Prediction method and device, communication equipment, storage medium and program product
By establishing a confidence threshold mechanism between the provider and the user of the model prediction result, processing low confidence prediction results, and using multiple models or non-AI algorithms, the problem of unreliability of the model prediction result is solved and the wireless system is improved Control decision reliability.
Patent Information
- Application Number
- CN202411161775.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-12-06
AI Technical Summary
The prior art fails to effectively describe and convey the confidence of the model's predicted results, resulting in unreliable use of the predicted results, which may lead to abnormal subsequent control decisions.
By establishing a confidence threshold mechanism between the provider and user of the model prediction result, if the prediction confidence is below the threshold, multiple model predictions, non-AI algorithms, or default values are used to process the target result, and the model is updated through the notification mechanism Or use non-AI algorithms to improve decision-making reliability.
This reduces the impact of uncertainty in the prediction results on the subsequent processing process and improves the reliability of control decisions of the wireless system.
Smart Images

Figure CN119095085A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of wireless technology, and in particular to a prediction method and apparatus, communication equipment, storage medium, and program product. Background Art
[0002] The confidence of a model's prediction is an important indicator to measure the certainty of the model's prediction results. It is usually used to express how "confident" the model is in its prediction results. When the confidence of the model's prediction results is low, there is a high probability that the prediction results are unreliable, and if subsequent processing is based on the prediction results, anomalies may occur.
[0003] In summary, the confidence level of the model prediction results (also referred to as prediction confidence level) is not described and communicated, so that the users of the prediction results do not know the credibility of the prediction results. When the confidence level of the prediction results is low, there is a risk of causing subsequent control decisions to be unreliable and abnormal processing. Summary of the invention
[0004] The embodiments of the present application provide a prediction method and apparatus, a communication device, a computer storage medium, and a computer program product.
[0005] The prediction method provided in the embodiment of the present application is applied to a first device, and the prediction method includes:
[0006] When it is determined that the prediction confidence of the first model is lower than the threshold, a first processing method is used to obtain the target result; the first processing method includes at least one of the following: using the second model to predict and obtain the target result; using multiple models to predict and obtain multiple results, and determining the target result based on the multiple results; using a non-AI algorithm to obtain the target result; using a default value or a conservative value as the target result;
[0007] A result message is sent to the second device, where the result message includes the target result.
[0008] The prediction method provided in the embodiment of the present application is applied to the second device, and the prediction method includes:
[0009] Receiving a result message sent by the first device, wherein the result message includes a target result;
[0010] When it is determined that the prediction confidence of the target result is lower than a threshold, a second processing method is adopted; the second processing method includes at least one of the following: using a third model to process the target result; sending a notification message to the first device, the notification message is used to notify the first device to obtain the target result through an updated model, and / or notify the first device to obtain the target result using a non-AI algorithm, and / or notify the first device to use a default value or a conservative value as the target result.
[0011] The prediction device provided in the embodiment of the present application is applied to a first device, and the device includes:
[0012] A first processing unit is used to determine that when the prediction confidence of the first model is lower than a threshold, a first processing method is used to obtain a target result; the first processing method includes at least one of the following: using the second model to predict and obtain the target result; using multiple models to predict and obtain multiple results, and determining the target result based on the multiple results; using a non-AI algorithm to obtain the target result; using a default value or a conservative value as the target result;
[0013] The first communication unit is used to send a result message to the second device, where the result message includes the target result.
[0014] The prediction device provided in the embodiment of the present application is applied to a second device, and the device includes:
[0015] A second communication unit, configured to receive a result message sent by the first device, wherein the result message includes a target result;
[0016] A second processing unit is used to determine that when the prediction confidence of the target result is lower than a threshold, a second processing method is adopted; the second processing method includes at least one of the following: using a third model to process the target result; sending a notification message to the first device, the notification message is used to notify the first device to obtain the target result through an updated model, and / or notify the first device to obtain the target result by using a non-AI algorithm, and / or notify the first device to use a default value or a conservative value as the target result.
[0017] The communication device provided in the embodiment of the present application includes: a processor and a memory, the memory is used to store a computer program, and the processor is used to call and run the computer program stored in the memory to execute any one of the above-mentioned prediction methods.
[0018] The computer-readable storage medium provided in the embodiments of the present application is used to store a computer program, and the computer program enables a computer to execute any one of the above-mentioned prediction methods.
[0019] The computer program product provided in the embodiments of the present application includes computer program instructions, which enable a computer to execute any one of the above prediction methods.
