Prediction information reporting method, apparatus and system, and storage medium

Abstract

The present application relates to the technical field of communications, and provides a prediction information reporting method, apparatus and system, and a storage medium. In the solution of the present application, a network device configures for a terminal device an occasion for executing RLF or HOF event prediction reporting, and the terminal device can obtain at the occasion a prediction result for an RLF or HOF event on the basis of an AI model, and reports the prediction result for the RLF or HOF event to the network device, thereby facilitating the network device to execute a corresponding handover (HO) strategy on the basis of the prediction result for the RLF or HOF event. HO is assisted by means of the AI model, so that the occurrence of a situation such as premature HO, late HO, or HO to a wrong cell can be reduced, improving the success rate of HO. In addition, RLF or HOF events within a period of time can be predicted on the basis of the AI model, and within the predicted period of time, the terminal device does not need to execute measurements, reducing the number of executions of a measurement action.

Description

Method, device and system for reporting prediction information and storage medium

[0001] The present application claims priority from the Chinese patent application No. 202410650545.0 filed on May 23, 2024, and entitled "Method, device and system for reporting prediction information and storage medium", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to the field of communication technology, and in particular to a method, device and system for reporting prediction information and storage medium. BACKGROUND

[0003] Handover (HO) is an important mobile communication technology. When a terminal device moves from one base station coverage area to another during a call, or the call quality is reduced due to external interference, it is necessary to switch from the original channel to another idle channel. This process is called HO.

[0004] Generally, handover failure (HOF) occurs in the following three situations: early HO, late HO, and HO to the wrong cell. Regardless of the situation, HOF can cause call drops and affect the running quality of the network. SUMMARY

[0005] The present application provides a method, device and system for reporting prediction information and storage medium to solve the technical problem of HOF affecting the running quality of the network.

[0006] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0007] In a first aspect, the present application provides a method for reporting prediction information. The method can be applied to a terminal device. The method can include: receiving configuration information from a network device, the configuration information including a reporting opportunity; and sending prediction information obtained based on an AI model to the network device at the reporting opportunity, the prediction information including a prediction result of an RLF or HOF event.

[0008] By the above scheme, the terminal device can obtain the prediction result of the RLF or HOF event by the AI model at the reporting time point configured by the network device, and report the prediction result of the RLF or HOF event to the network device, such as the prediction result can include the handover period, the recommended cell identifier to be switched, etc., so as to facilitate the network device to configure parameters according to the prediction result of the RLF or HOF event, and execute the corresponding handover strategy. By AI model assisted HO, the occurrence of premature HO, late HO or HO to the wrong cell can be reduced, the success rate of handover is improved, and the running quality of the network is ensured. In addition, based on the AI model, the RLF or HOF event in a period of time can be predicted, and the terminal device does not need to perform measurement in this prediction time period, thereby reducing the execution frequency of measurement action.

[0009] In a possible implementation, the AI model is stored in the terminal device. The AI model can take the real measurement value of the terminal device as input, predict and output the prediction result of the RLF or HOF event, that is, the AI model can predict whether the RLF or HOF event will occur in a future period of time based on the real measurement value, and formulate a strategy to reduce the occurrence of the RLF or HOF event. As an example, the AI model can include an RLF or HOF prediction model. As another example, the AI model can include an RRM prediction model and an RLF or HOF prediction model. The RRM prediction model is also called Model 1, which is used to predict the measurement result in a future period of time based on the current measurement result. The RLF or HOF prediction model is also called Model 2, which is used to predict the RLF or HOF event that may occur in a future period of time by taking the current measurement result or measurement value as input, or taking the prediction result or predicted value output by the RRM prediction model as input. It can be understood that if the AI model includes the RLF or HOF prediction model, the RLF / HOF prediction model can be used to predict the related result of the RLF or HOF event in a period of time, and the terminal device does not need to perform measurement in this predictable time period; if the AI model includes the RRM prediction model and the RLF or HOF prediction model, measurement is not needed in the period of time predicted by the RRM prediction model and the period of time predicted by the RLF / HOF prediction model when the two models are used for prediction. In this way, the execution frequency of the measurement action can be reduced, and more opportunities for energy saving and service transmission can be brought to the terminal device.

[0010] In a possible implementation, the above reporting time point can be divided into the following two cases:

[0011] The first case is that the reporting time can include a start time of reporting the prediction information and a time interval AT of reporting the prediction information. The start time is a time point of activating the AI model. It can be understood that the network device configures the start time of the AI model (such as Model 1) activated as the start time of the prediction reporting and the time interval of reporting the prediction information, so that the terminal device can perform the prediction information reporting immediately after activating the AI model.

[0012] The second case is that the reporting time can include a prediction window and a time point in the prediction window. The AI model is activated in the prediction window, and the terminal device performs the prediction information reporting at the time point. Exemplarily, the prediction window can include a start time point and an end time point of the prediction window, the start time point is a first preset frame or a first preset time slot, and the end time point is a second preset frame or a second preset time slot. The time point in the prediction window can include a third preset frame, a third preset time slot, or a time interval AT of reporting the prediction information in the prediction window. It can be understood that the network device configures the time point in the prediction window, so that the terminal device can perform the prediction information reporting at the time point in the prediction window.

[0013] In a possible implementation, the sending, to the network device, of the prediction information based on the AI model at the reporting time includes: inputting, at the reporting time, the measurement value into the AI model to obtain the prediction information, and sending the prediction information to the network device. The measurement value includes at least one of the following: the layer-1 measurement value and the layer-3 measurement value.

[0014] Exemplarily, if the network device configures multiple reporting times, the terminal device can perform the following operations at each reporting time configured by the network device: inputting the measurement value into the AI model to obtain the prediction information, and sending the prediction information corresponding to the reporting time to the network device. The measurement value includes at least one of the following: the layer-1 measurement value and the layer-3 measurement value. It can be understood that in each reporting time, if different measurement values are input into the AI model, the AI model can output different prediction results of the RLF or HOF event based on different measurement values.

[0015] In a possible implementation, the configuration information sent by the network device is configured based on the prediction capability of the terminal device. If the AI model includes Model1 and Model2, before the terminal device receives the configuration information from the network device, the method can further include: sending the prediction capability of the terminal device to the network device. For example, the prediction capability of the terminal device is sent to the network device through user equipment assistance information signaling or radio resource control signaling. The prediction capability includes at least one of the following: a first prediction time length Δt1, also referred to as a time length of Model1 measurement prediction, within the first prediction time length Δt1, Model1 supports obtaining a predicted value based on a measured value; an indication that the predicted value output by Model1 is supported as input of Model2, and the output of Model2 is a prediction result of an RLF or HOF event.

[0016] In a possible implementation, the reporting time includes at least a first reporting time and a second reporting time. The second reporting time is the last reporting time of the first reporting time. Accordingly, at the reporting time, the measured value is input into the AI model to obtain the prediction information, and the prediction information is sent to the network device, which can include: if the time interval between the first reporting time and the second reporting time is less than or equal to the first prediction time length, and the terminal device obtains a first measured value by performing measurement at the second reporting time, at the first reporting time, the first measured value is input into Model1 to obtain a first predicted value, the first predicted value is input into Model2 to obtain a first prediction result of an RLF or HOF event, and first prediction information including the first prediction result is sent to the network device. Alternatively, if the time interval between the first reporting time and the second reporting time is less than or equal to the first prediction time length, and the terminal device does not perform measurement at the second reporting time, at the first reporting time, a second measured value is obtained by performing measurement, the second measured value is input into Model2 to obtain a second prediction result of an RLF or HOF event, and second prediction information including the second prediction result is sent to the network device. Alternatively, if the time interval between the first reporting time and the second reporting time is greater than the first prediction time length, at the first reporting time, a second measured value is obtained by performing measurement, the second measured value is input into Model2 to obtain a second prediction result of an RLF or HOF event, and second prediction information including the second prediction result is sent to the network device.

[0017] It can be understood that if there is a measurement value less than or equal to the first prediction time length Δt1 from the current prediction reporting opportunity, it means that it is within the prediction capability of Model 1, so Model 1 can use the measurement value to predict the measurement value of the current opportunity (i.e. the predicted value), and use it as the input of Model 2; if there is no measurement value less than or equal to the first prediction time length Δt1 from the current prediction reporting opportunity, it means that although it is within the prediction capability of Model 1, there is no measurement value available for input of Model 1, so it is necessary to perform measurement at the current reporting opportunity, and use the measurement result as the input of Model 2; if the time interval between the first reporting opportunity and the second reporting opportunity is greater than the first prediction time length Δt1, it means that it has exceeded the prediction capability of Model 1, so it is also necessary to perform measurement at the current reporting opportunity, and use the measurement result as the input of Model 2.

[0018] In a possible implementation, the first prediction information can further include that the data type input into Model 2 is a predicted value; and the second prediction information can further include that the data type input into Model 2 is a measurement value. It can be understood that by carrying the data type input into Model 2 in the prediction information, the performance of the AI model can be facilitated to be monitored by the network device.

[0019] In a possible implementation, the configuration information sent by the network device is configured based on the prediction capability of the terminal device. If the AI model includes Model 2, before receiving the configuration information from the network device, the method can further include: sending the prediction capability of the terminal device to the network device. For example, the prediction capability of the terminal device is sent to the network device through user equipment assistance information signaling or radio resource control signaling. The prediction capability can include the second prediction time length Δt2, also referred to as the prediction time length or time interval of Model 2. Within the second prediction time length Δt2, Model 2 supports obtaining the prediction result of the RLF or HOF event based on the current measurement value or predicted value, that is, Model 2 can predict the RLF or HOF event within the future Δt2 time. The predicted value is obtained based on the measurement value.

[0020] In a possible implementation, the reporting occasions configured by the network device can include a third reporting occasion and a fourth reporting occasion, and the fourth reporting occasion is a previous reporting occasion of the third reporting occasion. Accordingly, at the reporting occasion, inputting the measurement value into the AI model to obtain the prediction information and sending the prediction information to the network device can include: if a time interval between the third reporting occasion and the fourth reporting occasion is less than or equal to the second prediction duration, at the third reporting occasion, inputting the measurement value or the prediction value into Model2 to obtain a third prediction result of the RLF or HOF event, and sending third prediction information including the third prediction result to the network device. The third prediction information can further include a data type input into Model2, and the data type is the measurement value or the prediction value.

[0021] In a possible implementation, at the third reporting occasion, inputting the measurement value or the prediction value into Model2 to obtain the third prediction result of the RLF or HOF event can include: if the AI model only includes Model2, at the third reporting occasion, performing measurement to obtain a third measurement value, and inputting the third measurement value into Model2 to obtain the third prediction result; or, if the AI model includes Model1 and Model2, at the third reporting occasion, inputting the third measurement value obtained by performing measurement or the second prediction value obtained based on Model1 into Model2 to obtain the third prediction result.

[0022] Exemplarily, Model1 supports obtaining the prediction value based on the measurement value within the first prediction duration. Accordingly, at the third reporting occasion, inputting the third measurement value obtained by performing measurement or the second prediction value obtained based on Model1 into Model2 to obtain the third prediction result can include: if a time interval between the third reporting occasion and the fourth reporting occasion is greater than the first prediction duration, at the third reporting occasion, performing measurement to obtain the third measurement value, and inputting the third measurement value into Model2 to obtain the third prediction result. Or, if the time interval between the third reporting occasion and the fourth reporting occasion is less than or equal to the first prediction duration, and measurement is not performed at the fourth reporting occasion, at the third reporting occasion, performing measurement to obtain the third measurement value, and inputting the third measurement value into Model2 to obtain the third prediction result. Or, if the time interval between the third reporting occasion and the fourth reporting occasion is less than or equal to the first prediction duration, and a fourth measurement value is obtained by performing measurement at the fourth reporting occasion, at the third reporting occasion, inputting the fourth measurement value into Model1 to obtain the second prediction value, and inputting the second prediction value into Model2 to obtain the third prediction result.

[0023] It can be understood that if the network device configures the predicted reporting time interval ΔT <= Δt2, it means that each reporting occasion is within the prediction capability of Model2, and therefore Model2 can obtain a prediction result of the RLF or HOF event based on the measurement value or the predicted value, that is, the AI model-based reporting manner is enabled, and the traditional measurement reporting manner is implicitly disabled.

