AI function is 85%, which is lower than the KPI threshold, then the first AI function or AI model can be activated.

[0451] In an optional embodiment, the method further includes:

[0452] The terminal receives a fourth indication message sent by the first network device; the fourth indication message is sent by the terminal to the terminal based on the second KPI information, and the second KPI information is the KPI information obtained by the first network device for the terminal to perform RLF prediction.

[0453] In this optional embodiment, the second KPI information may be determined by the first network device, that is, the first network device obtains the second KPI information for the terminal to perform RLF prediction based on the RLF prediction information.

[0454] Furthermore, based on the aforementioned second KPI information, the first network device decides whether to switch the AI ​​model and / or AI function, or deactivate the AI ​​model and / or AI function, or fall back to non-AI operation, and instructs the UE accordingly. That is, the fourth instruction information is used to instruct at least one of the following:

[0455] Instruct the terminal to switch to the second AI model and / or the second AI function;

[0456] The terminal is instructed to activate the second AI model and / or the second AI function;

[0457] Instruct the terminal to activate the first AI model and / or the first AI function;

[0458] Instruct the terminal to revert to non-AI operation, and instruct the terminal not to perform RLF prediction through AI models and / or AI functions.

[0459] The second AI model is a different model from the first AI model. The specific model can be determined by the network device based on the model's performance. The second AI function is a different function from the first AI function. The specific function can be determined by the network device based on the function's performance.

[0460] like Figure 8 As shown, this embodiment of the invention also provides a wireless link failure prediction method, applied to a first network device, the method comprising:

[0461] Step 801: Receive the first information sent by the terminal, and / or receive the second information sent by the terminal;

[0462] The first information includes information related to radio link failure (RLF) prediction;

[0463] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0464] It is understood that the RLF-related information is obtained by the terminal based on the first AI model and / or the first AI function, and is predicted based on the AI ​​model and / or the first AI function.

[0465] The RLF prediction-related information includes RLF prediction information, which indicates whether a wireless link failure and / or a wireless link failure prediction situation and / or a wireless link failure prediction result has been predicted.

[0466] Information related to the monitoring of the first artificial intelligence (AI) model and / or the first artificial intelligence (AI) function can be understood as information obtained by monitoring the AI ​​model and / or the AI ​​function.

[0467] This step clarifies the process by which the terminal sends RLF prediction-related information, AI model, and / or AI function monitoring-related information to the network device, so that the network device can use the predicted RLF results to assist the network in switching or configuration optimization.

[0468] In an optional embodiment, before receiving the first information sent by the receiving terminal, and / or before receiving the second information sent by the receiving terminal, the method further includes:

[0469] Send first configuration information to the terminal, the first configuration information being configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction.

[0470] It is understandable that this first configuration information is the configuration sent by the network device to the terminal, including measurement configuration, reporting information, and other content.

[0471] The terminal performs RLF prediction and / or RRM prediction based on the first configuration information.

[0472] Optionally, the first configuration information includes at least one of the following:

[0473] Measurement configuration information, including traditional RRM configuration information and RLM configuration information, such as the measured signal and the parameters mentioned above, such as T310, N310, N311, Qin, Qout, etc.

[0474] The observation window for performing RLF prediction is, for example, 100ms;

[0475] The prediction window for performing RLF prediction, such as 100ms;

[0476] The prediction period for performing RLF predictions is, for example, 40ms (predicting once every 40ms, using the measurement results of the past 100ms to predict the results of the next 100ms);

[0477] The probability threshold for RLF occurrence, such as 50%, can be used by the terminal to determine whether the RLF will occur, and / or to calculate the key performance indicator (KPI) for monitoring AI models and / or AI functions. For example, if the terminal predicts that the probability of RLF occurrence is 80%, which is higher than the RLF occurrence probability threshold of 50%, it is considered that RLF will occur.

[0478] The first information reported, that is, the first information reported by the network device configuration terminal, includes what specific content;

[0479] The reporting conditions for the first information can be understood as the network configuring the reporting conditions for the first information. When the reporting conditions are met, the terminal reports the first information to the network device.

[0480] RRM prediction configuration information is used in the indirect prediction process, whereby the terminal obtains RRM prediction information based on AI functions or models, and then derives RLF prediction information. The RRM prediction configuration information may include at least one of the following: measurement configuration information, the observation window for RRM prediction, the prediction window for RRM prediction, the prediction period for RRM prediction, filtering configuration (e.g., layer 1 filtering parameters, layer 3 filtering parameters), measurement reduction rate (e.g., 50%, 30%), model output (e.g., layer 1 beam quality, layer 3 cell quality), and cell quality information (e.g., RSRP, RSRQ, SINR).

[0481] This first configuration information can be used to implicitly instruct the UE to perform RLF prediction based on the AI ​​model and / or AI function. The UE receives the first configuration information from the network device for inference configuration of the AI ​​model and / or AI function.

[0482] Among them, the observation window for RLF prediction, the prediction window for RLF prediction, and the prediction period for RLF prediction are prediction-related configurations.

[0483] It should be noted that the RLF-related parameters include Qin, Qout, T310, N310, N311, and T311. The radio link monitoring process is as follows: Figure 4As shown, if N310 consecutive out-of-sync indications (i.e., link asynchrony indications, or signal quality (CQI) below the threshold Qout) are received, the UE starts T310; if N311 consecutive in-sync indications (i.e., link synchronization indications, or signal quality above the threshold Qin) are received during the operation of T310, T310 stops; if N311 consecutive in-sync indications are not received, T310 times out and is considered to have occurred RLF, the UE performs RRC reconstruction procedure and starts T311.

[0484] The reporting conditions for the first information include at least one of the following:

[0485] The difference between the predicted time of RLF occurrence and the current time is less than the preset time difference, that is, the difference between the predicted RLF occurrence time and the current time is less than a threshold, such as 100ms (if the predicted time is too long, the accuracy may be lower).

[0486] The terminal receives a first preset number of consecutive link asynchrony indications, wherein the first preset number is the maximum count value of N310, that is, timer T310 starts (T310 starts, which means that the wireless link is not good and the probability of RLF is higher. The network can be configured that the UE will only report when it predicts RLF after T310 starts).

[0487] If the reporting period is met, for example, if the reporting period is 40ms, the report is submitted once every 40ms. If a prediction window (e.g., 100ms) is also configured, then the prediction result for the next 100ms is submitted every 40ms.

[0488] If an RLF is predicted to occur, it should be reported; if an RLF is not predicted, it does not need to be reported.

[0489] If the predicted probability of an RLF occurrence is higher than the RLF occurrence probability threshold, then the report is submitted; otherwise, no report is submitted.

[0490] The reporting time is configured, which may include a reporting start time and a reporting end time. That is, the terminal starts reporting the first information at the reporting start time and stops reporting the first information at the reporting end time.

[0491] It should be noted that the prediction method for RLF prediction based on AI models / AI functions includes direct prediction and indirect prediction. Direct prediction means that the AI ​​model / AI function outputs the RLF prediction result (which indicates whether a radio link failure or a predicted situation of radio link failure has been predicted). Indirect prediction means that the AI ​​model / AI function outputs the Radio Resource Management (RRM) prediction result, such as the predicted Signal to Interference plus Noise Ratio (SINR) value. The UE derives whether an RLF will occur based on this RRM prediction result (SINR value), thus obtaining the RLF prediction result.

[0492] In an optional embodiment, the method further includes:

[0493] Receive RRM prediction information sent by the terminal.

[0494] In this optional embodiment, the terminal performs indirect prediction (RRM prediction) based on the first AI model and / or the first AI function to obtain RRM prediction information (such as SINR value), the terminal derives RLF prediction-related information based on the RRM prediction information, and the terminal sends the RLF prediction-related information and the RRM prediction information to the first network device.

[0495] In an optional embodiment, before receiving the first information sent by the receiving terminal, and / or before receiving the second information sent by the receiving terminal, the method further includes:

[0496] Receive the first indication information sent by the terminal;

[0497] Wherein, the first indication information is used to indicate at least one of the following:

[0498] The terminal is instructed to support RLF prediction based on AI models and / or AI functions;

[0499] The terminal is instructed to support RRM prediction based on AI models and / or AI functions;

[0500] Indicates the AI ​​models and / or AI functions supported by the terminal;

[0501] The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal;

[0502] Input information indicating the AI ​​models and / or AI functions supported by the terminal;

[0503] Output information indicating the AI ​​models and / or AI functions supported by the terminal.