[0020] In the technical solution of the embodiment of the present application, through the perception of the model prediction confidence, a processing method for the prediction result provider (i.e., the first device) when the prediction confidence is lower than the threshold is proposed, as well as a processing method for the prediction result user (i.e., the second device) when the prediction confidence is lower than the threshold, thereby reducing the impact of the uncertainty of the prediction results on subsequent control decisions. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 This is a schematic diagram of AI-based CSI compression feedback;
[0022] Figure 2 This is a schematic diagram of CSI time domain prediction based on AI;
[0023] Figure 3 This is a schematic diagram of user mobility optimization based on AI trajectory prediction;
[0024] Figure 4 This is a schematic diagram of the prediction method provided in the embodiment of the present application. Figure 1 ;
[0025] Figure 5 This is a schematic diagram of the prediction method provided in the embodiment of the present application. Figure 2 ;
[0026] Figure 6 It is a flowchart of the model confidence threshold configuration provided in the embodiment of the present application;
[0027] Figure 7 It is a schematic diagram of the processing flow of the result provider provided in the embodiment of the present application;
[0028] Figure 8 It is a schematic diagram of the processing flow of the result user provided in the embodiment of the present application;
[0029] Fig. 9 It is a flow chart of application example 1 provided in the embodiment of the present application;
[0030] Fig.10 This is a flow chart of application example 2 provided in the embodiment of the present application;
[0031] Fig.11 This is a schematic diagram of the structure of the prediction device provided in the embodiment of the present application. Figure 1 ;
[0032] Fig.12 This is a schematic diagram of the structure of the prediction device provided in the embodiment of the present application. Figure 2 ;
[0033] Fig.13 is a schematic structural diagram of a communication device provided in an embodiment of the present application;
[0034] Fig.14 It is a schematic structural diagram of the chip of an embodiment of the present application. DETAILED DESCRIPTION
[0035] It should be noted that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the objects associated with each other are in an "or" relationship. It should also be understood that the "indication" mentioned in this article can be a direct indication, an indirect indication, or an indication of an association relationship. For example, A indicates B, which can mean that A directly indicates B, for example, B can be obtained through A; it can also mean that A indirectly indicates B, for example, A indicates C, and B can be obtained through C; it can also mean that there is an association relationship between A and B. It should also be understood that the "protocol" mentioned in this article can refer to a standard protocol in the field of communications, such as the NR protocol and related protocols used in future communication systems, and this application does not limit this.
[0036] To facilitate understanding of the technical solutions of the embodiments of the present application, the relevant technologies of the embodiments of the present application are described below. The following related technologies can be arbitrarily combined with the technical solutions of the embodiments of the present application as optional solutions, and they all belong to the protection scope of the embodiments of the present application.
[0037] The wireless air interface protocol stack is divided into the control plane and the user plane. The control plane includes Radio Resource Control (RRC), Packet Data Convergence Protocol (PDCP), Radio Link Control (RLC), Media Access Control (MAC) and PHY, and the user plane includes Service Data Adaptation Protocol (SDAP), PDCP, RLC, MAC and PHY. With the deep integration of Artificial Intelligence (AI) / Machine Learning (ML) and wireless, the processing of each layer of the wireless protocol stack is gradually introducing AI models for reasoning / prediction / classification, etc., replacing the current traditional algorithms. For example, PHY introduces AI-based Channel State Information (CSI) compression feedback and CSI time domain prediction to improve the efficiency and accuracy of CSI feedback.
[0038] like Figure 1As shown in the figure, in some studies of the 3rd Generation Partnership Project (3GPP), CSI compression feedback adopts a dual-side model of the terminal and the base station. The terminal side model runs in the CSI encoder, compresses the CSI based on the feature vector through the AI / ML model, and feeds the result back to the decoder on the base station side through the CSI reporting message. The CSI decoder on the base station side also decodes the CSI based on the AI / ML model and submits the decoded CSI to the MAC layer for resource scheduling.
[0039] like Figure 2 As shown in the figure, the AI-based CSI time domain prediction adopts a terminal unilateral model and uses AI models such as the Long Short-Term Memory Network (LSTM) as the CSI prediction model. The CSI prediction model predicts the CSI at multiple moments in the future based on multiple historical CSIs and sends it to the base station through CSI reporting for MAC scheduling or beamforming and other processes.
[0040] like Figure 3 As shown in the figure, the process of user mobility optimization based on AI trajectory prediction collects measurement information through the Xn port (i.e., the interface between gNBs), uses AI / ML to predict the trajectory of user equipment (UE), makes a UE switching decision based on the prediction result, and completes the switching process.
[0041] Based on the above AI use cases, most of them utilize the prediction ability of AI and use the prediction results for subsequent corresponding processes, such as resource scheduling, beam management or switching.