[0024] In a possible implementation, the method can further include: if the time interval between the third reporting occasion and the fourth reporting occasion is greater than the second prediction duration, performing the following two operations: the first operation is that, at the third reporting occasion, performing measurement to obtain a fifth measurement value, inputting the fifth measurement value into Model2 to obtain a fourth prediction result, and sending the fourth prediction information to the network device, the fourth prediction information including the fourth prediction result; the second operation is that, enabling the traditional measurement reporting manner in the first gap, and sending the measurement result based on the traditional measurement reporting manner to the network device, the first gap being an idle duration between two second prediction durations. It can be understood that if the network device configures the predicted reporting time interval ΔT > Δt2, it means that the prediction capability of Model2 can be exceeded, and in this case, the AI model-based reporting manner is enabled in each second prediction duration, and the traditional measurement reporting manner is enabled between two second prediction durations Δt2.

[0025] In a possible implementation, if the AI model includes Model2, the configuration information can further include: a data type input into Model2 at the reporting occasion, the data type being a measurement value or a predicted value. It can be understood that by carrying the data type input into Model2 in the prediction information, the performance of the AI model can be conveniently monitored by the network device.

[0026] In a possible implementation, the method can further include: receiving first information from the network device, the first information indicating that the performance of Model2 is monitored, or the performance of Model1 and Model2 is monitored; in response to the first information, sending a monitoring result to the network device; receiving second information from the network device, the second information being determined based on the monitoring result. The second information indicates that the AI model is deactivated, or the AI model is retrained, or another AI model is reselected.

[0027] In a possible implementation, the first indication information can indicate that the performance of Model2 is monitored at the first occasion. Correspondingly, in response to the first information, sending the monitoring result to the network device can include: in response to the first information, monitoring the performance of Model2 at the first occasion, and sending a first monitoring result to the network device. The first monitoring result can include: a measurement value, an identifier of Model2, and an identifier of a first prediction characteristic corresponding to Model2.

[0028] In a possible implementation, the first indication information can indicate that performance monitoring is performed on the Model 1 and the Model 2 at the second time. Accordingly, in response to the first information, sending the monitoring result to the network device can include: in response to the first information, performing performance monitoring on the Model 1 and the Model 2 at the second time, and sending the second monitoring result to the network device. The second monitoring result can include: a measurement value, an identifier of the Model 1, an identifier of the Model 2, an identifier of the first predicted characteristic corresponding to the Model 2, and an identifier of the second predicted characteristic corresponding to the Model 1.

[0029] For example, the first information can include any of the following:

[0030] a start time point of performance monitoring of the AI model and a monitoring time interval;

[0031] indicating that performance monitoring of the AI model is triggered when a preset event occurs or a preset threshold is met;

[0032] a time point of performance monitoring of the AI model;

[0033] indicating that performance monitoring of the AI model is performed immediately.

[0034] In the above scheme, the network device can determine the management of each model based on the performance monitoring feedback of the terminal device, such as determining that the current AI model is no longer applicable, performing model deactivation, reselection, or reselecting a model. It can be understood that through monitoring of the AI model, it can be ensured that the AI model used for prediction is a suitable model, thereby bringing higher performance. In addition, performing management of the model by the network device can reduce the complexity of the terminal device.

[0035] In a possible implementation, the prediction information can further include a credibility parameter, and the credibility parameter is used to represent the credibility of the prediction information obtained based on the AI model. Accordingly, before sending the prediction information obtained based on the AI model to the network device at the reporting time, the method can further include: obtaining the credibility parameter by evaluating the credibility of the prediction information obtained based on the AI model this time.

[0036] In a possible implementation, the method can further include: in a case where the credibility parameter is less than or equal to a preset threshold, deactivating the AI model, or retraining the AI model, or reselecting another AI model. The preset threshold is configured by the network device.

[0037] In a possible implementation, the method can further include: sending the decision result of the AI model to the network device, and the decision result can include any one of the following: deactivating the AI model, or retraining the AI model, or reselecting another AI model.

[0038] In the above scheme, by monitoring the AI model on the terminal side, the process does not need to perform AI model performance feedback, thereby reducing signaling load.

[0039] In a second aspect, the present application provides a prediction information reporting method. The method can be applied to a network device. The method can include: sending configuration information to a terminal device, the configuration information can include a reporting opportunity; receiving prediction information from the terminal device at the reporting opportunity, the prediction information being a prediction result of an RLF or HOF event based on an AI model.

[0040] Through the above scheme, the network device configures the terminal device with a reporting opportunity for performing RLF or HOF event prediction, so that the terminal device can obtain a prediction result of an RLF or HOF event based on an AI model at the reporting opportunity, and report the prediction result of the RLF or HOF event to the network device. In this way, the occurrence of premature HO, late HO, or HO to a wrong cell can be reduced, the success rate of handover is improved, and the operation quality of the network is ensured.

[0041] In a possible implementation, the reporting opportunity includes: a prediction window, the AI model being activated within the prediction window; and a time point within the prediction window, the reporting of the prediction information being configured at the time point.

[0042] Exemplarily, the prediction window can include a start time point and an end time point of the prediction window, the start time point being a first preset frame or a first preset time slot, and the end time point being a second preset frame or a second preset time slot. The time point within the prediction window includes a third preset frame, or a third preset time slot, or a time interval for reporting the prediction information within the prediction window.

[0043] In a possible implementation, the reporting opportunity includes: a start opportunity for reporting the prediction information, the start opportunity being a time point at which the AI model is activated; and a time interval for reporting the prediction information.

[0044] In a possible implementation, before the configuration information is sent to the terminal device, the method further includes: receiving a prediction capability from the terminal device. The AI model can include Model 1 and Model 2, and the prediction capability can include at least one of the following: a first prediction duration, within which Model 1 supports obtaining a predicted value based on a measurement value; and an indication that a predicted value output by Model 1 is supported as an input of Model 2, and an output of Model 2 is a prediction result of an RLF or HOF event. Alternatively, the AI model can include Model 2, and the prediction capability can include a second prediction duration, within which Model 2 supports obtaining a prediction result of an RLF or HOF event based on a measurement value.

[0045] In a possible implementation, the method further includes: sending first information to the terminal device; and receiving a listening result from the terminal device. The first indication information indicates that performance of Model 2 is listened to at a first time; and the listening result can include a measurement value, an identifier of Model 2, and an identifier of a first prediction characteristic corresponding to Model 2. Alternatively, the first indication information indicates that performance of Model 1 and Model 2 is listened to at a second time; and the listening result can include a measurement value, an identifier of Model 1, an identifier of Model 2, an identifier of a first prediction characteristic corresponding to Model 2, and an identifier of a second prediction characteristic corresponding to Model 1.

[0046] In a possible implementation, the method further includes: receiving a decision result from the terminal device in a case where the credibility parameter is less than or equal to a preset threshold, the credibility parameter being used to represent a credibility degree of predicted information obtained based on the AI model. The decision result can include any one of the following: deactivating the AI model, or retraining the AI model, or reselecting another AI model.

[0047] In a third aspect, a communication apparatus is provided. The communication apparatus can include a processor, a communication interface, and a memory coupled to the processor and the communication interface. The memory stores instructions that, when executed by the processor, cause the communication apparatus to perform the prediction information reporting method of any one of the first aspect or the possible implementation of the first aspect.

[0048] In a fourth aspect, a terminal device is provided. The terminal device includes one or more processors and a memory. The memory is coupled to the one or more processors, and the memory is configured to store computer program codes including computer instructions. The one or more processors invoke the computer instructions to cause the terminal device to perform the method provided in the first aspect and any possible implementation of the first aspect.

[0049] In a fifth aspect, the present application provides a network device, comprising one or more processors, and a memory. The memory is coupled to the one or more processors, and the memory is configured to store computer program codes comprising computer instructions. The one or more processors invoke the computer instructions to cause the network device to perform the method according to any possible implementation of the second aspect.

[0050] In a sixth aspect, the present application provides a communication system, which can include a terminal device and a network device. The terminal device is configured to perform the method of predicting information reporting according to any one of the first aspect. The network device is configured to perform the method of predicting information reporting according to any one of the second aspect.

[0051] In a seventh aspect, the present application provides a computer readable storage medium storing a computer program. When the computer program is run on a terminal device, the terminal device performs the method of predicting information reporting according to any one of the first aspect. When the computer program is run on a network device, the network device performs the method of predicting information reporting according to any one of the second aspect.

[0052] In an eighth aspect, the present application provides a chip coupled to a memory. The chip is configured to read and execute a computer program stored in the memory, so as to implement the method of predicting information reporting according to any one of the first aspect or the second aspect.

[0053] In a ninth aspect, the present application provides a computer program product. When the computer program product is run on a computer, the computer performs the method of predicting information reporting according to any one of the first aspect or the second aspect.

[0054] It can be understood that the beneficial effects of the third aspect to the ninth aspect described above can be referred to the related description of the first aspect and the second aspect, which will not be repeated here. BRIEF DESCRIPTION OF DRAWINGS

[0055] FIG. 1 is a schematic diagram of five types of handover provided by the embodiments of the present application;

[0056] FIG. 2 is a flowchart of cross-gNB handover via Xn provided by the embodiments of the present application;

[0057] FIG. 3 is a flowchart of early HO provided by the embodiments of the present application;

[0058] FIG. 4 is a flowchart of late HO provided by the embodiments of the present application;

[0059] FIG. 5 is a flowchart of HO to an error cell provided by the embodiments of the present application;

[0060] FIG. 6 is a schematic diagram of a communication system according to an embodiment of the present application;

[0061] FIG. 7 is a schematic diagram of a hardware structure of a communication device according to an embodiment of the present application;

[0062] FIG. 8 is a schematic diagram of an AI model according to an embodiment of the present application;

[0063] FIG. 9 is a schematic diagram of a prediction information reporting method according to an embodiment of the present application;

[0064] FIG. 10 is a schematic diagram of another AI model according to an embodiment of the present application;

[0065] FIG. 11 is a schematic diagram of another AI model according to an embodiment of the present application;

[0066] FIG. 12 is a schematic diagram of a prediction window and an opportunity point in the prediction window according to an embodiment of the present application;

[0067] FIG. 13 is a schematic diagram of another prediction window and an opportunity point in the prediction window according to an embodiment of the present application;

[0068] FIG. 14 is a schematic diagram of another prediction window and an opportunity point in the prediction window according to an embodiment of the present application;

[0069] FIG. 15 is a schematic diagram of a starting opportunity and a time interval according to an embodiment of the present application;

[0070] FIG. 16 is a schematic diagram of determining Model2 input based on prediction capability according to an embodiment of the present application;

[0071] FIG. 17 is a schematic diagram of another determining Model2 input based on prediction capability according to an embodiment of the present application;

[0072] FIG. 18 is a schematic diagram of another determining Model2 input based on prediction capability according to an embodiment of the present application;

[0073] FIG. 19 is a schematic diagram of another determining Model2 input based on prediction capability according to an embodiment of the present application;

[0074] FIG. 20 is a schematic diagram of another determining Model2 input based on prediction capability according to an embodiment of the present application;

[0075] FIG. 21 is a schematic diagram of another determining Model2 input based on prediction capability according to an embodiment of the present application;

[0076] FIG. 22 is a schematic diagram of another determining Model2 input based on prediction capability according to an embodiment of the present application;

[0077] FIG. 23 is a schematic diagram of another method for determining the input of Model2 based on the prediction capability according to an embodiment of the present application;

[0078] FIG. 24 is a schematic diagram of a method for monitoring the performance of an AI model by a network device according to an embodiment of the present application;

[0079] FIG. 25 is a schematic diagram of a monitoring occasion and a monitoring model according to an embodiment of the present application;

[0080] FIG. 26 is a schematic diagram of a method for monitoring the performance of an AI model by a terminal device according to an embodiment of the present application. DETAILED DESCRIPTION

[0081] The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.

[0082] In a mobile communication network, when a terminal device is in a radio resource control (RRC) connected state, if the terminal device moves from one base station coverage area to another base station coverage area, or the call quality is reduced due to external interference, it needs to be transferred from the original channel to another idle channel. This process is called HO.

[0083] Taking a 5G network as an example, HO is needed when a terminal device roams or moves from one NR node (NR Node B, gNB) to another gNB. HO is crucial for both data and voice session connections. According to the routes involved in the handover, a 5G network can generally include five types of handover, which are: intra-gNB handover, cross-gNB handover via Xn, cross-gNB handover via N2, Inter-gNB-based cross-AMF N14 handover, and N26 handover based on cross-RAT.

[0084] Exemplarily, FIG. 1 shows a schematic diagram of the five types of handover.

[0085] As shown in FIG. 1, when a user equipment (UE) moves from one cell to another cell connected to the same gNB, the handover occurs within the gNB. Since the security termination point remains unchanged, there is no need to change the access stratum (AS) security algorithm during intra-gNB handover. If the UE does not receive an indication of a new AS security algorithm during intra-gNB handover, the UE can continue to use the same AS security algorithm as before.