[0504] Before the terminal receives the first configuration information sent by the first network device, the terminal reports its capabilities to the first network device. In one optional embodiment, an indication (i.e., first indication information) is added to the UE capability reporting information. This indication can indicate that the UE supports RLF prediction based on AI models and / or AI functions, or it can indicate whether the output information of the AI ​​models and / or AI functions supported by the UE is RLF prediction information (i.e., the UE supports direct prediction based on AI) or RRM prediction information (i.e., the UE supports indirect prediction based on AI). In another optional embodiment, the indication can also indicate the AI ​​models and / or AI functions supported by the UE, such as AI model identifiers (model ID or index) and / or AI function identifiers, as well as the accuracy information of the AI ​​models and / or AI functions supported by the UE (i.e., model accuracy and / or function accuracy), and the input and output information of the AI ​​models and / or AI functions supported by the UE (i.e., whether the AI ​​model and / or AI function is associated with direct prediction or indirect prediction). In this other optional embodiment, the indication can also be reported in the UE auxiliary information.

[0505] In an optional embodiment, before receiving the first information sent by the receiving terminal, and / or before receiving the second information sent by the receiving terminal, the method further includes:

[0506] Send a second instruction message to the terminal;

[0507] The second indication information is used to indicate at least one of the following:

[0508] The terminal is instructed to activate the AI ​​model and / or AI function, that is, to activate the RLF prediction function based on the direct prediction of the AI ​​model and / or AI function (the second indication information can be one indication) or the RRM prediction function based on the indirect prediction (the second indication information can be one indication or two indications).

[0509] The terminal is instructed to activate the RLF prediction function based on the AI ​​model and / or AI function, that is, to activate the RRM prediction function based on the direct prediction of the AI ​​model and / or AI function or the RRM prediction function based on the indirect prediction of the AI ​​model and / or AI function.

[0510] The AI ​​model and / or AI function used by the terminal for RLF prediction are indicated by the model identifier and / or function identifier, that is, the UE is instructed to use which AI model and / or AI function for RLF prediction by means of the model identifier and / or function identifier;

[0511] The terminal is instructed to access the first cell and perform RLF prediction. That is, after the terminal accesses the first cell where the first network device is located, it is instructed to perform RLF prediction based on the AI ​​model and / or AI function (the RLF prediction method includes direct prediction and indirect prediction).

[0512] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality (the RLF prediction includes direct RLF prediction and indirect RLF prediction);

[0513] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality (the RLF prediction includes direct RLF prediction and indirect RLF prediction);

[0514] The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication (the RLF prediction includes direct RLF prediction and indirect RLF prediction);

[0515] The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication (the RLF prediction includes direct RLF prediction and indirect RLF prediction);

[0516] AI models and / or AI functions are used for RLF prediction, i.e., AI models and / or AI functions are used for direct prediction;

[0517] AI models and / or AI functions are used for RRM prediction, i.e., AI models and / or AI functions are used for indirect prediction.

[0518] In one optional implementation, the network configures the UE to automatically activate / deactivate based on an AI model and / or AI function to perform RLF prediction via event triggering. Specifically, the network device configures the UE's triggering events to meet the conditions for automatic UE activation / deactivation based on the AI ​​model and / or AI function to perform RLF prediction. The triggering events include at least one of the following: based on cell signal quality, link asynchrony indication, T310 activation, KPI threshold, etc. For example, the signal quality of the serving cell is lower than threshold A (first quality threshold), the signal quality of the neighboring cell is higher than threshold B (second quality threshold), the signal quality of the neighboring cell is higher than the signal quality of the serving cell and the signal quality difference is greater than threshold C (third quality threshold), the KPI monitored by the AI ​​function or model is lower than the KPI threshold, the link asynchrony indication is higher than the threshold, etc.

[0519] Optionally, the second indication information includes at least one of the following:

[0520] RRC signaling;

[0521] MAC CE;

[0522] DCI;

[0523] SIB.

[0524] In one optional implementation, the network configures the terminal to activate the AI ​​model and / or AI function. For each UE, the terminal is configured to activate / deactivate RLF prediction based on the AI ​​model and / or AI function through explicit indication. Specifically, the UE is instructed to activate / deactivate the AI-based RLF prediction function (one indication) through RRC signaling, MAC CE, DCI, etc., or to activate the AI-based RLF indirect prediction (one or two indications), or to specifically indicate which model / function the UE uses (model identifier / function identifier).

[0525] In another optional implementation, the network configures the terminal to activate the AI ​​model and / or AI function. For each cell, the terminal is configured to activate / deactivate RLF prediction based on the AI ​​model and / or AI function by means of explicit indication. Specifically, the field carried in the SIB message indicates that the UE in the cell is allowed to perform AI-based RLF prediction. When the UE receives the SIB message containing the field or indicating permission, it activates / deactivates the AI-based RLF prediction.

[0526] Optionally, the first information includes at least one of the following:

[0527] The start status of the first timer, i.e. whether the first timer is started, wherein the first timer is started when the terminal receives a first preset number of consecutive out-of-sync indications, and the first timer is stopped when the terminal receives a second preset number of in-sync indications after it is started. The first preset number can be N310 (N310 can be a counter, and the first preset number is the maximum count value of the counter), the second preset number can be N311 (N311 can be a counter, and the second preset number is the maximum count value of the counter), and the first timer can be T310.

[0528] The runtime of the first timer, that is, the running time of T310 if T310 is started;

[0529] The remaining duration of the first timer, that is, the remaining time of T310 if T310 is started;

[0530] The number of link synchronization indications received when an RLF is predicted, i.e., the count value of N311 when an RLF is predicted if T310 is started;

[0531] The methods of forecasting include direct forecasting and indirect forecasting.

[0532] Predict whether RLF will occur or not, or predict whether RLF will occur within the prediction window;

[0533] If an RLF occurs, predict the time when the RLF will occur, including the specific time or time period (which can be a period of time less than or equal to the prediction window length).

[0534] Prediction window length: If the network device does not configure a prediction window for the UE, the UE can report the prediction window.

[0535] Predict the probability of RLF occurring within the window, such as 90%;

[0536] The predicted cell quality at the time of RLF occurrence can also be used to indicate whether the RLF is predicted directly or indirectly. If the RLF prediction information carries the predicted cell quality at the time of RLF occurrence, it is an indirect prediction; otherwise, it is a direct prediction.

[0537] It should be noted that the start status of the first timer, the runtime of the first timer, the remaining duration of the first timer, and the number of link synchronization indications received when an RLF is predicted are used by network devices to assess the likelihood of an actual RLF occurrence. If the T310 runtime is relatively long or the N311 count is relatively small, the likelihood of an actual RLF occurrence is relatively high.

[0538] In an optional embodiment, before receiving the first information sent by the receiving terminal, and / or before receiving the second information sent by the receiving terminal, the method further includes:

[0539] Send second configuration information to the terminal, the second configuration information being configuration information related to the second information.

[0540] The AI ​​model and / or AI functions can be configured for monitoring using the second configuration information.

[0541] The second configuration information includes at least one of the following:

[0542] The monitoring start time of the first AI model and / or the first AI function, i.e., the second configuration information explicitly indicates the monitoring start time T1;

[0543] The monitoring end time of the first AI model and / or the first AI function, i.e. the monitoring end time T2 is clearly indicated by the second configuration information;

[0544] The monitoring period of the first AI model and / or the first AI function, i.e., the monitoring period P explicitly indicated by the second configuration information;

[0545] The monitoring end indication information for the first AI model and / or the first AI function may not be an explicit indication, but rather a network release of monitoring configuration;

[0546] The reporting conditions for the second information are configured so that the second information is reported based on event triggers.

[0547] In one optional implementation, the second configuration information can explicitly indicate the monitoring period P, the monitoring start time T1, and the monitoring end time T2. That is, the UE continuously monitors during the time period {T1, T2} and reports the network monitoring results (i.e., reports the second information) every monitoring period P. In another optional implementation, only the monitoring period P can be configured, and monitoring can begin when the UE receives the configuration and continue until the network device indicates the end. In yet another optional implementation, the monitoring period P is configured, but monitoring is not started immediately; monitoring is started only after the network device gives an explicit indication.

[0548] The reporting conditions for the second information include at least one of the following:

[0549] The monitoring end time configured for the first network device must be met;

[0550] When the timer configured by the first network device stops or times out, that is, when the network device is configured with a timer, the second information is reported to the network device when the timer stops or times out.

[0551] The occurrence of RLF was predicted, and the RLF actually occurred;

[0552] The occurrence of RLF was predicted, but RLF did not actually occur;

[0553] The prediction was that the RLF would not occur, but the RLF actually did occur.

[0554] Among them, the following are triggering events: predicting that an RLF will occur and the RLF will actually occur, predicting that an RLF will occur and the RLF will not actually occur, and predicting that an RLF will not occur and the RLF will actually occur. That is, the occurrence of any one of these triggering events will trigger the reporting.

[0555] Optionally, the second information includes at least one of the following:

[0556] The actual time when RLF occurs;

[0557] The cell ID of the terminal;

[0558] After the first timer expires, the number of link synchronization indications received by the terminal is as follows: the first timer is started when the terminal receives a first preset number of consecutive link desynchronization indications, and the first timer is stopped when the terminal receives a second preset number of link synchronization indications after it is started. The first timer is T310 as described above. After the first timer expires, the number of link synchronization indications received by the terminal is the N311 count value after T310 expires.