[0042] It is foreseeable that after AI model prediction is introduced into each layer of the air interface protocol stack, the information interacted by each layer will contain many AI prediction results, so as to prepare for future events in advance. The processing of wireless processes changes from passive to active, such as passive switching to active switching.
[0043] The confidence of the model prediction is an important indicator to measure the certainty of the model's prediction results. Current technology does not describe and convey the confidence of the model prediction results (also referred to as prediction confidence), so that the user of the prediction results does not know the credibility of the prediction results. When the confidence of the prediction results is low, there is a risk of unreliable subsequent decisions and abnormal processing. For example, (1) The CSI prediction results are provided to the base station for subsequent MAC scheduling. When the confidence of the CSI prediction results is low, the scheduling of resources for the UE and the selection of the modulation and coding scheme (MCS) may be unreasonable, resulting in bit errors or waste of resources. (2) When the UE trajectory prediction results are unreliable, it will lead to incorrect switching decisions, which may cause users to drop offline and affect the service experience.
[0044] Therefore, when a network element entity and / or a layer of the protocol stack provides an AI prediction result, its confidence determines how the next layer or receiving entity processes it. Currently, the layers and / or entities do not interact with each other on the confidence of the model reasoning results, and do not involve how to process prediction results with different confidence levels.
[0045] To this end, a technical solution of the embodiment of the present application is proposed. In the technical solution of the embodiment of the present application, a solution for improving the reliability of control decision-making based on AI prediction result confidence perception is proposed.
[0046] To facilitate understanding of the technical solutions of the embodiments of the present application, the technical solutions of the present application are described in detail below through specific embodiments. The above related technologies can be combined arbitrarily with the technical solutions of the embodiments of the present application as optional solutions, and they all belong to the protection scope of the embodiments of the present application. The embodiments of the present application include at least part of the following contents.
[0047] It should be noted that the “model” described in the embodiments of the present application can also be replaced by “AI model”, “neural network model”, “network model”, “ML model”, etc.
[0048] It should be noted that the “prediction confidence” described in the embodiments of the present application refers to the confidence of the prediction results of the model, and the confidence here can also be described as the degree of credibility.
[0049] It should be noted that the "threshold" described in the embodiments of the present application refers to the threshold of prediction confidence, which can be referred to as prediction confidence threshold or confidence threshold for short.
[0050] Figure 4 This is a schematic diagram of the prediction method provided in the embodiment of the present application. Figure 1 The prediction method is applied to a first device, where the first device refers to a result provider, such as a terminal, a base station, etc. Figure 4As shown, the prediction method includes the following steps:
[0051] Step 401: When it is determined that the prediction confidence of the first model is lower than a threshold, a first processing method is used to obtain a target result.
[0052] In the embodiment of the present application, when the first device (ie, the result provider) determines that the prediction confidence of the first model is lower than a threshold, the first processing method is used to obtain the target result.
[0053] In some implementations, the first device may obtain the prediction confidence threshold in the following manner:
[0054] Method 1: The first device configures the threshold itself.
[0055] Further, the first device receives a configuration request message sent by the second device; the first device sends a configuration response message to the second device, where the configuration response message includes a threshold.
[0056] Mode 2: The first device sends a configuration request message to the third device; the first device receives a configuration response message sent by the third device, where the configuration response message includes a threshold.
[0057] The third device here refers to a wireless management and control platform or other network devices capable of configuring thresholds.
[0058] In some implementations, in addition to the threshold, the configuration response message further includes at least one of the following information corresponding to the threshold: model ID, model class, and model function ID.
[0059] In some implementations, the threshold is configured according to at least one of the following dimensions: model type, model, model function.
[0060] Exemplarily, different confidence thresholds are set for different model types, for example, corresponding confidence thresholds are set according to different types such as classification models and regression models.
[0061] Exemplarily, a corresponding confidence threshold is set for each model. The setting principle can be based on performance indicators such as model prediction accuracy and recall rate obtained during model training. For example, the confidence threshold of a model with high accuracy can be lower.
[0062] Exemplarily, the confidence threshold is set according to the importance of the function of the model service (referred to as the model function) or the application scenario requirements. For example, the confidence threshold of the CSI prediction function is higher (for example, 0.95), and the confidence threshold of the UE trajectory prediction function is lower, such as 0.9. The confidence threshold of the prediction of the UE's own cell or neighboring cell RSRP measurement function can be lower, such as 0.85.
[0063] The first device obtains a result through prediction by the first model. When the confidence of the result is lower than a threshold, or when the confidence of the result obtained by the first device through n consecutive predictions is lower than the threshold, the first device obtains the target result by using a first processing method. Here, the first processing method includes at least one of the following:
[0064] 1) Use the second model to predict the target result.