[0086] When the UE is handed over between gNBs through the Xn interface from the source gNB to the target gNB, the handover is called Xn-based inter-gNB handover. Regarding security issues, the source gNB includes the UE security capabilities, such as the encryption and integrity algorithms used in the original cell, in the handover request message. The target gNB can select the highest priority algorithm from the received UE security capabilities according to the locally configured algorithms in priority.

[0087] When the gNB source and the target gNB do not have an active Xn interface or do not allow handover over the Xn interface, the gNB can decide to perform handover over the N2 interface, which is called N2-based inter-gNB handover. In this type of handover, the AMF plays the role of an anchor to coordinate the coordination between the source gNB and the target gNB to make the handover successful.

[0088] 3GPP defines the N14 interface to connect two access and mobility management functions (AMFs) belonging to two different operators or serving two public land mobile networks (PLMNs). When the source gNB and the target gNB are connected to different AMFs, the handover can be triggered through the N14 interface.

[0089] In addition, in order to support inter-radio access technology (RAT) mobility, 3GPP defines the N26 interface to connect the 5G AMF to the 4G mobility management entity (MME). When the operator's 5G is not fully covered and the coverage gap is filled by 4G coverage, the UE can perform handover from 5G to 4G, which is called N26-based inter-RAT handover.

[0090] Generally, different handover types correspond to different handover procedures. In order to facilitate the understanding of the handover procedure, the specific procedure of the Xn-based inter-gNB handover will be described below.

[0091] As shown in FIG. 2, the Xn-based inter-gNB handover procedure includes the following steps 1 to 13.

[0092] Step 1: The source gNodeB sends a measurement control message to the UE through an RRC reconfiguration (RRCReconfiguration) message, such as measurement object (intra-frequency / inter-frequency), measurement reporting configuration, and measurement gap configuration.

[0093] Step 2, the UE replies to the source gNodeB with an RRC Reconfiguration Complete (RRCReconfigurationComplete) message.

[0094] Step 3, the UE performs measurements according to the received measurement control message. After the UE measures and determines that an event condition is reached, it reports a measurement report to the source gNodeB.

[0095] Correspondingly, the source gNodeB makes a handover strategy and target cell / frequency decision based on the measurement results.

[0096] Step 4, the source gNodeB initiates a handover request (HandoverRequest) to the gNodeB where the selected target cell is located (i.e., the target gNodeB).

[0097] Correspondingly, after receiving the handover request, the target gNodeB performs admission control, allocates UE instances and transmission resources after allowing admission.

[0098] Step 5, the target gNodeB replies to the source gNodeB with a handover request response (HandoverRequestAcknowledge), allowing handover access. If some PDU Sessions fail to access, the message needs to carry a list of failed protocol data units (PDUs) sessions.

[0099] Step 6, the source gNodeB sends an RRC Reconfiguration (RRCReconfiguration) message to the UE, requiring the UE to perform handover to the target cell.

[0100] Step 7, the source gNodeB sends packet data convergence protocol (PDCP) sequence numbers (SNs) to the target gNodeB through sequence number status transfer (SN Status Transfer).

[0101] Step 8, the UE sends an RRC Reconfiguration Complete (RRCReconfigurationComplete) message to the target gNodeB, and the UE air interface handover to the target cell is completed.

[0102] Step 9, the target gNodeB sends a Path Switch Request message to the AMF to inform that the UE has changed cell, the message contains the target cell identity and the list of converted PDU Sessions. After receiving the message, the AMF updates the downlink GPRS tunneling protocol user plane (GTPU) to modify the GTPU address on the radio access network (RAN) side to the target gNodeB.

[0103] Step 10, the AMF replies to the target gNodeB with a Path Switch Request Acknowledge message. If the AMF indicates in the Path Switch Request Acknowledge message that the AMF failed to establish a PDU Session, the gNodeB deletes the failed PDU Session.

[0104] Step 11, the target gNodeB sends a UE Context Release message to the source gNodeB, and the source gNodeB releases the switched user.

[0105] Step 12, after switching to the target cell, the target gNodeB sends a measurement control message to the UE through an RRC Reconfiguration message.

[0106] Step 13, after receiving the new measurement control from the target gNodeB, the UE replies with an RRC Reconfiguration Complete message.

[0107] The above steps 1 to 13 only illustrate the cross-gNB handover process via Xn, and if an error occurs in a certain step of the process, an HOF event may occur. It can be understood that for other several types of handover, an HOF event may also occur during the handover process, thereby causing call drop and affecting the running quality of the network.

[0108] The following describes three common scenarios that cause HOF in combination with FIGS. 3 to 5.

[0109] Scenario 1: Early HO.

[0110] As shown in FIG. 3, after the UE is handed over from a source gNodeB (i.e. the original cell) to a target gNodeB (the target cell), if a radio link failure (RLF) occurs between the UE and the target gNodeB in a short time, an RRC re-establishment is triggered, i.e. cell selection is started, the result of the selection is to select the original cell, and the UE and the source gNodeB complete the RRC re-establishment. Since the RLF is caused by the source gNodeB prematurely instructing the handover to the target gNodeB, this process is referred to as premature HO.

[0111] In a self-organizing network, configuration optimization of handover parameters can be performed by reporting failures. For example, after the RRC re-establishment is completed, the source gNodeB sends an RLF indication to the target gNodeB to indicate that the UE has an RLF event in the target cell. The target gNodeB returns an HO report to the source gNodeB, which carries the type of HOF failure (e.g. premature HO). The source gNodeB can modify or optimize the handover parameters according to the type of HOF failure to reduce the probability of premature HO.

[0112] Scenario two: late HO.

[0113] As shown in FIG. 4, during the process of the UE being handed over from a source gNodeB (i.e. the original cell) to a target gNodeB (the target cell), if an RLF event occurs between the UE and the source gNodeB, an RRC re-establishment is triggered, i.e. cell selection is started, the result of the selection is to select the target cell, and the UE and the target gNodeB complete the RRC re-establishment. Since the RLF is caused by the source gNodeB late instructing the handover to the target gNodeB, this process is referred to as late HO.

[0114] In a self-organizing network, configuration optimization of handover parameters can be performed by reporting failures. For example, after the RRC re-establishment is completed, the target gNodeB sends an RLF indication to the source gNodeB to indicate that the HOF failure is caused by late HO. The source gNodeB can modify or optimize the handover parameters according to the type of HOF failure to reduce the probability of late HO.

[0115] Scenario three: HO to a wrong cell.

[0116] As shown in FIG. 5, because the source gNodeB (i.e., the original cell) determines an inappropriate target gNodeB (i.e., the second gNodeB) at the time of handover decision, after the UE switches from the source gNodeB to the second gNodeB, RLF occurs in a short time, the UE performs RRC re-establishment, i.e., performs cell selection to select another cell (i.e., the third gNodeB), and completes RRC re-establishment between the UE and the third gNodeB. This process is referred to as handover to an error cell.

[0117] In the self-organizing network, after completing the re-establishment, the third gNodeB can send an RLE indication to the second gNodeB to indicate that an RLF event occurs between the UE and the second gNodeB, and then the second gNodeB sends an HO report to the source gNodeB, which carries the type of HOF failure (such as HO to an error cell). The source gNodeB can modify or optimize the handover parameters according to the type of HOF failure to reduce the probability of HO to an error cell.

[0118] As can be seen from the above embodiments of the three scenarios, due to the inappropriate handover timing indicated by the network device to the terminal device, or the terminal device switching to an error target gNodeB, etc., RLF or HOF events may occur.

[0119] In order to reduce the probability of HOF occurrence, an embodiment of the present application provides a prediction information reporting method. In this method, the network device configures a reporting time for the terminal device to perform prediction reporting, and the terminal device can obtain a prediction result of the RLF or HOF event based on an artificial intelligence (AI) model at the reporting time, and report the prediction result of the RLF or HOF event to the network device, such as the prediction result can include a handover period, a recommended handover cell identifier, etc., so as to facilitate the network device to configure parameters according to the prediction result of the RLF or HOF event, and perform corresponding handover strategy. By using AI model to assist HO, the occurrence of early HO, late HO or HO to an error cell can be reduced, and the success rate of handover is improved. In addition, based on the AI model, the RLF or HOF event in a period of time can be predicted, and the terminal device does not need to perform measurement in this period of prediction time, thereby reducing the number of execution times of measurement actions.

[0120] The prediction information reporting method provided by the embodiments of the present application can be applied to the communication system shown in FIG. 6. The communication system can be a cellular communication system, or a universal mobile telecommunication system terrestrial radio access network (UTRAN) system, or an evolved universal mobile telecommunications system (UMTS) terrestrial radio access network (E-UTRAN) system, or a long term evolution (LTE) system, or a new radio (NR) system, or another mobile communication system such as a future mobile communication system, and the embodiments of the present application do not limit the communication system.

[0121] Exemplarily, FIG. 6 is a schematic diagram of a communication system provided by an embodiment of the present application.

[0122] As shown in FIG. 6, the communication system 00 can include a network device 01 and one or more terminal devices 02 (only one terminal device 02 is shown in FIG. 6) connected to the network device 01. The network device 01 and the terminal device 02 can perform uplink (UL) data and downlink (DL) data transmission.

[0123] The network device 01 is a device capable of communicating with the terminal device 02. In some embodiments, the network device 01 can be an access network device, which can also be referred to as a RAN device, and is a device providing wireless communication functions for the terminal device 02. For example, the access network device can be a base station, which can be a generation node B (gNB) in 5G, an evolved node B (eNB or eNodeB) in long term evolution (LTE), a radio network controller (RNC), a node B (NB), a base station controller (BSC), a base station transceiver station (BTS), a home base station (such as a home node B or a home evolved node B), a base unit (BBU), a transmitting and receiving point (TRP), a transmitting point (TP), a mobile switching center, and the like. The access network device can also be a radio controller in a cloud radio access network (CRAN) scenario, a centralized unit (CU), and / or a distributed unit (DU), or the network device can be a relay station, an access point, a vehicle-mounted device, a terminal device, a wearable device, and a device in future mobile networks, and the like.

[0124] The terminal device 02 is a device with wireless transceiving functions. The terminal device 02 can be a mobile terminal device or a non-mobile terminal device. For example, the terminal device 02 can be a UE, a user unit, a user station, a mobile station, a mobile station, a remote station, a remote terminal, a mobile device, a terminal, an access terminal, a user terminal, a wireless communication device, a user agent, or a user equipment. The access terminal can be a cellular phone, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (WLL), a personal digital assistant (PAD), a handheld device with wireless communication functions, a computer device, or other processing devices connected to a wireless modem, a vehicle-mounted device, a wearable device, a terminal device in a future evolved PLMN, and the like, which are not limited in the embodiments of the present application.

[0125] The network device 01 and the terminal device 02 can perform data transmission through a radio resource. The radio resource can include at least one of a time domain resource, a frequency domain resource, and a code domain resource. Specifically, when the network device 01 and the terminal device 02 perform data transmission, the network device 01 can send control information to the terminal device 02 through a control channel, such as a physical downlink control channel (PDCCH), so as to allocate a data channel, such as a physical downlink shared channel (PDSCH) or a physical uplink shared channel (PUSCH), to the terminal device 02.

[0126] The network device 01 or the terminal device 02 in FIG. 6 of the embodiment of the present application can be implemented by one device or one functional module in the device, and the embodiment of the present application does not make a specific limitation thereon. It can be understood that the above function can be a network element in a hardware device, a software function running on a special hardware, or a virtualized function instantiated on a platform, or a chip system. In the embodiment of the present application, the chip system can be composed of a chip or can include the chip and other discrete devices.

[0127] In a specific implementation, the network device 01 or the terminal device 02 shown in FIG. 6 can have the components shown in FIG. 7.

[0128] Exemplarily, FIG. 7 is a schematic diagram of a hardware structure of a communication apparatus provided by the embodiment of the present application.

[0129] As shown in FIG. 7, the communication apparatus 700 can include a processor 701, a communication line 702, and at least one communication interface 703. Further, the communication apparatus 700 can also include a memory 704. The processor 701, the memory 704, and the communication interface 703 can be connected through the communication line 702. In the embodiment of the present application, the at least one can be one, two, three, or more.

[0130] The processor 701 can be a central processing unit (CPU), a general purpose processor, a network processor (NP), a digital signal processor (DSP), a microprocessor, a microcontroller, a programmable logic device (PLD), or any combination thereof. The processor can also be other devices with processing function, such as circuit, device, or software module, etc.

[0131] The communication line 702 can include a path for transmitting information between components included in the communication device.