[0559] Cell quality at the time when RLF actually occurs, where cell quality can be predicted and / or actually measured;

[0560] The cell quality in the RLF prediction information between the predicted time of RLF occurrence and the actual time of RLF occurrence is the cell quality during the period from the predicted RLF occurrence to the actual RLF occurrence. The cell quality can be predicted and / or actually measured.

[0561] Cell quality before the actual occurrence of RLF, i.e., cell quality for a period of time before the RLF occurs (which can be based on network configuration);

[0562] The startup status of the first timer, i.e., whether T310 is started;

[0563] The runtime of the first timer;

[0564] The indication information that the RLF did not actually occur means that if the RLF did not occur within the predicted time or time period, the UE will report an indication that the RLF did not occur or the prediction was wrong.

[0565] The quality of the cell where the terminal is located;

[0566] The first key performance indicator (KPI) information is the KPI information monitored by the first AI model and / or the first AI function. The first KPI information is calculated by the terminal and reported to the network based on the configuration of the network device.

[0567] It should also be noted that the second information reported by the UE is different when an RLF actually occurs and when an RLF does not actually occur. In an optional implementation, when an RLF actually occurs, the second information includes at least one of the following:

[0568] The actual time when RLF occurs;

[0569] The cell ID of the terminal;

[0570] After the first timer expires, the number of link synchronization indications received by the terminal;

[0571] Community quality when RLF actually occurs;

[0572] The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information;

[0573] Cell quality prior to the actual time of RLF occurrence.

[0574] In an alternative implementation, if the RLF does not occur after the prediction window has expired, i.e., if the RLF does not actually occur, the second information includes at least one of the following:

[0575] The quality of the cell where the terminal is located;

[0576] The start status of the first timer;

[0577] The runtime of the first timer;

[0578] RLF (Real-Frequency) indication information that did not actually occur;

[0579] The quality of the cell where the terminal is located.

[0580] The first KPI information includes at least one of the following:

[0581] The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality can be the average signal quality difference. Taking SINR as an example, the signal quality difference is (average) SINR difference.

[0582] The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence;

[0583] The time difference (early or late) between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0584] The time difference (early or late) between the end of the forecast window for RLF prediction and the actual time when the RLF occurs.

[0585] The first KPI information includes at least one of the following:

[0586] The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality can be the average signal quality difference. Taking SINR as an example, the signal quality difference is (average) SINR difference.

[0587] The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence;

[0588] The time difference (early or late) between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0589] The time difference (early or late) between the end of the forecast window for RLF prediction and the actual time when the RLF occurs.

[0590] One of the time relationships between the prediction window of RLF prediction and the actual time of RLF occurrence (actual RLF occurrence) is as follows (Case A): Figure 5 As shown, the second case (Case B) illustrates the temporal relationship between the prediction window and the actual occurrence of the RLF. Figure 6 As shown, the third case (Case C) illustrates the temporal relationship between the prediction window and the actual occurrence of the RLF. Figure 7 As shown, Figures 5 to 7 As shown, if the occurrence of RLF is predicted, Case B can be disregarded when calculating the time difference, or t2 can be calculated as negative, because Case B's prediction can be considered accurate.

[0591] The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring is the difference between the predicted probability (the probability of an RLF occurring within the prediction window) and the actual probability (the actual probability of an RLF occurring). If the actual probability of an RLF occurring is 100% and the probability of an RLF occurring within the prediction window (i.e., the predicted probability) is 70%, then the probability difference is 30%. If the actual probability of an RLF not occurring is 0% and the probability of an RLF occurring within the prediction window (i.e., the predicted probability) is 70%, then the probability difference is 70%.

[0592] Evaluation metrics for the first AI model and / or the first AI function.

[0593] The evaluation metrics are those used in traditional machine learning algorithms, and the evaluation metrics include at least one of the following:

[0594] Precision, recall, and F1 score.

[0595] The calculation table for the evaluation indicators is shown in Table 1 below.

[0596] Table 1

[0597]

[0598] Among them, “Baseline event” represents the baseline event, “Predicted event” represents the predicted event, “Positive” represents a positive sample, “Negative” represents a negative sample, “TP” represents the number of true positives, “FN” represents the number of negative positives, “FP” represents the number of false positives, and “TN” represents the number of false negatives.

[0599] Precision = TP / (TP+FP);

[0600] Recall = TP / (TP+FN);

[0601] F1 score=2*Precision*Recall / (Precision+Recall).

[0602] The output is the predicted probability of RLF occurring, and the statistical method is as follows:

[0603] Method (1): Set a probability threshold (e.g., 70%). If the probability exceeds the threshold, it is considered that the predicted RLF has occurred (TP or FP count is incremented by 1). If the probability is below the threshold, it is considered that the predicted RLF has not occurred (FN or TN count is incremented by 1). For example, if the predicted probability of RLF is 80% and RLF actually occurs, the TP value is incremented by 1; if the predicted probability is 30% and RLF actually occurs, the FN value is incremented by 1.

[0604] Method (2): The count is different for different probabilities, and it is not a simple addition of 1, such as:

[0605] Set a threshold. If the predicted probability exceeds the threshold, it is considered that an RLF has occurred, but the count is increased by the probability value (TP or FP count is increased by 0.X). If the probability is below the threshold, it is considered that an RLF has not occurred, and the count is increased by the probability value (FN or TN count is increased by 0.X). For example, assuming the threshold is 50%, if an RLF actually occurs, the predicted probability is 70%, the TP value is increased by 0.7, and the predicted probability is 40%, the FN value is increased by 0.4. If an RLF does not actually occur, the predicted probability is 70%, the FP value is increased by 0.7, and the predicted probability is 40%, the TN value is increased by 0.4.

[0606] In an alternative embodiment, if an RLF actually occurs, the method further includes:

[0607] Receive the second information sent by the second network device;

[0608] The first information is sent by the terminal to the second network device;

[0609] The second network device is the network device corresponding to the cell selected by the terminal after the RLF actually occurs.

[0610] In this optional embodiment, after an RLF actually occurs, if the cell selected by the UE for reconstruction is still the original cell, the recorded second information is reported to the source cell; if the UE selects another cell to perform reconstruction, the report is sent to the new cell (the second network device), and then the new cell forwards it to the source cell (the first network device), indicating that this is an AI-based RLF prediction monitoring report (i.e., the second information).

[0611] In an optional embodiment, the terminal can manage AI models and / or AI functions according to network instructions or based on network configuration. The method further includes:

[0612] Send a third instruction message to the terminal;

[0613] The third indication information is used to instruct the terminal to perform a first operation, and this configuration instructs the terminal to perform the first operation.

[0614] The first operation includes at least one of the following:

[0615] Switch to the second AI model and / or the second AI function;

[0616] Activate the second AI model and / or the second AI function;

[0617] To activate the first AI model and / or the first AI function;

[0618] Revert to non-AI operation.

[0619] The second AI model is different from the first AI model. Specifically, which model can be selected autonomously by the terminal based on the model's performance or configured by the network device. The second AI function is different from the first AI function. Specifically, which function can be selected autonomously by the terminal based on the function's performance or configured by the network device.

[0620] The third indication information is used for at least one of the following:

[0621] The terminal is instructed to access the first cell and perform the first operation, that is, after the terminal is instructed to access the first cell where the first network device is located, the first operation is triggered.

[0622] The terminal is instructed to perform the first operation based on the cell signal quality. For example, if the cell signal quality is lower than a certain preset signal quality, the first operation is triggered.

[0623] The terminal is instructed to perform the first operation according to the link asynchrony indication, such as T310 activation, that is, the UE receives N310 consecutive out-of-sync indications (i.e., asynchrony indications), the UE activates T310, and triggers the execution of the first operation;

[0624] The terminal is instructed to perform the first operation based on KPI information. For example, the first operation can be performed based on the relationship between the evaluation indicators of the first AI model and / or the first AI function and the corresponding thresholds.

[0625] KPI thresholds, for example, if the KPI threshold is set to 90%, and the evaluation index of the first AI model and / or the first AI function is 85%, which is lower than the KPI threshold, then the first AI function or AI model can be activated.

[0626] Furthermore, the method also includes:

[0627] The network device sends a fourth indication message to the terminal, indicating whether to switch the AI ​​model and / or AI function, or to deactivate the AI ​​model and / or AI function, or to fall back to non-AI operation, and instructs the UE accordingly. Specifically, the fourth indication message indicates at least one of the following:

[0628] Instruct the terminal to switch to the second AI model and / or the second AI function;

[0629] The terminal is instructed to activate the second AI model and / or the second AI function;

[0630] Instruct the terminal to activate the first AI model and / or the first AI function;

[0631] Instruct the terminal to revert to non-AI operation, and instruct the terminal not to perform RLF prediction through AI models and / or AI functions.