[0065] Here, the second model refers to a model other than the first model. The first device can re-execute the AI reasoning / prediction by replacing the model (ie, the second model) and use the obtained result as the target result.
[0066] In some implementations, when the result user and the result provider use a paired model (or matching model), the model replaced by the first device needs to be a model that matches the model identifier indicated by the reported result. The reported result here is provided by the data information collection end (which can be the result user or the result provider). Specifically, the data information collection end collects data information (such as channel information, service information, UE location information, etc.), and the result provider processes the data information through the model to obtain a prediction result.
[0067] 2) Use multiple models to predict multiple results, and determine the target result based on the multiple results.
[0068] As an implementation method, multiple results can be averaged to obtain the target result.
[0069] As another implementation, multiple results may be weighted according to the confidence level to obtain the target result. Here, each result corresponds to a confidence level, and the multiple results are weighted averaged according to the confidence levels of the multiple results to obtain the target result.
[0070] 3) Use non-AI algorithms to obtain target results.
[0071] Here, the non-AI algorithm can be a traditional rule-based processing algorithm. The non-AI algorithm can also be described as a traditional algorithm, or a rule-based algorithm, or a rule-based traditional algorithm.
[0072] 4) Use default or conservative values as target results.
[0073] Here, default values may be pre-configured.
[0074] Here, the conservative value can be calculated through confidence and configuration strategy.
[0075] Step 402: Send a result message to the second device, where the result message includes a target result.
[0076] In the embodiment of the present application, after the first device obtains the target result, it sends a result message to the second device, and the result message includes the target result.
[0077] In some implementations, the result message further includes at least one of the following information:
[0078] First information, the first information is used to indicate a processing method for obtaining a target result;
[0079] Second information, the second information is used to indicate a model identifier corresponding to a model for obtaining a target result;
[0080] The third information is used to indicate the model function corresponding to the model for obtaining the target result;
[0081] Fourth information, the fourth information is used to indicate the prediction confidence of the target result;
[0082] The fifth information is used to indicate whether the target result is obtained based on a non-AI algorithm or based on a default value or a conservative value.
[0083] The technical solution of the embodiment of the present application proposes a processing solution when the confidence of the model prediction result is low, reducing the impact of the uncertainty of the prediction result on the subsequent processing process, thereby improving the reliability of the control decision of the wireless system.
[0084] Figure 5 This is a schematic diagram of the prediction method provided in the embodiment of the present application. Figure 2 , the prediction method is applied to a second device, the second device refers to a result user, such as a base station, etc. Figure 5 As shown, the prediction method includes the following steps:
[0085] Step 501: Receive a result message sent by a first device, where the result message includes a target result.
[0086] In the embodiment of the present application, the second device receives a result message sent by the first device, and the result message includes a target result. The first device here refers to a result provider.
[0087] In some implementations, the result message may also include more content, for which the specific content may refer to the aforementioned Figure 4 Related description.
[0088] Step 502: When it is determined that the prediction confidence of the target result is lower than a threshold, a second processing method is adopted.
[0089] In some implementations, the second device may obtain the prediction confidence threshold in the following manner:
[0090] Method 1: The second device sends a configuration request message to the first device; the second device receives a configuration response message sent by the first device, where the configuration response message includes a threshold.
[0091] The first device here refers to the result provider.
[0092] Mode 2: The second device sends a configuration request message to the third device; the second device receives a configuration response message sent by the third device, where the configuration response message includes a threshold.
[0093] The third device here refers to a wireless management and control platform or other network devices capable of configuring thresholds.
[0094] In some implementations, in addition to the threshold, the configuration response message further includes at least one of the following information corresponding to the threshold: model ID, model class, and model function ID.
[0095] In some implementations, the threshold is configured according to at least one of the following dimensions: model type, model, model function.
[0096] Exemplarily, different confidence thresholds are set for different model types, for example, corresponding confidence thresholds are set according to different types such as classification models and regression models.
[0097] Exemplarily, a corresponding confidence threshold is set for each model. The setting principle can be based on performance indicators such as model prediction accuracy and recall rate obtained during model training. For example, the confidence threshold of a model with high accuracy can be lower.
[0098] Exemplarily, the confidence threshold is set according to the importance of the function of the model service (referred to as the model function) or the application scenario requirements. For example, the confidence threshold of the CSI prediction function is higher (for example, 0.95), and the confidence threshold of the UE trajectory prediction function is lower, such as 0.9. The confidence threshold of the prediction of the UE's own cell or neighboring cell RSRP measurement function can be lower, such as 0.85.