[0132] The communication interface 703 can be used for communication with other devices or communication networks (such as Ethernet, RAN, wireless local area networks (WLAN), etc.). The communication interface 703 can be a module, a circuit, a transceiver, or any device capable of realizing communication.

[0133] The memory 704 can be a read-only memory (ROM) or other type of static storage device that can store static information and / or instructions, or a random access memory (RAM), or other type of dynamic storage device that can store information and / or instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disk storage, a magnetic disk storage or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and that can be accessed by a computer.

[0134] In a possible design, the memory 704 can exist independently of the processor 701, that is, the memory 704 can be an external memory of the processor 701, and the memory 704 can be connected to the processor 701 through the communication line 702, for storing instructions or program codes. When the processor 701 invokes and executes the instructions or program codes stored in the memory 704, the prediction information reporting method provided in the embodiments of the present application can be implemented. In another possible design, the memory 704 can also be integrated with the processor 701, that is, the memory 704 can be an internal memory of the processor 701, for example, the memory 704 can be a cache, and can be used for temporarily storing some data and / or instruction information, etc.

[0135] As a possible implementation, the processor 701 can include one or more CPUs, for example, the CPU 0 and the CPU 1 in FIG. 7. As another possible implementation, the communication apparatus 700 can include multiple processors, for example, the processor 701 and the processor 707 in FIG. 7. As still another possible implementation, the communication apparatus 700 can further include the output device 705 and the input device 706. For example, the input device 706 can be a keyboard, a mouse, a microphone or a joystick, and the output device 705 can be a display screen or a speaker.

[0136] It should be noted that the communication apparatus 700 can be a general-purpose device or a special-purpose device. For example, the communication apparatus 700 can be a network server, a base station, a desktop computer, a laptop computer, a mobile phone, a tablet computer, a wireless terminal, an embedded device, a chip system or a device having a similar structure to that in FIG. 7. The embodiments of the present application do not limit the type of the communication apparatus 700.

[0137] For example, FIG. 8 is a schematic diagram of an AI model provided in the embodiments of the present application.

[0138] As shown in FIG. 8, the AI model can include a data collection module, an inference module, a management module, a model training module, and a model storage module. Taking the AI model used for positioning as an example, first, the AI model is established based on the training data collected by the data collection module, and the training process of the model is performed according to the use of the model, and a large number of positioning processes are required to complete the training of the model. During the use of the AI model, the AI model may no longer be suitable due to some factors, such as changes in the scene, and the management of the AI model based on the monitoring data collected by the data collection module is required, which includes the selection of the AI model, the activation, deactivation, and fallback (i.e., using a traditional method to realize the corresponding function) of the AI model, and the like. The inference stage of the AI model, that is, the corresponding function is realized based on the inference data collected by the data collection module, such as completing positioning. In addition, the model storage module can be used to store the trained / updated AI model, and in response to the model delivery / delivery request of the management module, the AI model is delivered / delivered to the inference module.

[0139] The prediction information reporting method provided by the embodiments of the present application will be described below in combination with the system shown in FIG. 6, the device shown in FIG. 7, and the AI model shown in FIG. 8.

[0140] FIG. 9 is a flowchart of a prediction information reporting method provided by an embodiment of the present application.

[0141] The method can be applied to the scenario of predicting the RLF or HOF event by the AI model stored in the terminal device before the network device sends the terminal device a handover indication, so as to facilitate the network device to configure or optimize the parameters according to the prediction result of the RLF or HOF event. As shown in FIG. 9, the method can include the following S11-S16.

[0142] S11, the terminal device reports the prediction capability of the terminal device to the network device.

[0143] Exemplarily, the terminal device can send the prediction capability of the terminal device to the network device through UE assistance information (UAI) signaling, RRC signaling, or other proprietary signaling. Correspondingly, the network device obtains the prediction capability of the terminal device by receiving the signaling.

[0144] In the embodiments of the present application, the AI model is stored in the terminal device. The AI model can take real measurement values of the terminal device as input, and output a prediction result of an RLF or HOF event, that is, the AI model can predict whether an RLF or HOF event will occur in a future period of time based on real measurement values, and formulate a strategy to reduce the occurrence of RLF or HOF events.

[0145] In some embodiments, as shown in FIG. 10, the AI model can include a radio resource management (RRM) prediction model, and an RLF or HOF prediction model. The RRM prediction model is also referred to as Model 1, which is used to predict measurement results in a future period of time based on current measurement results. The RLF or HOF prediction model is also referred to as Model 2, which is used to take current measurement results or measurement values as input, or take predicted results or predicted values output by the RRM prediction model as input, to predict RLF or HOF events that can occur in a future period of time. It can be understood that by using the predicted results or predicted values output by Model 1 as input of Model 2, the number of execution times of measurement actions can be reduced, and more opportunities for energy saving and service transmission of the terminal device can be brought.

[0146] In other embodiments, as shown in FIG. 11, the AI model can only include the RLF or HOF prediction model, that is, Model 2. It can be understood that compared with the AI model shown in FIG. 10, since the AI model shown in FIG. 11 does not set Model 1 and only sets Model 2, Model 2 can only take current measurement results or measurement values as input to predict RLF or HOF events that can occur in a future period of time, so that the terminal device can need to perform a larger number of measurement actions.

[0147] The prediction capability reported by the terminal device to the network device can specifically refer to the capability of the AI model to perform RLF or HOF event prediction. When the AI model includes different prediction models, the prediction capability can include different capability items.

[0148] Three implementation manners of the capability items are provided below.

[0149] In a first possible implementation manner, as shown in FIG. 10, if the AI model includes Model 1 and Model 2, the prediction capability reported by the terminal device to the network device can include at least one of the following:

[0150] ① A first prediction time length Δt1, also referred to as a time length of Model 1 measurement prediction. In the first prediction time length Δt1, Model 1 supports obtaining predicted values based on current measurement values, that is, Model 1 can predict measurement results in a future Δt1 period of time.

[0151] ② indicates that the predicted value output by Model1 is supported as the input of Model2.

[0152] As an example, the prediction capability reported by the terminal device to the network device only includes the first prediction duration Δt1, in which case the prediction capability implicitly indicates that the predicted value output by Model1 is supported as the input of Model2.

[0153] In a second possible implementation, as shown in FIG. 11, if the AI model only includes Model2, the prediction capability reported by the terminal device to the network device can include: the second prediction duration Δt2, also referred to as the duration or time interval of Model2 prediction. Within the second prediction duration Δt2, Model2 supports obtaining the prediction result of the RLF or HOF event based on the current measurement value, that is, Model2 can predict the RLF or HOF event within the future Δt2 time.

[0154] In a third possible implementation, as shown in FIG. 10, if the AI model includes Model1 and Model2, the prediction capability reported by the terminal device to the network device can at least include: the second prediction duration Δt2, also referred to as the duration or time interval of Model2 prediction. Within the second prediction duration Δt2, Model2 supports obtaining the prediction result of the RLF or HOF event based on the current measurement value or predicted value, that is, Model2 can predict the RLF or HOF event within the future Δt2 time. Wherein, if the input of Model2 is the predicted value, the predicted value is obtained based on the measurement value input into Model1.

[0155] Further, for the third possible implementation described above, the prediction capability reported by the terminal device to the network device can also include: the first prediction duration Δt1, also referred to as the duration of Model1 measurement prediction. Within the first prediction duration Δt1, Model1 supports obtaining the predicted value based on the measurement value, that is, Model1 can predict the measurement result within the future Δt1 time.

[0156] It should be noted that the first prediction duration Δt1 and the second prediction duration Δt2 described above can be equal or not equal. For details, please refer to the specific description of the following embodiments, which will not be described here. It can be understood that whether the terminal device reports the first prediction duration Δt1 or the second prediction duration Δt2 to the network device, it is convenient for the network device to configure the reporting occasion based on the prediction duration.

[0157] S12, the network device configures a reporting occasion for the terminal device based on the prediction capability of the terminal device.

[0158] The reporting occasion refers to an occasion at which the terminal device reports prediction information or a result obtained based on an AI model to a network device.

[0159] In some embodiments, the reporting occasion configured by the network device for the terminal device can include:

[0160] ① a prediction window.

[0161] In the prediction window, the terminal device activates or enables the AI model, and the AI model performs a prediction function. At a time outside the prediction window, such as a time gap between two prediction windows, the terminal device enables a legacy measurement reporting mode, and reports a measurement result based on the legacy measurement reporting mode to the network device.

[0162] As an example, the prediction window can include a start time point and an end time point of the prediction window. The start time point is a first preset frame, a first preset subframe, or a first preset time slot, and the end time point is a second preset frame, a second preset subframe, or a second preset time slot.

[0163] As another example, the prediction window can include a start time point and a time length of the prediction window. The start time point is a first preset frame, a first subframe, or a first preset time slot, and the time length of the prediction window is m preset frames, n subframes, p preset time slots, or a first time length, where m, n, and p are positive integers.

[0164] ② an occasion point within the prediction window.

[0165] The occasion point is a time point or time at which prediction results or information are reported within the prediction window. The number of occasion points within one prediction window can be one or more, and the terminal device performs prediction information reporting once at each occasion point.

[0166] By way of example, the occasion point within the prediction window can include a third preset frame, a third preset subframe, a third preset time slot, or a time interval at which prediction information is reported within the prediction window.

[0167] Taking the communication system of the present application as an example of 5G NR, the prediction window and the occasion point within the prediction window are described by way of example.

[0168] In 5G NR, the definitions of frame, subframe, and time slot are as follows:

[0169] Frame. The length of one frame is fixed at 10 milliseconds (ms), and contains 10 subframes.

[0170] Subframe. The length of each subframe is fixed at 1 ms, and one subframe includes a plurality of time slots.

[0171] Slot. The number and length of slots will vary under different configurations of data transmission and frequency bandwidth, and common slot lengths are 0.5 ms, 0.25 ms or shorter. In addition, a certain number of orthogonal frequency division multiplexing (OFDM) symbols are included in a slot.

[0172] Exemplarily, FIG. 12 is a schematic diagram of a prediction window and an opportunity point in the prediction window provided by an embodiment of the present application. As shown in FIG. 12, the starting time point of the prediction window is frame K, and the ending time point of the prediction window is frame K+i. Wherein, the length of each frame is 10 ms. In the prediction window, the opportunities for the network device to perform prediction reporting are frame K, frame K+3, frame K+6, …, frame K+i in sequence. That is, the time interval for performing prediction reporting in the prediction window is 3 frames.

[0173] Exemplarily, FIG. 13 is another schematic diagram of a prediction window and an opportunity point in the prediction window provided by an embodiment of the present application. As shown in FIG. 13, the starting time point of the prediction window is frame K, and the ending time point of the prediction window is frame K+i. Wherein, the length of each frame is 10 ms, and one frame includes 10 subframes. In the prediction window, the opportunities for the network device to perform prediction reporting are subframe 5 of frame K, subframe 10 of frame K, subframe 5 of frame K+1, subframe 10 of frame K+1, …, subframe 5 of frame K+i, subframe 10 of frame K+i in sequence. That is, the time interval for performing prediction reporting in the prediction window is 5 subframes.

[0174] Exemplarily, FIG. 14 is a schematic diagram of a prediction window and an opportunity point in the prediction window provided by an embodiment of the present application. As shown in FIG. 14, the starting time point of the prediction window is frame K, and the ending time point of the prediction window is frame K+i. Wherein, the length of each frame is 10 ms, one frame includes 10 subframes, and one subframe includes 2 slots. In the prediction window, the opportunities for the network device to perform prediction reporting are slot 1 of subframe 1 of frame K, slot 1 of subframe 6 of frame K, slot 1 of subframe 1 of frame K+1, slot 1 of subframe 6 of frame K+1, …, slot 1 of subframe 1 of frame K+i, slot 1 of subframe 6 of frame K+i in sequence. That is, the time interval for performing prediction reporting in the prediction window is 10 slots.

[0175] It can be understood that the network device configures the opportunity points in the prediction window, so that the terminal device can perform prediction information reporting at the opportunity points in the prediction window.

[0176] It should be noted that FIGS. 12 to 14 are described by taking the fixed interval ΔT of the timing point of reporting the prediction information as an example, which does not limit the present application. In actual implementation, the interval between any two adjacent timing points can be equal or unequal, and can be configured according to actual use requirements.

[0177] In some other embodiments, the reporting timing configured by the network device for the terminal device can include:

[0178] ① The start timing of reporting the prediction information.

[0179] The network device indicates that the activation of the AI model (such as Model1) is the start timing of prediction reporting, that is, the start timing of reporting the prediction information is the time point of activating the AI model.

[0180] ② The time interval ΔT of reporting the prediction information.