[0632] like Figure 9 As shown, this embodiment of the invention also provides a wireless link failure prediction device, the device comprising:

[0633] The first sending module 901 is used to send first information to the first network device, and / or send second information to the first network device;

[0634] The first information includes information related to radio link failure (RLF) prediction;

[0635] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0636] Optionally, the device further includes:

[0637] The first information receiving module is configured to receive first configuration information sent by the first network device, wherein the first configuration information is configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction.

[0638] Optionally, the device further includes:

[0639] The first processing module is used to obtain the first information based on the first AI model and / or the first AI function.

[0640] Optionally, the first configuration information includes at least one of the following:

[0641] Measurement configuration information;

[0642] Observation window for RLF prediction;

[0643] The forecast window for RLF forecasting;

[0644] The forecast period for RLF forecasts;

[0645] RLF occurrence probability threshold;

[0646] The first piece of information to be reported;

[0647] The reporting conditions for the first information;

[0648] RRM predicts configuration information.

[0649] Optionally, the reporting conditions for the first information include at least one of the following:

[0650] The difference between the predicted time of RLF occurrence and the current time is less than the preset time difference;

[0651] The terminal receives a first preset number of consecutive link asynchrony indications;

[0652] Meet the reporting cycle;

[0653] RLF was predicted to occur;

[0654] The predicted probability of RLF occurrence is higher than the RLF occurrence probability threshold;

[0655] The reporting time meets the configuration.

[0656] Optionally, the device further includes:

[0657] The second processing module is used to obtain RRM prediction information based on the first AI model and / or the first AI function;

[0658] The third processing module is used to obtain the first information based on the RRM prediction information.

[0659] Optionally, the device further includes:

[0660] The first information sending module is used to send first indication information to the first network device;

[0661] Wherein, the first indication information is used to indicate at least one of the following:

[0662] The terminal is instructed to support RLF prediction based on AI models and / or AI functions;

[0663] The terminal is instructed to support RRM prediction based on AI models and / or AI functions;

[0664] Indicates the AI ​​models and / or AI functions supported by the terminal;

[0665] The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal;

[0666] Input information indicating the AI ​​models and / or AI functions supported by the terminal;

[0667] Output information indicating the AI ​​models and / or AI functions supported by the terminal.

[0668] Optionally, the device further includes:

[0669] The second information receiving module is configured to receive second indication information sent by the first network device; wherein the second indication information is configured to indicate at least one of the following:

[0670] The terminal is instructed to activate the AI ​​model and / or AI function;

[0671] Instruct the terminal to activate the AI ​​model and / or AI function;

[0672] Instruct the terminal to use the AI ​​model and / or AI function for RLF prediction;

[0673] The terminal is instructed to access the first cell and perform RLF prediction;

[0674] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality;

[0675] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality;

[0676] The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication;

[0677] The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication; the AI ​​model and / or AI function are used to perform RLF prediction;

[0678] AI models and / or AI functions are used for RRM prediction.

[0679] Optionally, the second indication information includes at least one of the following:

[0680] Radio Resource Control (RRC) signaling;

[0681] Media intervention control layer control cell MAC CE;

[0682] Downlink Control Information (DCI);

[0683] System Information Block (SIB).

[0684] Optionally, the first information includes at least one of the following:

[0685] The start status of the first timer;

[0686] The runtime of the first timer;

[0687] Remaining duration of the first timer;

[0688] The number of link synchronization indications received when an RLF is predicted;

[0689] Methods of prediction;

[0690] Predicting whether RLF will occur or not;

[0691] Predict the timing of RLF occurrence;

[0692] Prediction window length;

[0693] Predict the probability of RLF occurring within the window;

[0694] Predict the cell quality when RLF occurs.

[0695] Optionally, the device further includes:

[0696] The third information receiving module is used to receive the second configuration information sent by the first network device, wherein the second configuration information is configuration information related to the second information.

[0697] Optionally, the second configuration information includes at least one of the following:

[0698] The start time of monitoring the first AI model and / or the first AI function;

[0699] The end time of monitoring the first AI model and / or the first AI function;

[0700] Monitoring cycle of the first AI model and / or the first AI function;

[0701] Monitoring termination indication information for the first AI model and / or the first AI function;

[0702] The reporting conditions for the second information.

[0703] Optionally, the reporting conditions for the second information include at least one of the following:

[0704] The monitoring end time configured for the first network device must be met;

[0705] The timer configured on the first network device stops or times out;

[0706] The occurrence of RLF was predicted, and the RLF actually occurred;

[0707] The occurrence of RLF was predicted, but RLF did not actually occur;

[0708] The prediction was that the RLF would not occur, but the RLF actually did occur.

[0709] Optionally, the second information includes at least one of the following:

[0710] The actual time when RLF occurs;

[0711] The cell identifier where the terminal is located;

[0712] After the first timer expires, the number of link synchronization indications received by the terminal;

[0713] Community quality when RLF actually occurs;

[0714] The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information;

[0715] Cell quality prior to the actual time of RLF occurrence;

[0716] The start status of the first timer;

[0717] The runtime of the first timer;

[0718] RLF (Real-Frequency) indication information that did not actually occur;

[0719] The quality of the cell where the terminal is located;

[0720] First Key Performance Indicator (KPI) information.

[0721] Optionally, the first KPI information includes at least one of the following:

[0722] The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality;

[0723] The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence;

[0724] The time difference between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0725] The time difference between the end of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0726] The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring;

[0727] Evaluation metrics for the first AI model and / or the first AI function.

[0728] Optionally, the device further includes:

[0729] The second information sending module is used to send the second information to the second network device;

[0730] The second network device sends the second information to the first network device;

[0731] The second network device is the network device corresponding to the cell selected by the terminal for reconstruction after the RLF actually occurs.

[0732] Optionally, the device further includes:

[0733] The fourth information receiving module is used to receive third indication information sent by the first network device, the third indication information being used to instruct the terminal to perform the first operation;

[0734] The first operation includes at least one of the following:

[0735] Switch to the second AI model and / or the second AI function;

[0736] Activate the second AI model and / or the second AI function;

[0737] To activate the first AI model and / or the first AI function;

[0738] Revert to non-AI operation.

[0739] Optionally, the device further includes:

[0740] The fourth processing module is used to execute the first operation based on the third instruction information and the first KPI information.

[0741] Optionally, the third indication information is used for at least one of the following:

[0742] The terminal is instructed to access the first cell and execute the first operation;

[0743] The terminal is instructed to perform the first operation based on the cell signal quality;

[0744] The terminal is instructed to perform the first operation according to the link asynchrony indication;

[0745] The terminal is instructed to perform the first operation based on the KPI information;

[0746] KPI thresholds.

[0747] Optionally, the device further includes:

[0748] The fifth information receiving module is used to receive the fourth indication information sent by the first network device;

[0749] The fourth indication information is used to indicate at least one of the following:

[0750] Instruct the terminal to switch to the second AI model and / or the second AI function;

[0751] The terminal is instructed to activate the second AI model and / or the second AI function;

[0752] Instruct the terminal to activate the first AI model and / or the first AI function;

[0753] Instruct the terminal to revert to non-AI operation.

[0754] It should be noted that the wireless link failure prediction device provided in the embodiments of the present invention is a device capable of executing the above-described wireless link failure prediction method applied to a terminal. Therefore, all embodiments of the above-described wireless link failure prediction method applied to a terminal are applicable to this device and can achieve the same or similar technical effects.

[0755] like Figure 10 As shown in the figure, this embodiment of the invention also provides a wireless link failure prediction device, applied to a terminal, the device comprising:

[0756] The first receiving module 1001 is used to receive first information sent by the terminal, and / or to receive second information sent by the terminal;

[0757] The first information includes information related to radio link failure (RLF) prediction;

[0758] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0759] Optionally, the device further includes:

[0760] The third information sending module is used to send first configuration information to the terminal, wherein the first configuration information is configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction.

[0761] Optionally, the first configuration information includes at least one of the following:

[0762] Measurement configuration information;

[0763] Observation window for RLF prediction;

[0764] The forecast window for RLF forecasting;

[0765] The forecast period for RLF forecasts;

[0766] RLF occurrence probability threshold;

[0767] The first piece of information to be reported;

[0768] The reporting conditions for the first information;

[0769] RRM predicts configuration information.

[0770] Optionally, the reporting conditions for the first information include at least one of the following:

[0771] The difference between the predicted time of RLF occurrence and the current time is less than a preset time difference; the terminal receives a first preset number of consecutive link asynchrony indications;

[0772] Meet the reporting cycle;

[0773] RLF was predicted to occur;

[0774] The predicted probability of RLF occurrence is higher than the RLF occurrence probability threshold;

[0775] The reporting time meets the configuration.