[0099] When the second device determines that the confidence level of the target result is lower than the threshold, or when the second device determines that the confidence level of the target result is lower than the threshold for n consecutive times, the second device adopts a second processing method. Here, the second processing method includes at least one of the following:
[0100] 1) Use the third model to process the target results.
[0101] Here, the third model is one of the candidate models supported by the second device, and the candidate model can process the target result. The second device selects a model with higher complexity and better performance (i.e., the third model) from the candidate models to process the target result, which can improve processing efficiency and accuracy. It should be noted that processing the target result through a model with higher complexity and better performance requires higher computing power and / or processing latency.
[0102] 2) Sending a notification message to the first device, where the notification message is used to notify the first device to obtain the target result through the updated model, and / or to notify the first device to obtain the target result using a non-AI algorithm, and / or to notify the first device to use a default value or a conservative value as the target result.
[0103] As an implementation method, the second device replaces the currently used model and sends a paired model update indication (corresponding to the above notification) to the first device. In this way, the first device will also replace the currently used model. It should be noted that the models replaced by the first device and the second device also need to be paired. The first device uses the replaced model to re-obtain the target result and sends it to the first device, and the first device uses the replaced model to process the received target result.
[0104] As another implementation method, the second device will use a non-AI algorithm to process the target result. In this case, the second device sends a non-AI algorithm indication information (corresponding to the above notification) to the first device, and the first device will use the non-AI algorithm to regain the target result and send it to the first device, and the first device uses the non-AI algorithm to process the target result.
[0105] In some implementations, the notification message includes at least one of the following information:
[0106] Sixth information, where the sixth information is used to indicate a model identifier corresponding to the model updated by the first device;
[0107] seventh information, where the seventh information is used to indicate a model function corresponding to the model updated by the first device;
[0108] Eighth information, the eighth information is used to instruct the first device to obtain a target result by adopting a non-AI algorithm or based on a default value or a conservative value;
[0109] Ninth information: The ninth information is used to indicate the effectiveness condition of the notification message.
[0110] The technical solution of the embodiment of the present application proposes a processing solution when the confidence of the model prediction result is low, reducing the impact of the uncertainty of the prediction result on the subsequent processing process, thereby improving the reliability of the control decision of the wireless system.
[0111] Figure 6 is a flow chart of the model confidence threshold configuration provided in the embodiment of the present application, such as Figure 6 As shown, the process includes the following steps:
[0112] Step 601: The second device sends a configuration request message to the first device.
[0113] Here, the configuration request message may also be described as a model prediction confidence threshold configuration request message.
[0114] Step 602: The first device sends a configuration response message to the second device.
[0115] Here, the configuration response message may also be described as a model prediction confidence threshold configuration response message.
[0116] Here, the content included in the configuration response message can refer to the following Table 1.
[0117] Table 1: Parameters in the configuration response message
[0118]
[0119] In the above solution, the threshold is configured according to at least one of the following dimensions: model type, model, and model function.
[0120] Exemplarily, different confidence thresholds are set for different model types, for example, corresponding confidence thresholds are set according to different types such as classification models and regression models.
[0121] Exemplarily, a corresponding confidence threshold is set for each model. The setting principle can be based on performance indicators such as model prediction accuracy and recall rate obtained during model training. For example, the confidence threshold of a model with high accuracy can be lower.
[0122] Exemplarily, the confidence threshold is set according to the importance of the function of the model service (referred to as the model function) or the application scenario requirements. For example, the confidence threshold of the CSI prediction function is higher (for example, 0.95), and the confidence threshold of the UE trajectory prediction function is lower, such as 0.9. The confidence threshold of the prediction of the UE's own cell or neighboring cell RSRP measurement function can be lower, such as 0.85.
[0123] Figure 7 is a schematic diagram of the processing flow of the result provider provided in the embodiment of the present application, such as Figure 7 As shown, the process includes the following steps:
[0124] Step 701: The first device determines that the confidence level of the current model prediction result is lower than a threshold.
[0125] Here, the first device determines that the confidence level of the current model prediction result is lower than a threshold or the confidence level of the current model prediction result is lower than the threshold for n consecutive times, where n is a preconfigured value.
[0126] Step 702: Option 1) the first device still uses the AI algorithm to predict the target result; Option 2) the first device does not use the AI algorithm to predict the target result.
[0127] For option 1), this can be achieved by:
[0128] Method 1) The first device selects other models to obtain the target result.
[0129] Specifically, the first device selects other models to perform AI reasoning / prediction to obtain the target result.
[0130] Optionally, when the first device and the second device use a pairing model, the first device selects a model that matches the model identifier indicated by the reported result. The reported result here is provided by a data information collection end (which may be a result user or a result provider). Specifically, the data information collection end collects data information (such as channel information, service information, UE location information, etc.), and the result provider processes the data information through the model to obtain a prediction result.