[0181] Further, the reporting timing configured by the network device for the terminal device can further include the duration of the prediction window. Alternatively, the network device indicates that the deactivation of the AI model is the end time of prediction reporting.

[0182] Exemplarily, FIG. 15 is a schematic diagram of a start timing and a time interval provided by an embodiment of the present application. As shown in FIG. 15, the terminal device can take the activation of the AI model (such as Model1) as the start timing of prediction reporting, and perform prediction reporting once every interval ΔT, for example, at time t1, time t2, time t3, …, time ti, the terminal device can obtain the prediction result based on the AI model and report the prediction result to the network device.

[0183] It can be understood that the network device configures the activation of the AI model (such as Model1) as the start timing of prediction reporting, and the time interval of reporting the prediction information, so that the terminal device can perform the reporting of the prediction information immediately after activating the AI model.

[0184] In some other embodiments, if the AI model includes Model2, the configuration information can further include: the data type input by Model2 at the reporting timing, which can be a measurement value or a prediction value. It can be understood that by carrying the data type input by Model2 in the prediction information, the network device can facilitate monitoring the performance of the AI model.

[0185] It should be noted that the embodiments of the present application are described by taking the network device configuring the reporting timing for the terminal device based on the prediction capability of the terminal device (that is, performing S11 and S12 first, and then performing S13) as an example, which does not limit the present application. In actual implementation, the network device can also directly configure the reporting timing for the terminal device.

[0186] S13, the network device sends the configuration information.

[0187] Correspondingly, the terminal device receives the configuration information from the network device.

[0188] The configuration information can include a reporting occasion configured by the network device for the terminal device. For implementation of the network device configuring the reporting occasion, refer to the specific description of embodiment S12 above, which will not be repeated here.

[0189] In some embodiments, the configuration information can be included in broadcast signaling, or can be included in control signaling or other signaling, which can be set according to actual use requirements, and the embodiments of the present application are not limited.

[0190] Exemplarily, the network device can use a semi-static configuration manner to send RRC signaling to the terminal device, and the RRC signaling carries the configuration information. Of course, the network device can also use other ways to configure information for the terminal device, such as by sending a medium access control (MAC) control element (CE) carrying the configuration information, a system message, a physical layer signaling, a downlink control information (DCI) or a paging message manner, and the embodiments of the present application are not limited.

[0191] S14, the terminal device inputs the measurement value into the AI model to obtain prediction information at the reporting occasion configured by the network device.

[0192] The prediction information can include a prediction result of the RLF or HOF event.

[0193] Specifically, the terminal device can perform the following operations at each reporting occasion configured by the network device: inputting the measurement value into the AI model to obtain prediction information, and sending the prediction information corresponding to the reporting occasion to the network device. Exemplarily, the measurement value includes at least one of the following: a layer one measurement result (L1 measurement result) and a layer three measurement result (L3 measurement result). It can be understood that at each reporting occasion, if different measurement values are input into the AI model, the AI model can output different prediction results of the RLF or HOF event based on different measurement values.

[0194] Referring to the description of S11 above, the AI model can include Model1 and Model2, or at least Model2.

[0195] The reporting opportunities include at least a first reporting opportunity and a second reporting opportunity. The second reporting opportunity is the last reporting opportunity of the first reporting opportunity. If the AI model includes Model1 and Model2, then the step of inputting the measurement value into the AI model to obtain the prediction information at the reporting opportunity configured by the network device can include the following three cases.

[0196] Case 1: If the time interval between the first reporting opportunity and the second reporting opportunity is less than or equal to the first prediction time length Δt1, and the terminal device obtains the first measurement value by performing measurement at the second reporting opportunity, then at the first reporting opportunity, the first measurement value is input into Model1 to obtain the first prediction value, the first prediction value is input into Model2 to obtain the first prediction result of the RLF or HOF event, and the first prediction information is sent to the network device. The first prediction information can include the first prediction result.

[0197] Case 2: If the time interval between the first reporting opportunity and the second reporting opportunity is less than or equal to the first prediction time length Δt1, and the terminal device does not perform measurement at the second reporting opportunity, then at the first reporting opportunity, the second measurement value is obtained by performing measurement, the second measurement value is input into Model2 to obtain the second prediction result of the RLF or HOF event, and the second prediction information is sent to the network device. The second prediction information can include the second prediction result.

[0198] Case 3: If the time interval between the first reporting opportunity and the second reporting opportunity is greater than the first prediction time length Δt1, then at the first reporting opportunity, the second measurement value is obtained by performing measurement, the second measurement value is input into Model2 to obtain the second prediction result of the RLF or HOF event, and the second prediction information is sent to the network device. The second prediction information can include the second prediction result.

[0199] Further, the first prediction information further includes that the data type input into Model2 is a prediction value, and the second prediction information further includes that the data type input into Model2 is a measurement value. It can be understood that by carrying the data type input into Model2 in the prediction information, the performance of the AI model can be monitored by the network device.

[0200] In the above scheme, if there is a measurement value less than or equal to the first prediction duration Δt1 from the current prediction reporting opportunity, it means that it is within the prediction capability of Model 1, so Model 1 can use the measurement value to predict the measurement value of the current opportunity (i.e. the predicted value), and use it as the input of Model 2; if there is no measurement value less than or equal to the first prediction duration Δt1 from the current prediction reporting opportunity, it means that although it is within the prediction capability of Model 1, there is no measurement value available for input to Model 1, so it is necessary to perform measurement at the current reporting opportunity, and use the measurement result as the input of Model 2; if the time interval between the first reporting opportunity and the second reporting opportunity is greater than the first prediction duration Δt1, it means that it has exceeded the prediction capability of Model 1, so it is also necessary to perform measurement at the current reporting opportunity, and use the measurement result as the input of Model 2.

[0201] The reporting opportunity includes at least a third reporting opportunity and a fourth reporting opportunity, and the fourth reporting opportunity is the last reporting opportunity of the third reporting opportunity. If the AI model includes at least Model 2, then "inputting the measurement value into the AI model at the reporting opportunity configured by the network device to obtain the prediction information" can specifically include the following three cases:

[0202] Case 1, if the time interval between the third reporting opportunity and the fourth reporting opportunity is less than or equal to the second prediction duration Δt2, then at the third reporting opportunity, input the measurement value or the predicted value into Model 2 to obtain the third prediction result of the RLF or HOF event, and send the third prediction information to the network device, the third prediction information including the third prediction result. Further, the third prediction information can also include the data type input into Model 2, which is the measurement value or the predicted value.

[0203] For example, if the AI model only includes Model 2, then at the third reporting opportunity, perform measurement to obtain a third measurement value, and input the third measurement value into Model 2 to obtain a third prediction result.

[0204] For example, if the AI model includes Model 1 and Model 2, the following steps are performed:

[0205] If the time interval between the third reporting opportunity and the fourth reporting opportunity is greater than the first prediction duration, then at the third reporting opportunity, perform measurement to obtain a third measurement value, and input the third measurement value into Model 2 to obtain a third prediction result.

[0206] Or, if the time interval between the third reporting opportunity and the fourth reporting opportunity is less than or equal to the first prediction duration, and no measurement is performed at the fourth reporting opportunity, at the third reporting opportunity, a measurement is performed to obtain a third measurement value, and the third measurement value is input into Model2 to obtain a third prediction result.

[0207] Or, if the time interval between the third reporting opportunity and the fourth reporting opportunity is less than or equal to the first prediction duration, and a fourth measurement value is obtained by performing measurement at the fourth reporting opportunity, at the third reporting opportunity, the fourth measurement value is input into Model1 to obtain a second prediction value, and the second prediction value is input into Model2 to obtain a third prediction result.

[0208] Case 2, if the time interval between the third reporting opportunity and the fourth reporting opportunity is greater than the second prediction duration Δt2, at the third reporting opportunity, a fifth measurement value is obtained by performing measurement, and the fifth measurement value is input into Model2 to obtain a fourth prediction result, and fourth prediction information including the fourth prediction result is sent to the network device.

[0209] Case 3, if the time interval between the third reporting opportunity and the fourth reporting opportunity is greater than the second prediction duration Δt2, a traditional measurement reporting mode is enabled in the first gap, and a measurement result based on the traditional measurement reporting mode is sent to the network device. Wherein, the first gap is an idle duration between two second prediction durations.

[0210] In the above scheme, if the time interval ΔT of the prediction reporting configured by the network device is less than or equal to Δt2, it means that each reporting opportunity is within the prediction capability range of Model2, so Model2 can obtain the prediction result of the RLF or HOF event based on the measurement value or the prediction value, that is, the reporting mode based on the AI model is enabled, and the traditional measurement reporting mode is implicitly disabled. If the time interval ΔT of the prediction reporting configured by the network device is greater than Δt2, it means that it may exceed the prediction capability range of Model2, in which case the reporting mode based on the AI model is enabled within each second prediction duration, and the traditional measurement reporting mode is enabled between two second prediction durations Δt2.

[0211] S15, the terminal device reports the prediction information to the network device.

[0212] S16, the network device configures or optimizes the handover parameter based on the prediction information.

[0213] Referring to the description of S14 above, the prediction information can include the prediction result of the RLF or HOF event.

[0214] Further, the prediction information can also include the data type input into Model2, which is a measurement value or a prediction value.

[0215] Correspondingly, the switching parameter can be a relevant parameter associated with the HO.

[0216] Exemplarily, the prediction result of the RLF or HOF event can include at least one of the following:

[0217] ① A first time period, i.e., a switching time period. The terminal device can predict a time period (referred to as a second time period) in which the RLF or HOF event is more likely to occur or a time period (referred to as a first time period) in which the RLF or HOF event is less likely to occur based on the Model2, and carry the first time period in the prediction information. The first time period is different from the second time period. In this way, the network device can set the time for issuing the HO instruction to the terminal device within the first time period based on the prediction information, so as to reduce the probability of occurrence of early HO and late HO.

[0218] ② An identifier of the first cell, i.e., a cell identifier of the recommended switching cell. The terminal device can predict a target cell (referred to as a second cell) in which the RLF or HOF event is more likely to occur or a target cell (referred to as a first cell) in which the RLF or HOF event is less likely to occur based on the Model2, and carry the identifier of the first cell in the prediction information. The first cell is different from the second cell. In this way, the network device can set the first cell as the target cell to be switched based on the prediction information, so as to reduce the probability of occurrence of HO to an error cell.

[0219] In the above prediction information reporting method provided in the application, the network device configures a reporting occasion for the terminal device to perform prediction reporting. The terminal device can obtain the prediction result of the RLF or HOF event based on the AI model at the reporting occasion, and report the prediction result of the RLF or HOF event to the network device. The prediction result can include a switching time period, a recommended cell identifier to be switched, etc., so as to facilitate the network device to configure parameters according to the prediction result of the RLF or HOF event and perform a corresponding switching strategy. Through AI model assisted HO, the occurrence of early HO, late HO or HO to an error cell can be reduced, the success rate of switching is improved, and the running quality of the network is ensured. In addition, the RLF or HOF event in a period of time can be predicted based on the AI model. The terminal device does not need to perform measurement within the prediction time period, and the number of execution times of the measurement action is reduced.

[0220] In order to facilitate understanding of the above prediction information reporting method, the following takes the activation of the AI model (such as Model1) as the starting time of prediction reporting, and the time interval of prediction reporting is ΔT as an example, and combines FIG. 16 to FIG. 23 to exemplarily illustrate the determination of the Model2 input based on the prediction capability of the AI model.

[0221] Example 1

[0222] As shown in FIG. 16, the AI model includes Model 1 and Model 2, and the reporting occasions are t1, t2, t3, t4, t5, t6, …, wherein the time interval between the first reporting occasion and the second reporting occasion is ΔT, and Model 1 supports obtaining a predicted value based on a measurement value within the first prediction duration Δt1. The second reporting occasion is the last reporting occasion of the first reporting occasion, for example, the first reporting occasion is t2, and the second reporting occasion is t1. Since the first prediction duration Δt1 starting from the second reporting occasion is within the capability range of Model 1, and the time interval ΔT between the first reporting occasion and the second reporting occasion is less than the first prediction duration Δt1, it indicates that the first reporting occasion is within the capability range of Model 1, and therefore the terminal device can determine whether the input type of Model 1 at the first reporting occasion is a measurement value or a predicted value based on whether a measurement value is obtained by performing measurement at the second reporting occasion.

[0223] For the reporting occasion t1, since the reporting occasion t1 is the first reporting occasion, there is no measurement value obtained within the time range less than or equal to Δt from the reporting occasion t1, and therefore the terminal device performs measurement at the reporting occasion t1 to obtain a measurement value, inputs this measurement value into Model 2 to obtain a predicted result of the RLF or HOF event, and sends the predicted information including the predicted result to the network device.