[0776] Optionally, the device further includes:

[0777] The sixth information receiving module is configured to receive first indication information sent by the terminal; wherein the first indication information is configured to indicate at least one of the following:

[0778] The terminal is instructed to support RLF prediction based on AI models and / or AI functions;

[0779] The terminal is instructed to support RRM prediction based on AI models and / or AI functions;

[0780] Indicates the AI ​​models and / or AI functions supported by the terminal;

[0781] The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal;

[0782] Input information indicating the AI ​​models and / or AI functions supported by the terminal;

[0783] Output information indicating the AI ​​models and / or AI functions supported by the terminal.

[0784] Optionally, the device further includes:

[0785] The fourth information sending module is used to send second indication information to the terminal;

[0786] The second indication information is used to indicate at least one of the following:

[0787] The terminal is instructed to activate the AI ​​model and / or AI function;

[0788] Instruct the terminal to activate the AI ​​model and / or AI function;

[0789] Instruct the terminal to use the AI ​​model and / or AI function for RLF prediction;

[0790] The terminal is instructed to access the first cell and perform RLF prediction;

[0791] The terminal is instructed to activate the AI ​​model and / or AI function based on cell signal quality; the terminal is instructed to deactivate the AI ​​model and / or AI function based on cell signal quality; the terminal is instructed to activate the AI ​​model and / or AI function based on link asynchrony.

[0792] The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication;

[0793] AI models and / or AI functions are used for RLF prediction;

[0794] AI models and / or AI functions are used for RRM prediction.

[0795] Optionally, the second indication information includes at least one of the following:

[0796] RRC signaling;

[0797] MAC CE;

[0798] DCI;

[0799] SIB.

[0800] Optionally, the first information includes at least one of the following:

[0801] The start status of the first timer;

[0802] The runtime of the first timer;

[0803] Remaining duration of the first timer;

[0804] The number of link synchronization indications received when an RLF is predicted;

[0805] Methods of prediction;

[0806] Predicting whether RLF will occur or not;

[0807] Predict the timing of RLF occurrence;

[0808] Prediction window length;

[0809] Predict the probability of RLF occurring within the window;

[0810] Predict the cell quality when RLF occurs.

[0811] Optionally, the device further includes:

[0812] The fifth information sending module is used to send second configuration information to the terminal, wherein the second configuration information is configuration information related to the second information.

[0813] Optionally, the second configuration information includes at least one of the following:

[0814] The start time of monitoring the first AI model and / or the first AI function;

[0815] The end time of monitoring the first AI model and / or the first AI function;

[0816] Monitoring cycle of the first AI model and / or the first AI function;

[0817] Monitoring termination indication information for the first AI model and / or the first AI function;

[0818] The reporting conditions for the second information.

[0819] Optionally, the reporting conditions for the second information include at least one of the following:

[0820] The monitoring end time configured for the first network device must be met;

[0821] The timer configured on the first network device stops or times out;

[0822] The occurrence of RLF was predicted, and the RLF actually occurred;

[0823] The occurrence of RLF was predicted, but RLF did not actually occur;

[0824] The prediction was that the RLF would not occur, but the RLF actually did occur.

[0825] Optionally, the second information includes at least one of the following:

[0826] The actual time when RLF occurs;

[0827] The cell identifier where the terminal is located;

[0828] After the first timer expires, the number of link synchronization indications received by the terminal;

[0829] Community quality when RLF actually occurs;

[0830] The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information;

[0831] Cell quality prior to the actual time of RLF occurrence;

[0832] The start status of the first timer;

[0833] The runtime of the first timer;

[0834] RLF (Real-Frequency) indication information that did not actually occur;

[0835] The quality of the cell where the terminal is located;

[0836] First Key Performance Indicator (KPI) information.

[0837] Optionally, the first KPI information includes at least one of the following:

[0838] The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality;

[0839] The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence;

[0840] The time difference between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0841] The time difference between the end of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0842] The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring;

[0843] Evaluation metrics for the first AI model and / or the first AI function.

[0844] Optionally, the device further includes:

[0845] The seventh information receiving module is used to receive the second information sent by the second network device;

[0846] The first information is sent by the terminal to the second network device;

[0847] The second network device is the network device corresponding to the cell selected by the terminal after the RLF actually occurs.

[0848] Optionally, the device further includes:

[0849] The sixth information sending module is used to send third indication information to the terminal, the third indication information being used to instruct the terminal to perform the first operation;

[0850] The first operation includes at least one of the following:

[0851] Switch to the second AI model and / or the second AI function;

[0852] Activate the second AI model and / or the second AI function;

[0853] To activate the first AI model and / or the first AI function;

[0854] Revert to non-AI operation.

[0855] Optionally, the third indication information is used for at least one of the following:

[0856] The terminal is instructed to access the first cell and execute the first operation;

[0857] The terminal is instructed to perform the first operation based on the cell signal quality;

[0858] The terminal is instructed to perform the first operation according to the link asynchrony indication;

[0859] The terminal is instructed to perform the first operation based on the KPI information;

[0860] KPI thresholds.

[0861] Optionally, the device further includes:

[0862] The seventh information sending module sends the fourth instruction information to the terminal;

[0863] The fourth indication information is used to indicate at least one of the following:

[0864] Instruct the terminal to switch to the second AI model and / or the second AI function;

[0865] The terminal is instructed to activate the second AI model and / or the second AI function;

[0866] Instruct the terminal to activate the first AI model and / or the first AI function;

[0867] Instruct the terminal to revert to non-AI operation.

[0868] It should be noted that the wireless link failure prediction device provided in the embodiments of the present invention is a device capable of executing the above-described wireless link failure prediction method applied to the first network device. Therefore, all embodiments of the above-described wireless link failure prediction method applied to the first network device are applicable to this device and can achieve the same or similar technical effects.

[0869] like Figure 11 As shown, this embodiment of the invention also provides a terminal, including: a processor 1101; and a memory 1103 connected to the processor 1101 via a bus interface 1102, the memory 1103 being used to store programs and data used by the processor 1101 when performing operations, and the processor 1101 calling and executing the programs and data stored in the memory 1103.

[0870] The transceiver 1104 is connected to the bus interface 1102 and is used to receive and send data under the control of the processor 1101. Specifically, the processor 1101 is used to read the program in the memory 1103, and the transceiver 1104 is used to execute the following processes:

[0871] Send first information to the first network device, and / or send second information to the first network device;

[0872] The first information includes information related to radio link failure (RLF) prediction;

[0873] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0874] Optionally, the transceiver 1104 is further configured to:

[0875] The first configuration information sent by the first network device is configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction.

[0876] Optionally, the processor 1101 is used for:

[0877] The first information is obtained based on the first AI model and / or the first AI function.

[0878] Optionally, the first configuration information includes at least one of the following:

[0879] Measurement configuration information;

[0880] Observation window for RLF prediction;

[0881] The forecast window for RLF forecasting;

[0882] The forecast period for RLF forecasts;

[0883] RLF occurrence probability threshold;

[0884] The first piece of information to be reported;

[0885] The reporting conditions for the first information;

[0886] RRM predicts configuration information.

[0887] Optionally, the reporting conditions for the first information include at least one of the following:

[0888] The difference between the predicted time of RLF occurrence and the current time is less than the preset time difference;

[0889] The terminal receives a first preset number of consecutive link asynchrony indications;

[0890] Meet the reporting cycle;

[0891] RLF was predicted to occur;

[0892] The predicted probability of RLF occurrence is higher than the RLF occurrence probability threshold;

[0893] The reporting time meets the configuration.

[0894] Optionally, the processor 1101 is used for:

[0895] Based on the first AI model and / or the first AI function, RRM prediction information is obtained;

[0896] The first information is obtained based on the RRM prediction information.

[0897] Optionally, the transceiver 1104 is further configured to:

[0898] Send the first instruction information to the first network device;

[0899] Wherein, the first indication information is used to indicate at least one of the following:

[0900] The terminal is instructed to support RLF prediction based on AI models and / or AI functions;

[0901] The terminal is instructed to support RRM prediction based on AI models and / or AI functions;

[0902] Indicates the AI ​​models and / or AI functions supported by the terminal;

[0903] The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal;

[0904] Input information indicating the AI ​​models and / or AI functions supported by the terminal;

[0905] Output information indicating the AI ​​models and / or AI functions supported by the terminal.

[0906] Optionally, the transceiver 1104 is further configured to:

[0907] Receive the second indication information sent by the first network device;

[0908] The second indication information is used to indicate at least one of the following:

[0909] The terminal is instructed to activate the AI ​​model and / or AI function;

[0910] Instruct the terminal to activate the AI ​​model and / or AI function;

[0911] Instruct the terminal to use the AI ​​model and / or AI function for RLF prediction;

[0912] The terminal is instructed to access the first cell and perform RLF prediction;

[0913] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality;

[0914] The terminal is instructed to activate the AI ​​model and / or AI function based on cell signal quality; the terminal is instructed to activate the AI ​​model and / or AI function based on link asynchrony indication; the AI ​​model and / or AI function is used for RLF prediction;

[0915] AI models and / or AI functions are used for RRM prediction.