[0131] Method 2) The first device uses multiple models to perform AI reasoning / prediction to obtain the target result.
[0132] Specifically, the first device uses multiple models to perform AI reasoning / prediction to obtain multiple results; as an implementation method, the multiple results can be averaged to obtain a target result; as another implementation method, the multiple results can be weighted according to the confidence level to obtain the target result.
[0133] For option 2), this can be achieved by:
[0134] Method 3) Use traditional rule-based algorithms to obtain the target results.
[0135] Method 4) Use the configured default or conservative value as the target result.
[0136] Step 703: The first device sends a result message to the second device.
[0137] Here, the content included in the result message can refer to the following Table 2.
[0138] Table 2: Parameters in the result message
[0139]
[0140]
[0141] Step 704: The second device selects an appropriate processing method according to the result message.
[0142] Figure 8 is a schematic diagram of the processing flow of the result user provided in the embodiment of the present application, such as Figure 8 As shown, the process includes the following steps:
[0143] Step 801: The second device receives the target result sent by the first device.
[0144] Step 802: The second device determines that the confidence level of the current target result is lower than a threshold.
[0145] Here, the second device determines that the confidence level of the current target result is lower than a threshold or the confidence level of the current target result is lower than the threshold for n consecutive times, where n is a preconfigured value.
[0146] Step 803: Option 1) the second device processes the target result using the model; Option 2) the second device does not process the target result using the model.
[0147] For option 1), this can be achieved by:
[0148] Method 1) The second device selects other models to process the target result.
[0149] Here, the second device selects a model with higher complexity and better performance to process the target result, which requires higher computing power and / or processing delay.
[0150] Method 2) The second device and the first device synchronously update the model, and the second device uses the updated model to process the target result.
[0151] Here, the second device and the first device use a pairing model. If the second device updates the model, the second device sends a notification message to the first device to notify the first device of the pairing model update. The first device will also update the model, thereby achieving synchronous update of the pairing model.
[0152] For option 2), this can be achieved by:
[0153] Method 3) Use traditional rule-based algorithms to process the target results.
[0154] Here, the second device sends a notification message to the first device to notify the first device to calculate the target result using a traditional rule-based algorithm, and accordingly, the second device processes the target result based on the traditional rule-based algorithm.
[0155] Method 4) Use the default strategy to process the target result.
[0156] Here, the second device sends a notification message to the first device to notify the first device to use the default value configured by the system or calculate a conservative value according to the confidence and the predicted value, and use the default value or conservative value as the target result. The second device processes the target result based on the default policy.
[0157] For the above method, the content included in the notification message sent by the second device to the first device can refer to the following Table 3.
[0158] Table 3: Parameters in the notification message
[0159]
[0160] The following is an example of the technical solution of the embodiment of the present application with reference to a specific application example.
[0161] Application Example 1
[0162] In this application example, the first device is a UE and the second device is a base station. Fig. 9 As shown, in Figure 7 Based on the processing flow shown, Fig. 9 Added the following steps:
[0163] S1: The base station sends a model prediction confidence threshold configuration request message to the UE.
[0164] S2: The UE sends a model prediction confidence threshold configuration response message to the base station.
[0165] For the specific implementation of the above steps S1 and S2, please refer to the above Figure 6 Description.
[0166] S3: UE uses the current model to make predictions.
[0167] Steps 701 to 704 after S3 can refer to the aforementioned Figure 7 Related description.
[0168] In this application example, when the UE side makes predictions through the AI algorithm and reports the results to the base station, the base station performs subsequent processing based on the reported results. For example, CSI prediction and RRM measurement prediction. During the initial configuration, the UE configures the model prediction confidence threshold to the base station through the air interface. The UE determines the confidence of each prediction. If it is lower than the threshold, or is lower than the threshold for n consecutive times (n is a preconfigured value), a new prediction model is considered, or the target result is obtained using a traditional non-AI algorithm, and the target result is reported to the base station for processing.
[0169] Application Example 2
[0170] In this application example, the first device is the destination base station, and the second device is the source base station. Fig.10 As shown, the following steps are included:
[0171] Step 1001: The model prediction confidence threshold is pre-configured on the source base station side.
[0172] Step 1002: The source base station sends UE information, including service information, location information, etc., to the destination base station.
[0173] Step 1003: The target base station uses the model to predict the load and service quality according to the UE information.
[0174] Step 1004: The destination base station sends the predicted target result and the confidence level corresponding to the target result to the source base station.
[0175] Step 1005: The source base station determines that the confidence level of the target result (n consecutive times) is lower than a threshold.