[0224] For the reporting occasion t2, since the time interval between the reporting occasion t2 and the reporting occasion t1 is less than the first prediction duration Δt1, and the terminal device performs measurement at the reporting occasion t1 to obtain a measurement value, at the reporting occasion t2, the terminal device can input the measurement value obtained at the reporting occasion t1 into Model 1 to obtain a predicted value, input this predicted value into Model 2 to obtain a predicted result of the RLF or HOF event, and send the predicted information including the predicted result to the network device.

[0225] For the reporting occasion t3, although the time interval between the reporting occasion t3 and the reporting occasion t2 is less than the first prediction duration Δt1, the terminal device does not actually perform measurement at the reporting occasion t2, and therefore the terminal device performs measurement at the reporting occasion t3 to obtain a measurement value, inputs this measurement value into Model 2 to obtain a predicted result of the RLF or HOF event, and sends the predicted information including the predicted result to the network device.

[0226] Similarly, the terminal device performs measurement at the reporting time t1, the reporting time t3, the reporting time t5, and the like, and takes the measurement result / measurement value as the input of Model 2. The terminal device inputs the measurement result / measurement value of the last reporting period into Model 1 to obtain a predicted value at the reporting time t2, the reporting time t4, the reporting time t6, and the like, and takes the predicted value as the input of Model 2. It can be understood that in Example 1, the network device configures the time interval ΔT of the prediction report based on the first prediction time Δt1 reported by the terminal device, ΔT≤Δt1, so that the terminal device can obtain the prediction result of the RLF or HOF event based on the measurement value or the predicted value at each reporting time. Since the terminal device does not need to perform measurement at the reporting time t2, the reporting time t4, and the reporting time t6, the terminal device has more opportunities for energy saving and service transmission in this prediction period.

[0227] Example 2

[0228] As shown in FIG. 17, the AI model includes Model 1 and Model 2, and the reporting times are t1, t2, t3, t4, t5, t6, and the like. The time interval between the first reporting time and the second reporting time is ΔT, and Model 1 supports obtaining a predicted value based on a measurement value within the first prediction time Δt1. The second reporting time is the last reporting time of the first reporting time, for example, the first reporting time is t2, and the second reporting time is t1. Since the first prediction time Δt1 starting from the second reporting time is within the capability range of Model 1, and the time interval ΔT between the first reporting time and the second reporting time is greater than the first prediction time Δt1, it indicates that the first reporting time is not within the capability range of Model 1, so the terminal device needs to perform measurement at the first reporting time, and takes the measurement value as the input of Model 2.

[0229] For the reporting time t1, since the reporting time t1 is the first reporting time, there is no measurement value obtained within the time range less than or equal to Δt from the reporting time t1, so the terminal device performs measurement at the reporting time t1 to obtain a measurement value, and takes the measurement value as the input of Model 2 to obtain the prediction result of the RLF or HOF event, and sends the prediction information including the prediction result to the network device.

[0230] For the reporting time t2, since the reporting time t2 is not within the capability range of Model 1, the terminal device needs to perform measurement at the reporting time t2, and takes the measurement value as the input of Model 2.

[0231] By analogy, the terminal device performs measurement at the reporting time t1, the reporting time t2, the reporting time t3, the reporting time t4, the reporting time t5, the reporting time t6, and the like, and takes the measurement result / measurement value as the input of Model2. It can be understood that in Example 2, the network device configures the time interval ΔT of the prediction report based on the first prediction time Δt1 reported by the terminal device, ΔT > Δt1, so that the terminal device can only obtain the prediction result of the RLF or HOF event based on the measurement value at each reporting time.

[0232] Example 3:

[0233] As shown in FIG. 18, the AI model only includes Model2, and the reporting times are t1, t2, t3, t4, t5, t6, and the like, wherein the time interval between the third reporting time and the fourth reporting time is ΔT, and Model2 supports obtaining the prediction result of the RLF or HOF event based on the measurement value within the second prediction time Δt2. The fourth reporting time is the last reporting time of the third reporting time, for example, the third reporting time is t2, and the fourth reporting time is t1. Since the second prediction time Δt2 starting from the fourth reporting time is within the capability range of Model2, and the time interval ΔT between the third reporting time and the fourth reporting time is less than the second prediction time Δt2, it indicates that the third reporting time is within the capability range of Model2, so that the terminal device can obtain the prediction result of the RLF or HOF event based on Model2 at the third reporting time.

[0234] Specifically, for the reporting time t1, the reporting time t2, the reporting time t3, the reporting time t4, the reporting time t5, the reporting time t6, and the like, since the AI model only includes Model2 and does not include Model1, the terminal device needs to perform measurement at each reporting time, and input the measurement value into Model2 to obtain the prediction result of the RLF or HOF event, and send the prediction information including the prediction result to the network device.

[0235] It can be understood that in Example 3, the network device configures the time interval ΔT of the prediction report based on the second prediction time Δt2 reported by the terminal device, ΔT ≤ Δt2, so that the terminal device can obtain the prediction result of the RLF or HOF event based on the measurement value at each reporting time.

[0236] Example 4:

[0237] As shown in FIG. 19, the AI model only includes Model2, and the reporting time points are t1, t2, t3, t4, t5, t6, and the like. The time interval between the third reporting time point and the fourth reporting time point is ΔT, and Model2 supports obtaining the prediction result of the RLF or HOF event based on the measurement value within the second prediction time length Δt2. The fourth reporting time point is the previous reporting time point of the third reporting time point, for example, the third reporting time point is t2, and the fourth reporting time point is t1. Since the second prediction time length Δt2 starting from the fourth reporting time point is within the capability range of Model2, and the time interval ΔT between the third reporting time point and the fourth reporting time point is greater than the second prediction time length Δt2, it is indicated that the third reporting time point is not within the capability range of Model2, and therefore, at the third reporting time point, the terminal device can obtain the prediction result of the RLF or HOF event based on Model2.

[0238] Specifically, for the reporting time point t1, the reporting time point t2, the reporting time point t3, the reporting time point t4, the reporting time point t5, the reporting time point t6, and the like, since the AI model only includes Model2 and does not include Model1, the terminal device needs to perform measurement at each reporting time point, input the measurement value into Model2 to obtain the prediction result of the RLF or HOF event, and send the prediction information including the prediction result to the network device. In addition, for the gap between the two second prediction time lengths Δt2, the terminal device can enable the traditional measurement reporting mode, and send the measurement result based on the traditional measurement reporting mode to the network device.

[0239] It can be understood that in Example 4, the network device configures the time interval ΔT of the prediction reporting based on the second prediction time length Δt2 reported by the terminal device, and ΔT>Δt2, so that the terminal device can obtain the prediction result of the RLF or HOF event based on the measurement value within the second prediction time length Δt2, and enable the traditional measurement reporting mode between the two second prediction time lengths Δt2.

[0240] Example 5:

[0241] As shown in FIG. 20, the AI model includes Model 1 and Model 2, and the reporting occasions are t1, t2, t3, t4, t5, t6, …, wherein the time interval between the third reporting occasion and the fourth reporting occasion is ΔT, Model 1 supports obtaining a predicted value based on a measurement value within the first prediction duration Δt1, and Model 2 supports obtaining a predicted result of an RLF or HOF event based on a measurement value within the second prediction duration Δt2. The fourth reporting occasion is the last reporting occasion of the third reporting occasion, for example, the third reporting occasion is t2, and the fourth reporting occasion is t1. Since the second prediction duration Δt2 starting from the fourth reporting occasion is within the capability range of Model 2, and the time interval ΔT between the third reporting occasion and the fourth reporting occasion is less than the second prediction duration Δt2, it is indicated that the third reporting occasion is within the capability range of Model 2. Therefore, at the third reporting occasion, the terminal device can further determine, based on the size relationship between the time interval ΔT and the first prediction duration Δt1, that the measurement value or the predicted value is taken as the input of Model 2 to obtain the predicted result of the RLF or HOF event.

[0242] For the reporting occasion t1, since the reporting occasion t1 is the first reporting occasion, there is no measurement value obtained within the time range less than or equal to Δt from the reporting occasion t1. Therefore, the terminal device performs measurement at the reporting occasion t1 to obtain a measurement value, inputs the measurement value into Model 2 to obtain a predicted result of an RLF or HOF event, and sends prediction information including the predicted result to the network device.

[0243] For the reporting occasion t2, since the time interval between the reporting occasion t2 and the reporting occasion t1 is less than the first prediction duration Δt1, and the terminal device performs measurement at the reporting occasion t1 to obtain a measurement value, at the reporting occasion t2, the terminal device can input the measurement value obtained at the reporting occasion t1 into Model 1 to obtain a predicted value, input the predicted value into Model 2 to obtain a predicted result of an RLF or HOF event, and send prediction information including the predicted result to the network device.

[0244] For the reporting occasion t3, although the time interval between the reporting occasion t3 and the reporting occasion t2 is less than the first prediction duration Δt1, the terminal device does not actually perform measurement at the reporting occasion t2. Therefore, the terminal device performs measurement at the reporting occasion t3 to obtain a measurement value, inputs the measurement value into Model 2 to obtain a predicted result of an RLF or HOF event, and sends prediction information including the predicted result to the network device.

[0245] Similarly, the terminal device performs measurement at the reporting time t1, the reporting time t3, the reporting time t5, and the like, and takes the measurement result / measurement value as the input of Model 2. The terminal device inputs the measurement result / measurement value of the last reporting period into Model 1 to obtain a predicted value at the reporting time t2, the reporting time t4, the reporting time t6, and the like, and takes the predicted value as the input of Model 2. It can be understood that in Example 5, the network device configures the time interval ΔT of the prediction report based on the first prediction time Δt1 and the second prediction time Δt2 reported by the terminal device, ΔT≤Δt1 and ΔT≤Δt2, so that the terminal device can obtain the prediction result of the RLF or HOF event based on the measurement value or the predicted value at each reporting time. Since the terminal device does not need to perform measurement at the reporting time t2, the reporting time t4, and the reporting time t6, the terminal device has more opportunities for energy saving and service transmission during the prediction period.

[0246] It should be noted that FIG. 20 is described by taking Δt1=Δt2 as an example, which does not limit the present application. In actual implementation, the first prediction time Δt1 and the second prediction time Δt2 can also satisfy the following relationship: Δt1<Δt2, or Δt1>Δt2.

[0247] Example 6:

[0248] As shown in FIG. 21, the AI model includes Model 1 and Model 2, and the reporting times are t1, t2, t3, t4, t5, t6, and the like, wherein the time interval between the third reporting time and the fourth reporting time is ΔT, Model 1 supports obtaining a predicted value based on a measurement value within the first prediction time Δt1, and Model 2 supports obtaining a prediction result of an RLF or HOF event based on a measurement value within the second prediction time Δt2. The fourth reporting time is the last reporting time of the third reporting time, for example, the third reporting time is t2, and the fourth reporting time is t1. Since the second prediction time Δt2 starting from the fourth reporting time is within the capability range of Model 2, and the time interval ΔT between the third reporting time and the fourth reporting time is less than the second prediction time Δt2, it indicates that the third reporting time is within the capability range of Model 2, so at the third reporting time, the terminal device can further determine to take the measurement value or the predicted value as the input of Model 2 based on the size relationship between the time interval ΔT and the first prediction time Δt1, to obtain the prediction result of the RLF or HOF event.

[0249] Specifically, for the reporting occasion t1, the reporting occasion t2, the reporting occasion t3, the reporting occasion t4, the reporting occasion t5, the reporting occasion t6, …, since the first prediction duration Δt1 starting from the fourth reporting occasion is within the capability range of Model1, and the time interval ΔT between the third reporting occasion and the fourth reporting occasion is greater than the first prediction duration Δt1, it is indicated that the third reporting occasion is not within the capability range of Model1, and therefore the terminal device needs to perform measurement at the third reporting occasion and input the measurement value as the input of Model2.

[0250] It can be understood that in Example 6, the network device configures the time interval ΔT of the prediction reporting based on the first prediction duration Δt1 and the second prediction duration Δt2 reported by the terminal device, ΔT > Δt1 and ΔT ≤ Δt2, so that the terminal device can obtain the prediction result of the RLF or HOF event based on the measurement value at each reporting occasion.

[0251] Example 7:

[0252] As shown in FIG. 22 or FIG. 23, the AI model includes Model1 and Model2, and the reporting occasions are t1, t2, t3, t4, t5, t6, …, wherein the time interval ΔT between the third reporting occasion and the fourth reporting occasion is ΔT, Model1 supports obtaining the prediction value based on the measurement value within the first prediction duration Δt1, and Model2 supports obtaining the prediction result of the RLF or HOF event based on the measurement value within the second prediction duration Δt2. The fourth reporting occasion is the last reporting occasion of the third reporting occasion, for example, the third reporting occasion is t2, and the fourth reporting occasion is t1.