[0916] Optionally, the second indication information includes at least one of the following:

[0917] Radio Resource Control (RRC) signaling;

[0918] Media intervention control layer control cell MAC CE;

[0919] Downlink Control Information (DCI);

[0920] System Information Block (SIB).

[0921] Optionally, the first information includes at least one of the following:

[0922] The start status of the first timer;

[0923] The runtime of the first timer;

[0924] Remaining duration of the first timer;

[0925] The number of link synchronization indications received when an RLF is predicted;

[0926] Methods of prediction;

[0927] Predicting whether RLF will occur or not;

[0928] Predict the timing of RLF occurrence;

[0929] Prediction window length;

[0930] Predict the probability of RLF occurring within the window;

[0931] Predict the cell quality when RLF occurs.

[0932] Optionally, the transceiver 1104 is further configured to:

[0933] The system receives second configuration information sent by the first network device, wherein the second configuration information is configuration information related to the second information.

[0934] Optionally, the second configuration information includes at least one of the following:

[0935] The start time of monitoring the first AI model and / or the first AI function;

[0936] The end time of monitoring the first AI model and / or the first AI function;

[0937] Monitoring cycle of the first AI model and / or the first AI function;

[0938] Monitoring termination indication information for the first AI model and / or the first AI function;

[0939] The reporting conditions for the second information.

[0940] Optionally, the reporting conditions for the second information include at least one of the following:

[0941] The monitoring end time configured for the first network device must be met;

[0942] The timer configured on the first network device stops or times out;

[0943] The occurrence of RLF was predicted, and the RLF actually occurred;

[0944] The occurrence of RLF was predicted, but RLF did not actually occur;

[0945] The prediction was that the RLF would not occur, but the RLF actually did occur.

[0946] Optionally, the second information includes at least one of the following:

[0947] The actual time when RLF occurs;

[0948] The cell identifier where the terminal is located;

[0949] After the first timer expires, the number of link synchronization indications received by the terminal;

[0950] Community quality when RLF actually occurs;

[0951] The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information;

[0952] Cell quality prior to the actual time of RLF occurrence;

[0953] The start status of the first timer;

[0954] The runtime of the first timer;

[0955] RLF (Real-Frequency) indication information that did not actually occur;

[0956] The quality of the cell where the terminal is located;

[0957] First Key Performance Indicator (KPI) information.

[0958] Optionally, the first KPI information includes at least one of the following:

[0959] The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality;

[0960] The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence;

[0961] The time difference between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0962] The time difference between the end of the prediction window for RLF prediction and the actual time when the RLF occurs;

[0963] The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring;

[0964] Evaluation metrics for the first AI model and / or the first AI function.

[0965] Optionally, the transceiver 1104 is further configured to:

[0966] The second information sent to the second network device;

[0967] The second network device sends the second information to the first network device;

[0968] The second network device is the network device corresponding to the cell selected by the terminal for reconstruction after the RLF actually occurs.

[0969] Optionally, the transceiver 1104 is further configured to:

[0970] The terminal receives a third indication message sent by the first network device, the third indication message being used to instruct the terminal to perform a first operation;

[0971] The first operation includes at least one of the following:

[0972] Switch to the second AI model and / or the second AI function;

[0973] Activate the second AI model and / or the second AI function;

[0974] To activate the first AI model and / or the first AI function;

[0975] Revert to non-AI operation.

[0976] Optionally, the processor 1101 is used for:

[0977] The first operation is performed based on the third instruction information and the first KPI information.

[0978] Optionally, the third indication information is used for at least one of the following:

[0979] The terminal is instructed to access the first cell and execute the first operation;

[0980] The terminal is instructed to perform the first operation based on the cell signal quality;

[0981] The terminal is instructed to perform the first operation according to the link asynchrony indication;

[0982] The terminal is instructed to perform the first operation based on the KPI information;

[0983] KPI thresholds.

[0984] Optionally, the transceiver 1104 is further configured to:

[0985] Receive the fourth indication information sent by the first network device;

[0986] The fourth indication information is used to indicate at least one of the following:

[0987] Instruct the terminal to switch to the second AI model and / or the second AI function;

[0988] The terminal is instructed to activate the second AI model and / or the second AI function;

[0989] Instruct the terminal to activate the first AI model and / or the first AI function;

[0990] Instruct the terminal to revert to non-AI operation.

[0991] Among them, Figure 11 In this context, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits together, represented by one or more processors (processor 1101) and memory (memory 1103). The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. A bus interface provides a user interface 1105. A transceiver 1104 may be multiple elements, including transmitters and receivers, providing units for communicating with various other devices over a transmission medium. Processor 1101 is responsible for managing the bus architecture and general processing, and memory 1103 may store data used by processor 1101 during operation.

[0992] like Figure 12 As shown, this embodiment of the invention also provides a network device, which is a first network device, including: a processor 1201; and a memory 1203 connected to the processor 1201 via a bus interface 1202. The memory 1203 is used to store programs and data used by the processor 1201 when performing operations, and the processor 1201 calls and executes the programs and data stored in the memory 1203.

[0993] The transceiver 1204 is connected to the bus interface 1202 and is used to receive and send data under the control of the processor 1201. Specifically, the processor 1201 is used to read the program in the memory 1203, and the transceiver 1204 is used to execute the following processes:

[0994] The receiving terminal sends a first message, and / or the receiving terminal sends a second message;

[0995] The first information includes information related to radio link failure (RLF) prediction;

[0996] The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

[0997] Optionally, the transceiver 1204 is further configured to:

[0998] Send first configuration information to the terminal, the first configuration information being configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction.

[0999] Optionally, the first configuration information includes at least one of the following:

[1000] Measurement configuration information;

[1001] Observation window for RLF prediction;

[1002] The forecast window for RLF forecasting;

[1003] The forecast period for RLF forecasts;

[1004] RLF occurrence probability threshold;

[1005] The first piece of information to be reported;

[1006] The reporting conditions for the first information;

[1007] RRM predicts configuration information.

[1008] Optionally, the reporting conditions for the first information include at least one of the following:

[1009] The difference between the predicted time of RLF occurrence and the current time is less than the preset time difference;

[1010] The terminal receives a first preset number of consecutive link asynchrony indications;

[1011] Meet the reporting cycle;

[1012] RLF was predicted to occur;

[1013] The predicted probability of RLF occurrence is higher than the RLF occurrence probability threshold;

[1014] The reporting time meets the configuration.

[1015] Optionally, the transceiver 1204 is further configured to:

[1016] Receive the first indication information sent by the terminal;

[1017] Wherein, the first indication information is used to indicate at least one of the following:

[1018] The terminal is instructed to support RLF prediction based on AI models and / or AI functions;

[1019] The terminal is instructed to support RRM prediction based on AI models and / or AI functions;

[1020] Indicates the AI ​​models and / or AI functions supported by the terminal;

[1021] The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal;

[1022] Input information indicating the AI ​​models and / or AI functions supported by the terminal;

[1023] Output information indicating the AI ​​models and / or AI functions supported by the terminal.

[1024] Optionally, the transceiver 1204 is further configured to:

[1025] Send a second instruction message to the terminal;

[1026] The second indication information is used to indicate at least one of the following:

[1027] The terminal is instructed to activate the AI ​​model and / or AI function;

[1028] Instruct the terminal to activate the AI ​​model and / or AI function;

[1029] Instruct the terminal to use the AI ​​model and / or AI function for RLF prediction;

[1030] The terminal is instructed to access the first cell and perform RLF prediction;

[1031] The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality;

[1032] The terminal is instructed to activate the AI ​​model and / or AI function based on cell signal quality; the terminal is instructed to activate the AI ​​model and / or AI function based on link asynchrony indication; the AI ​​model and / or AI function is used for RLF prediction;

[1033] AI models and / or AI functions are used for RRM prediction.

[1034] Optionally, the second indication information includes at least one of the following:

[1035] RRC signaling;

[1036] MAC CE;

[1037] DCI;

[1038] SIB.

[1039] Optionally, the first information includes at least one of the following:

[1040] The start status of the first timer;

[1041] The runtime of the first timer;

[1042] Remaining duration of the first timer;

[1043] The number of link synchronization indications received when an RLF is predicted;

[1044] Methods of prediction;

[1045] Predicting whether RLF will occur or not;

[1046] Predict the timing of RLF occurrence;

[1047] Prediction window length;

[1048] Predict the probability of RLF occurring within the window;

[1049] Predict the cell quality when RLF occurs.

[1050] Optionally, the transceiver 1204 is further configured to:

[1051] Send second configuration information to the terminal, the second configuration information being configuration information related to the second information.

[1052] Optionally, the second configuration information includes at least one of the following:

[1053] The start time of monitoring the first AI model and / or the first AI function;

[1054] The end time of monitoring the first AI model and / or the first AI function;

[1055] Monitoring cycle of the first AI model and / or the first AI function;

[1056] Monitoring termination indication information for the first AI model and / or the first AI function;

[1057] The reporting conditions for the second information.