[0176] Step 1006: Option 1) the source base station uses the AI algorithm to process the target result; Option 2) the source base station does not use the AI algorithm to process the target result.
[0177] For option 1), this can be achieved by:
[0178] Method 1) The source base station selects other models to process the target results.
[0179] Here, the source base station selects a model with higher complexity and better performance (or a more advanced model) to process the target result.
[0180] Mode 2) The source base station and the destination base station update the model synchronously, and the source base station uses the updated model to process the target result.
[0181] Here, the source base station and the destination base station use a pairing model. If the source base station updates the model, the source base station sends a notification message to the destination base station to notify the destination base station of the pairing model update. The destination base station will also update the model and re-predict the target result through the updated model.
[0182] For option 2), this can be achieved by:
[0183] The source base station sends a notification message to the destination base station to notify the destination base station to recalculate the target result using a rule-based traditional algorithm. Accordingly, the source base station processes the target result based on the rule-based traditional algorithm.
[0184] Step 1007: The source base station determines whether the UE should be transferred according to the processing result.
[0185] This application example can be applied to load balancing scenarios. The destination base station will predict the service conditions of the UE after the UE is transferred based on the information provided by the source base station and the load, service, and channel conditions of the base station.
[0186] Fig.11 This is a schematic diagram of the structure of the prediction device provided in the embodiment of the present application. Figure 1 , applied to the first device, such as Fig.11 As shown, the prediction device comprises:
[0187] The first processing unit 1101 is used to determine that when the prediction confidence of the first model is lower than a threshold, a first processing method is used to obtain a target result; the first processing method includes at least one of the following: using the second model to predict and obtain the target result; using multiple models to predict and obtain multiple results, and determining the target result based on the multiple results; using a non-AI algorithm to obtain the target result; using a default value or a conservative value as the target result;
[0188] The first communication unit 1102 is configured to send a result message to the second device, where the result message includes the target result.
[0189] In some implementations, the first communication unit 1102 is configured to send a configuration request message to the second device and / or the third device; and receive a configuration response message sent by the second device and / or the third device, wherein the configuration response message includes the threshold.
[0190] In some implementations, the configuration response message further includes at least one of the following information corresponding to the threshold: a model identifier, a model type, and a model function identifier.
[0191] In some implementations, the threshold is configured according to at least one of the following dimensions: model type, model, model function.
[0192] In some implementations, the result message further includes at least one of the following information:
[0193] first information, the first information being used to indicate a processing method for obtaining the target result;
[0194] second information, where the second information is used to indicate a model identifier corresponding to a model for obtaining the target result;
[0195] third information, the third information being used to indicate a model function corresponding to the model for obtaining the target result;
[0196] fourth information, the fourth information being used to indicate a prediction confidence of the target result;
[0197] The fifth information is used to indicate whether the target result is obtained based on a non-AI algorithm or based on a default value or a conservative value.
[0198] Those skilled in the art should understand that Fig.11 The implementation functions of each unit in the prediction device shown can be understood by referring to the relevant description of the aforementioned method. Fig.11 The functions of each unit in the prediction device shown can be implemented by a program running on a processor, or can be implemented by a specific logic circuit.
[0199] Fig.12 This is a schematic diagram of the structure of the prediction device provided in the embodiment of the present application. Figure 2 , applied to the second device, such as Fig.12 As shown, the prediction device comprises:
[0200] The second communication unit 1201 is used to receive a result message sent by the first device, where the result message includes a target result;
[0201] The second processing unit 1202 is used to determine that when the prediction confidence of the target result is lower than a threshold, a second processing method is adopted; the second processing method includes at least one of the following: using a third model to process the target result; sending a notification message to the first device, the notification message is used to notify the first device to obtain the target result through an updated model, and / or notify the first device to obtain the target result by using a non-AI algorithm, and / or notify the first device to use a default value or a conservative value as the target result.
[0202] In some implementations, the notification message includes at least one of the following information:
[0203] Sixth information, where the sixth information is used to indicate a model identifier corresponding to the model updated by the first device;
[0204] seventh information, where the seventh information is used to indicate a model function corresponding to the model updated by the first device;
[0205] Eighth information, where the eighth information is used to instruct the first device to obtain the target result by adopting a non-AI algorithm or based on a default value or a conservative value;
[0206] Ninth information, the ninth information is used to indicate the effectiveness condition of the notification message.
[0207] In some implementations, the second communication unit 1201 is configured to receive a configuration request message sent by the first device; and send a configuration response message to the first device, wherein the configuration response message includes the threshold.
[0208] In some implementations, the configuration response message further includes at least one of the following information corresponding to the threshold: a model identifier, a model type, and a model function identifier.