[0253] The second prediction duration Δt2 starting from the fourth reporting occasion is within the capability range of Model2, and the time interval ΔT between the third reporting occasion and the fourth reporting occasion is greater than the second prediction duration Δt2, which indicates that the third reporting occasion is not within the capability range of Model2. Under this premise, whether ΔT > Δt1 or ΔT ≤ Δt1 (that is, regardless of the capability range of Model1), the terminal device can only perform measurement at each reporting occasion and input the measurement value into Model2 to obtain the prediction result of the RLF or HOF event, and send the prediction information including the prediction result to the network device. For the gap between the two second prediction durations Δt2, the terminal device can enable the traditional measurement reporting mode, and send the measurement result based on the traditional measurement reporting mode to the network device.

[0254] It can be understood that in Example 7, the network device configures the time interval ΔT of the prediction report based on the second prediction time length Δt2 reported by the terminal device, ΔT > Δt2, so that the terminal device can obtain the prediction result of the RLF or HOF event based on the measurement value within the second prediction time length Δt2, and enable the traditional measurement report mode between two second prediction time lengths Δt2.

[0255] The above embodiments introduce the execution of the prediction of the RLF or HOF event by the AI model. Due to the influence of the transformation of the wireless environment or the mobility of the terminal device (such as the terminal device moving to another scene), the AI model is no longer applicable. In order to solve this problem, the present application also provides two methods for monitoring the performance of the AI model.

[0256] Exemplarily, FIG. 24 is a flowchart of the network device monitoring the performance of the AI model according to an embodiment of the present application.

[0257] As shown in FIG. 24, the method can include the following S21-S24.

[0258] S21, the network device sends first information to the terminal device.

[0259] Correspondingly, the terminal device receives the first information from the network device.

[0260] In some embodiments, if the AI model at least includes Model2, the first indication information can be used to indicate that the performance of Model2 is monitored at the first time. Or, if the AI model includes Model1 and Model2, and supports taking the prediction value as the input of Model2, the first indication information can be used to indicate that the performance of Model1 and Model2 is monitored at the second time.

[0261] In some embodiments, the first information can include any one of the following:

[0262] ① the starting time point and the monitoring time interval of the performance monitoring of the AI model, that is, the periodic performance monitoring is performed;

[0263] ② indicating that the performance monitoring of the AI model is triggered when a preset event occurs or a preset threshold is met, for example, if the proportion of the prediction result error in a period of time triggers the performance monitoring of the AI model;

[0264] ③ the time point of the performance monitoring of the AI model, that is, the network configures the performance monitoring at certain time points;

[0265] ④ indicating that the performance monitoring of the AI model is performed immediately, for example, the network device triggers the performance monitoring immediately through signaling.

[0266] S22, the terminal device performs performance monitoring on the AI model and reports the monitoring result to the network device.

[0267] Correspondingly, the network device receives the monitoring result from the terminal device.

[0268] The network device can indicate to the terminal device to perform performance monitoring on Model2, or on Model1 and Model2. As an example, the network device can indicate to the terminal device to perform performance monitoring on which Model at which occasion. As another example, if the network device does not indicate to the terminal device the specific occasion to perform performance monitoring, the terminal device can perform performance monitoring on Model2, or on Model1 and Model2 at each occasion configured by the network device.

[0269] If the first indication information indicates to perform performance monitoring on Model2 at the first occasion, the monitoring result (may be referred to as first monitoring result) can include at least one of the following: a measurement value input at Model2 (such as L1 measurement value and / or L3 measurement value), an identifier of Model2, and an identifier of a first prediction functionality corresponding to Model2. The identifier of the first prediction functionality is an identifier of a certain prediction functionality or certain prediction functionalities corresponding to Model2.

[0270] If the first indication information indicates to perform performance monitoring on Model1 and Model2 at the second occasion, the monitoring result (may be referred to as second monitoring result) can include at least one of the following: a measurement value, a measurement value input at Model1 (such as L1 measurement value and / or L3 measurement value), a measurement value input at Model2 (such as L1 measurement value and / or L3 measurement value), an identifier of Model1, an identifier of Model2, an identifier of a first prediction functionality corresponding to Model2, and an identifier of a second prediction functionality corresponding to Model1. The identifier of the first prediction functionality is an identifier of a certain prediction functionality or certain prediction functionalities corresponding to Model2, and the identifier of the second prediction functionality is an identifier of a certain prediction functionality or certain prediction functionalities corresponding to Model1.

[0271] Exemplarily, as shown in FIG. 25, the network device can instruct to monitor the performance of Model2 at t3, and monitor the performance of Model1 and Model2 at t4. At t3, the content fed back by the terminal device to the network device includes: the L1 measurement value and / or the L3 measurement value, the identification of Model2, and the identification of the first prediction characteristic corresponding to Model2. At t4, the content fed back by the terminal device to the network device includes: the measurement value, the L1 measurement value and / or the L3 measurement value input into Model1, the L1 measurement value and / or the L3 measurement value input into Model2, the identification of Model1, the identification of Model2, the identification of the first prediction characteristic corresponding to Model2, and the identification of the second prediction characteristic corresponding to Model1.

[0272] It should be noted that the specific content of the first prediction characteristic and the second prediction characteristic can be adjusted according to the use requirement, and the present application does not make specific limitation. For example, the first prediction characteristic corresponding to Model2 can be the second prediction time length Δt2 or other characteristics representing the prediction performance of Model2, and the second prediction characteristic corresponding to Model1 can be the first prediction time length Δt1 or other characteristics representing the prediction performance of Model1.

[0273] S23, the network device determines, based on the monitoring result, to deactivate the AI model, or to retrain the AI model, or to reselect another AI model.

[0274] S24, the network device sends second information to the terminal device.

[0275] Correspondingly, the terminal device receives the second information from the network device.

[0276] The second information can be used to instruct to deactivate the AI model, or to retrain the AI model, or to reselect another AI model.

[0277] In the method for monitoring the performance of the AI model by the network device provided in the present application, the network device can determine the management of each model based on the performance monitoring feedback of the terminal device, such as determining that the current AI model is no longer applicable, and performing the deactivation, reselection or reselection of the model. It can be understood that through the monitoring of the AI model, it can be ensured that the AI model used for prediction is a suitable model, thereby bringing higher performance. In addition, the management of the model performed by the network device can reduce the complexity of the terminal device.

[0278] Exemplarily, FIG. 26 is a flowchart of the terminal device monitoring the performance of the AI model provided in the embodiments of the present application.

[0279] In combination with FIG. 9, as shown in FIG. 26, before S15, the method can include the following S31; S15 can be specifically implemented by the following S32. As an example, after S32, the method can further include the following S33 and S34.

[0280] S31, the terminal device obtains a confidence parameter by evaluating the confidence degree of the prediction information obtained based on the AI model this time.

[0281] S32, the terminal device reports the prediction information to the network device. The prediction information can include not only the prediction result of RLF or HOF, but also the confidence parameter.

[0282] In the process of reporting the prediction result, the prediction information can also include the confidence parameter. The confidence parameter is used to represent the confidence degree of the prediction information obtained based on the AI model. As an example, the confidence parameter can be a percentage parameter, such as 60%, 80%, etc. Before performing the prediction result reporting, the terminal device needs to evaluate the confidence degree of the prediction information obtained based on the AI model this time, and include the confidence parameter in the prediction result information, so as to facilitate the network device to determine the confidence degree of the prediction information this time based on the confidence parameter.

[0283] S33, the terminal device deactivates the AI model, or re-trains the AI model, or re-selects other AI model.

[0284] In one way, the network device configures a threshold of confidence. In the case that the terminal evaluates that the confidence parameter is less than or equal to the preset threshold, the terminal device deactivates the AI model, or re-trains the AI model, or re-selects other AI model.

[0285] In another way, the network device does not need to configure the threshold of confidence. It is completely controlled by the terminal. The terminal evaluates the confidence of each prediction, and further performs the management of the AI model, such as deactivating the AI model, or re-training the AI model, or re-selecting other AI model.

[0286] S34, the terminal device sends a decision result of the AI model to the network device, and the decision result includes any one of the following: deactivating the AI model, or re-training the AI model, or re-selecting other AI model.

[0287] It should be noted that S33 and S34 are optional steps. For example, in the case that the terminal evaluates that the confidence parameter is greater than the preset threshold, S33 and S34 do not need to be performed.

[0288] In the method for terminal device to perform performance monitoring on an AI model provided in the application, the AI model performance feedback is not required in the process by using the monitoring of the AI model on the terminal side, and thus the signaling load is reduced.

[0289] The application further provides a computer readable storage medium, which stores a computer program; when the computer readable storage medium is run on a terminal device or a network device, the terminal device or the network device performs the method as shown above. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, a data center, etc. integrated with one or more media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk or a magnetic tape), an optical medium or a semiconductor medium (for example, a solid state disk (SSD)) and the like.

[0290] The application further provides a computer program product, which includes computer program code, when the computer program code is run on a computer, the computer performs the method in each of the above embodiments.

[0291] The application further provides a chip, which is coupled with a memory, and is used to read and execute a computer program or instructions stored in the memory to perform the method in each of the above embodiments. The chip can be a general-purpose processor or a special-purpose processor. It should be noted that the chip can be implemented by using one or more field programmable gate arrays (FPGA), programmable logic devices (PLD), controllers, state machines, gate logic, discrete hardware components, any other suitable circuit, or any combination of circuits capable of performing the various functions described throughout the application.

[0292] It should be noted that the terms "first" and "second" and the like in the specification, claims and drawings of the application are used to distinguish different objects, and are not used to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed or can optionally include other steps or units inherent to the process, method, product or device.

[0293] It should be understood that, in the present application, "at least one" means one or more, "multiple" means two or more, "at least two" means two or three and three or more, and "and / or" is used to describe the association relationship of the associated objects, which means that there can be three relationships, for example, "A and / or B" can mean: only A, only B, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after it. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0294] It should be understood that in the embodiments of the present application, "B corresponding to A" means that B is associated with A. For example, B can be determined according to A. It should also be understood that determining B according to A does not mean that B is determined only according to A, but B can also be determined according to A and / or other information. In addition, "connection" appearing in the embodiments of the present application means direct connection or indirect connection and various connection modes to achieve communication between devices, which is not limited by the embodiments of the present application.

[0295] The "transmit / transmission" appearing in the embodiments of the present application means bidirectional transmission containing sending and / or receiving actions, unless otherwise specified. Specifically, the "transmit / transmission" in the embodiments of the present application contains sending of data, receiving of data, or sending of data and receiving of data. Or, the data transmission here includes uplink and / or downlink data transmission. The data can include channels and / or signals, uplink data transmission is uplink channel and / or uplink signal transmission, and downlink data transmission is downlink channel and / or downlink signal transmission.

[0296] Through the description of the above embodiments, those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional modules is taken as an example for illustration, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above.

[0297] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. For example, the division of the apparatus embodiments is merely an example, and for example, the division of the modules or units can be different, and for example, multiple modules or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0298] The units described as separate components may or may not be physically separate, and the components shown as units may be one physical unit or multiple physical units, i.e., may be located in one place, or may be distributed in multiple different places. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0299] In addition, each functional unit in the various embodiments of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0300] If the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present application essentially or the parts that make contributions to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium, includes several instructions to make a device, such as a single-chip microcomputer, chip, or processor, execute all or part of the steps of the methods provided in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, ROM, RAM, magnetic disk or optical disk, and various program codes that can be stored.

[0301] The above is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any changes or replacements within the technical scope disclosed in the present application should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for reporting prediction information, characterized in that, The method comprises: receiving configuration information from a network device, the configuration information comprising a reporting occasion; sending, to the network device, prediction information obtained based on an artificial intelligence (AI) model at the reporting occasion, the prediction information comprising a prediction result of a radio link failure (RLF) or handover failure (HOF) event.

2. The method of claim 1, wherein, The reporting occasion comprises: a prediction window in which the AI model is activated; and a time point in the prediction window at which the reporting of the prediction information is performed.

3. The method of claim 2, wherein: the prediction window comprises a start time point and an end time point of the prediction window, the start time point being a first preset frame or a first preset time slot, and the end time point being a second preset frame or a second preset time slot; the time point in the prediction window comprises a third preset frame, a third preset time slot, or a time interval for reporting the prediction information in the prediction window.