[1058] Optionally, the reporting conditions for the second information include at least one of the following:

[1059] The monitoring end time configured for the first network device must be met;

[1060] The timer configured on the first network device stops or times out;

[1061] The occurrence of RLF was predicted, and the RLF actually occurred;

[1062] The occurrence of RLF was predicted, but RLF did not actually occur;

[1063] The prediction was that the RLF would not occur, but the RLF actually did occur.

[1064] Optionally, the second information includes at least one of the following:

[1065] The actual time when RLF occurs;

[1066] The cell identifier where the terminal is located;

[1067] After the first timer expires, the number of link synchronization indications received by the terminal;

[1068] Community quality when RLF actually occurs;

[1069] The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information;

[1070] Cell quality prior to the actual time of RLF occurrence;

[1071] The start status of the first timer;

[1072] The runtime of the first timer;

[1073] RLF (Real-Frequency) indication information that did not actually occur;

[1074] The quality of the cell where the terminal is located;

[1075] First Key Performance Indicator (KPI) information.

[1076] Optionally, the first KPI information includes at least one of the following:

[1077] The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality;

[1078] The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence;

[1079] The time difference between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs;

[1080] The time difference between the end of the prediction window for RLF prediction and the actual time when the RLF occurs;

[1081] The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring;

[1082] Evaluation metrics for the first AI model and / or the first AI function.

[1083] Optionally, the transceiver 1204 is further configured to:

[1084] Receive the second information sent by the second network device;

[1085] The first information is sent by the terminal to the second network device;

[1086] The second network device is the network device corresponding to the cell selected by the terminal after the RLF actually occurs.

[1087] Optionally, the transceiver 1204 is further configured to:

[1088] Send a third instruction message to the terminal, the third instruction message being used to instruct the terminal to perform a first operation;

[1089] The first operation includes at least one of the following:

[1090] Switch to the second AI model and / or the second AI function;

[1091] Activate the second AI model and / or the second AI function;

[1092] To activate the first AI model and / or the first AI function;

[1093] Revert to non-AI operation.

[1094] Optionally, the third indication information is used for at least one of the following:

[1095] The terminal is instructed to access the first cell and execute the first operation;

[1096] The terminal is instructed to perform the first operation based on the cell signal quality;

[1097] The terminal is instructed to perform the first operation according to the link asynchrony indication;

[1098] The terminal is instructed to perform the first operation based on the KPI information;

[1099] KPI thresholds.

[1100] Optionally, the transceiver 1204 is further configured to:

[1101] Send a fourth instruction message to the terminal;

[1102] The fourth indication information is used to indicate at least one of the following:

[1103] Instruct the terminal to switch to the second AI model and / or the second AI function;

[1104] The terminal is instructed to activate the second AI model and / or the second AI function;

[1105] Instruct the terminal to activate the first AI model and / or the first AI function;

[1106] Instruct the terminal to revert to non-AI operation.

[1107] Among them, Figure 12In this context, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits together, represented by one or more processors (processor 1201) and memory (memory 1203). The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. The transceiver 1204 may be multiple elements, including transmitters and receivers, providing a unit for communicating with various other devices over a transmission medium. The processor 1201 is responsible for managing the bus architecture and general processing, and the memory 1203 may store data used by the processor 1201 during operation.

[1108] In addition, specific embodiments of the present invention also provide a readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, it implements the steps in the wireless link failure prediction method for a terminal as described above, or implements the steps in the wireless link failure prediction method for a first network device as described above.

[1109] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[1110] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can be physically comprised separately, or two or more units can be integrated into one unit. The integrated unit described above can be implemented in hardware or in the form of hardware plus software functional units.

[1111] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions that cause a computer device (which may be a personal computer, server, or network device, etc.) to execute partial steps of the resource selection method described in the various embodiments of the present invention, or to execute partial steps of the information transmission method described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[1112] A specific embodiment of the present invention also provides a computer program product, including computer instructions, which, when executed by a processor, implement the above-described functionality. Figure 1 or Figure 8 The various processes of the method embodiments shown can achieve the same technical effect, and will not be described again here to avoid repetition.

[1113] The above describes the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also within the scope of protection of the present invention.

Claims

1. A method for predicting wireless link failures, characterized in that, Applied to a terminal, the method includes: Send first information to the first network device, and / or send second information to the first network device; The first information includes information related to radio link failure (RLF) prediction; The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

2. The method according to claim 1, characterized in that, The method further includes: The first configuration information sent by the first network device is configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction.

3. The method according to claim 1 or 2, characterized in that, The method further includes: The first information is obtained based on the first AI model and / or the first AI function.

4. The method according to claim 2, characterized in that, The first configuration information includes at least one of the following: Measurement configuration information; Observation window for RLF prediction; The forecast window for RLF forecasting; The forecast period for RLF forecasts; RLF occurrence probability threshold; The first piece of information to be reported; The reporting conditions for the first information; RRM predicts configuration information.

5. The method according to claim 4, characterized in that, The reporting conditions for the first information include at least one of the following: The difference between the predicted time of RLF occurrence and the current time is less than the preset time difference; The terminal receives a first preset number of consecutive link asynchrony indications; Meet the reporting cycle; RLF was predicted to occur; The predicted probability of RLF occurrence is higher than the RLF occurrence probability threshold; The reporting time meets the configuration.

6. The method according to claim 1 or 4, characterized in that, The method further includes: Based on the first AI model and / or the first AI function, RRM prediction information is obtained; The first information is obtained based on the RRM prediction information.

7. The method according to claim 1, characterized in that, The method further includes: Send the first instruction information to the first network device; Wherein, the first indication information is used to indicate at least one of the following: The terminal is instructed to support RLF prediction based on AI models and / or AI functions; The terminal is instructed to support RRM prediction based on AI models and / or AI functions; Indicates the AI ​​models and / or AI functions supported by the terminal; The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal; Input information indicating the AI ​​models and / or AI functions supported by the terminal; Output information indicating the AI ​​models and / or AI functions supported by the terminal.

8. The method according to claim 1, characterized in that, The method further includes: Receive the second indication information sent by the first network device; The second indication information is used to indicate at least one of the following: The terminal is instructed to activate the AI ​​model and / or AI function; Instruct the terminal to activate the AI ​​model and / or AI function; Instruct the terminal to use the AI ​​model and / or AI function for RLF prediction; The terminal is instructed to access the first cell and perform RLF prediction; The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality; The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality; The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication; The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication; AI models and / or AI functions are used for RLF prediction; AI models and / or AI functions are used for RRM prediction.

9. The method according to claim 8, characterized in that, The second instruction information includes at least one of the following: Radio Resource Control (RRC) signaling; Media intervention control layer control cell MAC CE; Downlink Control Information (DCI); System Information Block (SIB).

10. The method according to claim 1, characterized in that, The first information includes at least one of the following: The start status of the first timer; The runtime of the first timer; Remaining duration of the first timer; The number of link synchronization indications received when an RLF is predicted; Methods of prediction; Predicting whether RLF will occur or not; Predict the timing of RLF occurrence; Prediction window length; Predict the probability of RLF occurring within the window; Predicting cell quality when RLF occurs; RRM prediction information.

11. The method according to claim 1, characterized in that, The method further includes: The system receives second configuration information sent by the first network device, wherein the second configuration information is configuration information related to the second information.

12. The method according to claim 11, characterized in that, The second configuration information includes at least one of the following: The start time of monitoring the first AI model and / or the first AI function; The end time of monitoring the first AI model and / or the first AI function; Monitoring cycle of the first AI model and / or the first AI function; Monitoring termination indication information for the first AI model and / or the first AI function; The reporting conditions for the second information.

13. The method according to claim 12, characterized in that, The reporting conditions for the second information include at least one of the following: The monitoring end time configured for the first network device must be met; The timer configured on the first network device stops or times out; The occurrence of RLF was predicted, and the RLF actually occurred; The occurrence of RLF was predicted, but RLF did not actually occur; The prediction was that the RLF would not occur, but the RLF actually did occur.

14. The method according to claim 1, characterized in that, The second information includes at least one of the following: The actual time when RLF occurs; The cell identifier where the terminal is located; After the first timer expires, the number of link synchronization indications received by the terminal; Community quality when RLF actually occurs; The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information; Cell quality prior to the actual time of RLF occurrence; The start status of the first timer; The runtime of the first timer; RLF (Real-Frequency) indication information that did not actually occur; The quality of the cell where the terminal is located; First Key Performance Indicator (KPI) information.

15. The method according to claim 14, characterized in that, The first KPI information includes at least one of the following: The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality; The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence; The time difference between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs; The time difference between the end of the prediction window for RLF prediction and the actual time when the RLF occurs; The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring; Evaluation metrics for the first AI model and / or the first AI function.