4. The method of claim 1, wherein, The reporting occasion comprises: a start occasion for reporting the prediction information, the start occasion being a time point at which the AI model is activated; and a time interval for reporting the prediction information.

5. The method according to any one of claims 1 to 4, characterized in that, The sending, to the network device, of the prediction information obtained based on the AI model at the reporting occasion comprises: inputting, at the reporting occasion, a measurement value into the AI model to obtain the prediction information, and sending the prediction information to the network device; wherein the measurement value comprises at least one of the following: a layer-one measurement value, a layer-three measurement value.

6. The method of claim 5, wherein, The configuration information is configured based on a prediction capability of a terminal device, and the AI model comprises a radio resource management (RRM) prediction model and an RLF or HOF prediction model. Before the receiving of the configuration information from the network device, the method further comprises: sending, to the network device, a prediction capability of the terminal device; wherein the prediction capability comprises at least one of the following: a first prediction duration, in which the RRM prediction model supports obtaining a prediction value based on a measurement value; an indication that a prediction value output by the RRM prediction model is supported as an input of the RLF or HOF prediction model, and an output of the RLF or HOF prediction model is a prediction result of an RLF or HOF event.

7. The method of claim 6, wherein, The reporting occasion comprises at least a first reporting occasion and a second reporting occasion, the second reporting occasion being a previous reporting occasion of the first reporting occasion. The inputting, at the reporting occasion, of the measurement value into the AI model to obtain the prediction information, and the sending of the prediction information to the network device, comprises: If a time interval between the first reporting occasion and the second reporting occasion is less than or equal to the first prediction duration, and the terminal device performs measurement at the second reporting occasion to obtain a first measurement value, at the first reporting occasion, the first measurement value is input into the RRM prediction model to obtain a first prediction value, the first prediction value is input into the RLF or HOF prediction model to obtain a first prediction result of the RLF or HOF event, and first prediction information including the first prediction result is sent to the network device; or, If the time interval between the first reporting occasion and the second reporting occasion is less than or equal to the first prediction duration, and the terminal device does not perform measurement at the second reporting occasion; Or, if the time interval between the first reporting occasion and the second reporting occasion is greater than the first prediction duration, at the first reporting occasion, measurement is performed to obtain a second measurement value, the second measurement value is input into the RLF or HOF prediction model to obtain a second prediction result of the RLF or HOF event, and second prediction information including the second prediction result is sent to the network device.

8. The method of claim 7, wherein: the first prediction information further includes a data type input into the RLF or HOF prediction model, the data type being a prediction value; the second prediction information further includes a data type input into the RLF or HOF prediction model, the data type being a measurement value.

9. The method of claim 5, wherein, the configuration information is configured based on a prediction capability of the terminal device; the AI model includes the RLF or HOF prediction model; before the receiving of the configuration information from the network device, the method further includes: sending, to the network device, the prediction capability of the terminal device; wherein the prediction capability includes a second prediction duration, within which the RLF or HOF prediction model supports obtaining a prediction result of the RLF or HOF event based on a measurement value or a prediction value, the prediction value being obtained based on the measurement value.

10. The method of claim 9, wherein, the reporting occasions include at least a third reporting occasion and a fourth reporting occasion, the fourth reporting occasion being a previous reporting occasion of the third reporting occasion; the inputting of the measurement value into the AI model at the reporting occasion to obtain the prediction information and the sending of the prediction information to the network device include: if a time interval between the third reporting occasion and the fourth reporting occasion is less than or equal to the second prediction duration, at the third reporting occasion, the measurement value or the prediction value is input into the RLF or HOF prediction model to obtain a third prediction result of the RLF or HOF event, and third prediction information including the third prediction result is sent to the network device.

11. The method of claim 10, wherein, the third prediction information further includes a data type input into the RLF or HOF prediction model, the data type being a measurement value or a prediction value.

12. The method of claim 10, wherein, The third prediction result of the RLF or HOF event is obtained by inputting the measurement value or the predicted value obtained at the third reporting time into the RLF or HOF prediction model. If the AI model only includes the RLF or HOF prediction model, the third prediction result is obtained by inputting a third measurement value obtained by performing measurement at the third reporting time into the RLF or HOF prediction model. If the AI model includes the RRM prediction model and the RLF or HOF prediction model, the third prediction result is obtained by inputting a third measurement value obtained by performing measurement at the third reporting time or a second predicted value obtained based on the RRM prediction model into the RLF or HOF prediction model.

13. The method of claim 12, wherein, The RRM prediction model supports obtaining a predicted value based on a measurement value within a first prediction duration. The third prediction result of the RLF or HOF event is obtained by inputting the measurement value or the predicted value obtained at the third reporting time into the RLF or HOF prediction model. If the time interval between the third reporting time and the fourth reporting time is greater than the first prediction duration, or if the time interval between the third reporting time and the fourth reporting time is less than or equal to the first prediction duration and measurement is not performed at the fourth reporting time, the third prediction result is obtained by inputting a third measurement value obtained by performing measurement at the third reporting time into the RLF or HOF prediction model; or If the time interval between the third reporting time and the fourth reporting time is less than or equal to the first prediction duration and a fourth measurement value is obtained by performing measurement at the fourth reporting time, the third prediction result is obtained by inputting the fourth measurement value into the RRM prediction model to obtain the second predicted value and inputting the second predicted value into the RLF or HOF prediction model.

14. The method of claim 10, wherein, The method further includes: If the time interval between the third reporting time and the fourth reporting time is greater than the second prediction duration, a fifth measurement value is obtained by performing measurement at the third reporting time, a fourth prediction result is obtained by inputting the fifth measurement value into the RLF or HOF prediction model, and fourth prediction information including the fourth prediction result is sent to the network device.

15. The method of claim 10, wherein, The method further includes: If the time interval between the third reporting time and the fourth reporting time is greater than the second prediction duration, a traditional measurement reporting mode is enabled in a first interval, and a measurement result based on the traditional measurement reporting mode is sent to the network device, the first interval being an idle duration between two second prediction durations.

16. The method of any one of claims 6 to 15, wherein The prediction capability of the terminal device is sent to the network device, including: The prediction capability of the terminal device is sent to the network device through user equipment assistance information signaling or radio resource control signaling.

17. The method of any one of claims 1 to 16, wherein, The AI model comprises an RLF or HOF prediction model; the configuration information further comprises: a data type input to the RLF or HOF prediction model at the reporting occasion, the data type being a measurement value or a prediction value.

18. The method of any one of claims 1 to 17, wherein, The method further comprises: receiving first information from the network device, the first information indicating performance monitoring of an RLF or HOF prediction model, or performance monitoring of an RRM prediction model and the RLF or HOF prediction model; in response to the first information, sending monitoring results to the network device; receiving second information from the network device, the second information being determined based on the monitoring results; wherein the second information indicates deactivating the AI model, or retraining the AI model, or reselecting another AI model.

19. The method of claim 18, wherein, The first indication information indicates that performance monitoring of the RLF or HOF prediction model is performed at a first occasion; The response to the first information, sending monitoring results to the network device, comprises: in response to the first information, performing performance monitoring of the RLF or HOF prediction model at the first occasion, and sending first monitoring results to the network device; wherein the first monitoring results comprise: a measurement value, an identifier of the RLF or HOF prediction model, and an identifier of a first prediction characteristic corresponding to the RLF or HOF prediction model.

20. The method of claim 18, wherein, The first indication information indicates that performance monitoring of the RRM prediction model and the RLF or HOF prediction model is performed at a second occasion; The response to the first information, sending monitoring results to the network device, comprises: in response to the first information, performing performance monitoring of the RRM prediction model and the RLF or HOF prediction model at the second occasion, and sending second monitoring results to the network device; wherein the second monitoring results comprise: a measurement value, an identifier of the RRM prediction model, an identifier of the RLF or HOF prediction model, an identifier of a first prediction characteristic corresponding to the RLF or HOF prediction model, and an identifier of a second prediction characteristic corresponding to the RRM prediction model.

21. The method of any one of claims 18 to 20, wherein The first information comprises any one of the following: a starting time point and a monitoring time interval for performance monitoring of the AI model; indicates that performance monitoring of the AI model is triggered when a preset event occurs or a preset threshold is met; a time point for performance monitoring of the AI model; indicates that performance monitoring of the AI model is performed immediately.

22. The method of any one of claims 1 to 17, wherein, The prediction information further comprises a credibility parameter, the credibility parameter being used to represent a credibility degree of the prediction information obtained based on the AI model; Before sending the prediction information obtained based on the AI model to the network device at the reporting occasion, the method further comprises: obtaining the credibility parameter by evaluating the credibility degree of the prediction information obtained based on the AI model this time.

23. The method of claim 22, wherein, The method further comprises: In a case where the credibility parameter is less than or equal to a preset threshold, the AI model is deactivated, or the AI model is retrained, or another AI model is reselected.

24. The method of claim 23, wherein, The method further includes: sending, to the network device, a decision result of the AI model, the decision result including any one of the following: deactivating the AI model, retraining the AI model, or reselecting another AI model.

25. The method of claim 23, wherein, The preset threshold is configured by the network device.

26. A method for reporting prediction information, the method comprising: The method includes: sending, to a terminal device, configuration information, the configuration information including a reporting occasion; receiving, from the terminal device, prediction information at the reporting occasion, the prediction information being a prediction result of a radio link failure (RLF) or handover failure (HOF) event obtained based on an AI model.

27. The method of claim 26, wherein, The reporting occasion includes: a prediction window in which the AI model is activated, and a time point in the prediction window at which reporting of the prediction information is performed.

28. The method of claim 27, wherein the prediction window includes a start time point and an end time point of the prediction window, the start time point being a first preset frame or a first preset time slot, and the end time point being a second preset frame or a second preset time slot; the time point in the prediction window includes a third preset frame, a third preset time slot, or a time interval for reporting the prediction information in the prediction window.

29. The method of claim 26, wherein, The reporting occasion includes: a start time of reporting the prediction information, the start time being a time point at which the AI model is activated, and a time interval for reporting the prediction information.

30. The method of claim 26, wherein, Before the configuration information is sent to the terminal device, the method further includes: receiving, from the terminal device, a prediction capability; wherein the AI model includes an RRM prediction model and an RLF or HOF prediction model, and the prediction capability includes at least one of the following: a first prediction duration in which the RRM prediction model supports obtaining a prediction value based on a measurement value, and an indication that a prediction value output by the RRM prediction model is supported as an input of the RLF or HOF prediction model, an output of the RLF or HOF prediction model being a prediction result of an RLF or HOF event; or, the AI model includes an RLF or HOF prediction model, and the prediction capability includes a second prediction duration in which the RLF or HOF prediction model supports obtaining a prediction result of an RLF or HOF event based on a measurement value.

31. The method of claim 30, wherein, The method further includes: sending, to the terminal device, first information; receiving, from the terminal device, a listening result; wherein the first indication information indicates that performance listening is performed on the RLF or HOF prediction model at a first time point, and the listening result includes a measurement value, an identifier of the RLF or HOF prediction model, and an identifier of a first prediction characteristic corresponding to the RLF or HOF prediction model. Alternatively, the first indication information indicates that performance monitoring is performed on the RRM prediction model and the RLF or HOF prediction model at a second occasion; and the monitoring result includes a measurement value, an identifier of the RRM prediction model, an identifier of the RLF or HOF prediction model, an identifier of a first prediction characteristic corresponding to the RLF or HOF prediction model, and an identifier of a second prediction characteristic corresponding to the RRM prediction model.

32. The method of any one of claims 26-31, wherein, The method further includes: receiving a decision result from the terminal device in a case where the credibility parameter is less than or equal to a preset threshold, the credibility parameter being used to represent a credibility degree of the prediction information obtained based on the AI model; wherein the decision result includes any one of the following: deactivating the AI model, or retraining the AI model, or reselecting another AI model.

33. A communications device, characterized by The communication device includes a processor, a communication interface, and a memory coupled to the processor and the communication interface; wherein the memory stores instructions, and the processor executes the instructions to cause the communication device to perform the prediction information reporting method according to any one of claims 1 to 25, or to cause the communication device to perform the prediction information reporting method according to any one of claims 26 to 32.

34. A communication system, characterized by The communication system includes a terminal device and a network device; wherein the terminal device is configured to perform the prediction information reporting method according to any one of claims 1 to 25, and the network device is configured to perform the prediction information reporting method according to any one of claims 26 to 32.

35. A computer readable storage medium, characterized in that, The computer readable storage medium stores a computer program; wherein when the computer program runs on a terminal device, the terminal device is caused to perform the prediction information reporting method according to any one of claims 1 to 25; or when the computer program runs on a network device, the network device is caused to perform the prediction information reporting method according to any one of claims 26 to 32.

Images