16. The method according to claim 1, characterized in that, The method further includes: The second information sent to the second network device; The second network device sends the second information to the first network device; The second network device is the network device corresponding to the cell selected by the terminal for reconstruction after the RLF actually occurs.

17. The method according to claim 1, characterized in that, The method further includes: The terminal receives a third indication message sent by the first network device, the third indication message being used to instruct the terminal to perform a first operation; The first operation includes at least one of the following: Switch to the second AI model and / or the second AI function; Activate the second AI model and / or the second AI function; To activate the first AI model and / or the first AI function; Revert to non-AI operation.

18. The method according to claim 17, characterized in that, The method further includes: The first operation is performed based on the third instruction information and the first KPI information.

19. The method according to claim 17 or 18, characterized in that, The third instruction information is used for at least one of the following: The terminal is instructed to access the first cell and execute the first operation; The terminal is instructed to perform the first operation based on the cell signal quality; The terminal is instructed to perform the first operation according to the link asynchrony indication; The terminal is instructed to perform the first operation based on the KPI information; KPI thresholds.

20. The method according to claim 1, characterized in that, The method further includes: Receive the fourth indication information sent by the first network device; The fourth indication information is used to indicate at least one of the following: Instruct the terminal to switch to the second AI model and / or the second AI function; The terminal is instructed to activate the second AI model and / or the second AI function; Instruct the terminal to activate the first AI model and / or the first AI function; Instruct the terminal to revert to non-AI operation.

21. A method for predicting wireless link failures, characterized in that, Applied to a first network device, the method includes: The receiving terminal sends a first message, and / or the receiving terminal sends a second message; The first information includes information related to radio link failure (RLF) prediction; The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

22. The method according to claim 21, characterized in that, The method further includes: Send first configuration information to the terminal, the first configuration information being configuration information related to RLF prediction and / or Radio Resource Management measurement RRM prediction.

23. The method according to claim 22, characterized in that, The first configuration information includes at least one of the following: Measurement configuration information; Observation window for RLF prediction; The forecast window for RLF forecasting; The forecast period for RLF forecasts; RLF occurrence probability threshold; The first piece of information to be reported; The reporting conditions for the first information; RRM predicts configuration information.

24. The method according to claim 23, characterized in that, The reporting conditions for the first information include at least one of the following: The difference between the predicted time of RLF occurrence and the current time is less than the preset time difference; The terminal receives a first preset number of consecutive link asynchrony indications; Meet the reporting cycle; RLF was predicted to occur; The predicted probability of RLF occurrence is higher than the RLF occurrence probability threshold; The reporting time meets the configuration.

25. The method according to claim 21, characterized in that, The method further includes: Receive the first indication information sent by the terminal; Wherein, the first indication information is used to indicate at least one of the following: The terminal is instructed to support RLF prediction based on AI models and / or AI functions; The terminal is instructed to support RRM prediction based on AI models and / or AI functions; Indicates the AI ​​models and / or AI functions supported by the terminal; The accuracy information indicates the AI ​​models and / or AI functions supported by the terminal; Input information indicating the AI ​​models and / or AI functions supported by the terminal; Output information indicating the AI ​​models and / or AI functions supported by the terminal.

26. The method according to claim 21, characterized in that, The method further includes: Send a second instruction message to the terminal; The second indication information is used to indicate at least one of the following: The terminal is instructed to activate the AI ​​model and / or AI function; Instruct the terminal to activate the AI ​​model and / or AI function; Instruct the terminal to use the AI ​​model and / or AI function for RLF prediction; The terminal is instructed to access the first cell and perform RLF prediction; The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality; The terminal is instructed to activate the AI ​​model and / or AI function based on the cell signal quality; The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication; The terminal is instructed to activate the AI ​​model and / or AI function according to the link asynchrony indication; AI models and / or AI functions are used for RLF prediction; AI models and / or AI functions are used for RRM prediction.

27. The method according to claim 26, characterized in that, The second instruction information includes at least one of the following: RRC signaling; MAC CE; DCI; SIB.

28. The method according to claim 21, characterized in that, The first information includes at least one of the following: The start status of the first timer; The runtime of the first timer; Remaining duration of the first timer; The number of link synchronization indications received when an RLF is predicted; Methods of prediction; Predicting whether RLF will occur or not; Predict the timing of RLF occurrence; Prediction window length; Predict the probability of RLF occurring within the window; Predict the cell quality when RLF occurs.

29. The method according to claim 21, characterized in that, The method further includes: Send second configuration information to the terminal, the second configuration information being configuration information related to the second information.

30. The method according to claim 29, characterized in that, The second configuration information includes at least one of the following: The start time of monitoring the first AI model and / or the first AI function; The end time of monitoring the first AI model and / or the first AI function; Monitoring cycle of the first AI model and / or the first AI function; Monitoring termination indication information for the first AI model and / or the first AI function; The reporting conditions for the second information.

31. The method according to claim 30, characterized in that, The reporting conditions for the second information include at least one of the following: The monitoring end time configured for the first network device must be met; The timer configured on the first network device stops or times out; The occurrence of RLF was predicted, and the RLF actually occurred; The occurrence of RLF was predicted, but RLF did not actually occur; The prediction was that the RLF would not occur, but the RLF actually did occur.

32. The method according to claim 22, characterized in that, The second information includes at least one of the following: The actual time when RLF occurs; The cell identifier where the terminal is located; After the first timer expires, the number of link synchronization indications received by the terminal; Community quality when RLF actually occurs; The cell quality between the predicted time of RLF occurrence and the actual time of RLF occurrence in the RLF prediction information; Cell quality prior to the actual time of RLF occurrence; The start status of the first timer; The runtime of the first timer; RLF (Real-Frequency) indication information that did not actually occur; The quality of the cell where the terminal is located; First Key Performance Indicator (KPI) information.

33. The method according to claim 32, characterized in that, The first KPI information includes at least one of the following: The signal quality difference between the signal quality predicted by the first AI model and / or the first AI function and the actual signal quality; The time difference between the predicted time of RLF occurrence and the actual time of RLF occurrence; The time difference between the start time of the prediction window for RLF prediction and the actual time when the RLF occurs; The time difference between the end of the prediction window for RLF prediction and the actual time when the RLF occurs; The probability difference between the probability of an RLF occurring within the prediction window and the actual probability of an RLF occurring; Evaluation metrics for the first AI model and / or the first AI function.

34. The method according to claim 21, characterized in that, The method further includes: Receive the second information sent by the second network device; The first information is sent by the terminal to the second network device; The second network device is the network device corresponding to the cell selected by the terminal after the RLF actually occurs.

35. The method according to claim 21, characterized in that, The method further includes: Send a third instruction message to the terminal, the third instruction message being used to instruct the terminal to perform a first operation; The first operation includes at least one of the following: Switch to the second AI model and / or the second AI function; Activate the second AI model and / or the second AI function; To activate the first AI model and / or the first AI function; Revert to non-AI operation.

36. The method according to claim 35, characterized in that, The third instruction information is used for at least one of the following: The terminal is instructed to access the first cell and execute the first operation; The terminal is instructed to perform the first operation based on the cell signal quality; The terminal is instructed to perform the first operation according to the link asynchrony indication; The terminal is instructed to perform the first operation based on the KPI information; KPI thresholds.

37. The method according to claim 21, characterized in that, The method further includes: Send a fourth instruction message to the terminal; The fourth indication information is used to indicate at least one of the following: Instruct the terminal to switch to the second AI model and / or the second AI function; The terminal is instructed to activate the second AI model and / or the second AI function; Instruct the terminal to activate the first AI model and / or the first AI function; Instruct the terminal to revert to non-AI operation.

38. A wireless link failure prediction device, characterized in that, The device includes: The first sending module is used to send first information to the first network device, and / or send second information to the first network device; The first information includes information related to radio link failure (RLF) prediction; The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

39. A wireless link failure prediction device, characterized in that, The device includes: The first receiving module is used to receive first information sent by the terminal, and / or to receive second information sent by the terminal; The first information includes information related to radio link failure (RLF) prediction; The second information includes information related to the monitoring of the first artificial intelligence (AI) model and / or the first AI function.

40. A terminal, characterized in that, include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the wireless link failure prediction method as claimed in any one of claims 1 to 20.

41. A network device, characterized in that, include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the wireless link failure prediction method as claimed in any one of claims 21 to 37.

42. A readable storage medium, characterized in that, The readable storage medium stores a program that, when executed by a processor, implements the steps of the wireless link failure prediction method as described in any one of claims 1 to 20, or, when executed, implements the steps of the wireless link failure prediction method as described in any one of claims 21 to 37.

43. A computer program product, characterized in that, The method includes computer instructions that, when executed by a processor, implement the steps of the wireless link failure prediction method as described in any one of claims 1 to 20, or, when executed, implement the steps of the wireless link failure prediction method as described in any one of claims 21 to 37.

Images