[0209] In some implementations, the threshold is configured according to at least one of the following dimensions: model type, model, model function.
[0210] Those skilled in the art should understand that Fig.12 The implementation functions of each unit in the prediction device shown can be understood by referring to the relevant description of the aforementioned method. Fig.12 The functions of each unit in the prediction device shown can be implemented by a program running on a processor, or can be implemented by a specific logic circuit.
[0211] Fig.13 1300 is a schematic structural diagram of a communication device 1300 provided in an embodiment of the present application. The communication device may be a terminal device or a first device. Fig.13 The communication device 1300 shown includes a processor 1310, and the processor 1310 can call and run a computer program from a memory to implement the method in the embodiment of the present application.
[0212] Alternatively, if Fig.13 As shown, the communication device 1300 may further include a memory 1320. The processor 1310 may call and run a computer program from the memory 1320 to implement the method in the embodiment of the present application.
[0213] The memory 1320 may be a separate device independent of the processor 1310 , or may be integrated into the processor 1310 .
[0214] Alternatively, if Fig.13 As shown, the communication device 1300 may further include a transceiver 1330, and the processor 1310 may control the transceiver 1330 to communicate with other devices, specifically, may send information or data to other devices, or receive information or data sent by other devices.
[0215] The transceiver 1330 may include a transmitter and a receiver. The transceiver 1330 may further include an antenna, and the number of antennas may be one or more.
[0216] Optionally, the communication device 1300 may specifically be the first device of the embodiment of the present application, and the communication device 1300 may implement the corresponding processes implemented by the first device in each method of the embodiment of the present application, which will not be described in detail here for the sake of brevity.
[0217] Optionally, the communication device 1300 may specifically be the second device of the embodiment of the present application, and the communication device 1300 may implement the corresponding processes implemented by the second device in each method of the embodiment of the present application, which will not be described in detail here for the sake of brevity.
[0218] Fig.14 It is a schematic structural diagram of the chip of an embodiment of the present application. Fig.14 The chip 1400 shown includes a processor 1410, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.
[0219] Alternatively, if Fig.14 As shown, the chip 1400 may further include a memory 1420. The processor 1410 may call and run a computer program from the memory 1420 to implement the method in the embodiment of the present application.
[0220] The memory 1420 may be a separate device independent of the processor 1410 , or may be integrated into the processor 1410 .
[0221] Optionally, the chip 1400 may further include an input interface 1430. The processor 1410 may control the input interface 1430 to communicate with other devices or chips, and specifically, may obtain information or data sent by other devices or chips.
[0222] Optionally, the chip 1400 may further include an output interface 1440. The processor 1410 may control the output interface 1440 to communicate with other devices or chips, and specifically, may output information or data to other devices or chips.
[0223] Optionally, the chip can be applied to the first device in the embodiment of the present application, and the chip can implement the corresponding processes implemented by the first device in each method of the embodiment of the present application. For the sake of brevity, it will not be repeated here.
[0224] Optionally, the chip can be applied to the second device in the embodiments of the present application, and the chip can implement the corresponding processes implemented by the second device in each method of the embodiments of the present application. For the sake of brevity, they will not be repeated here.
[0225] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.
[0226] It should be understood that the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiment can be completed by the hardware integrated logic circuit or software instructions in the processor. The above processor can be a general processor, a digital signal processor (Digital Signal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to perform, or the hardware and software modules in the decoding processor are combined and performed. The software module can be located in a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, and other mature storage media in the art. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.
[0227] It can be understood that the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAM bus random access memory (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.
[0228] It should be understood that the above-mentioned memory is exemplary but not restrictive. For example, the memory in the embodiments of the present application may also be static random access memory (static RAM, SRAM), dynamic random access memory (dynamic RAM, DRAM), synchronous dynamic random access memory (synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (double data rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (synch link DRAM, SLDRAM) and direct memory bus random access memory (Direct Rambus RAM, DR RAM), etc. That is to say, the memory in the embodiments of the present application is intended to include but not limited to these and any other suitable types of memory.
[0229] An embodiment of the present application also provides a computer-readable storage medium for storing a computer program.
[0230] Optionally, the computer-readable storage medium can be applied to the first device in the embodiments of the present application, and the computer program enables the computer to execute the corresponding processes implemented by the first device in the various methods of the embodiments of the present application. For the sake of brevity, they are not repeated here.
[0231] Optionally, the computer-readable storage medium can be applied to the second device in the embodiments of the present application, and the computer program enables the computer to execute the corresponding processes implemented by the second device in the various methods of the embodiments of the present application. For the sake of brevity, they are not repeated here.
[0232] An embodiment of the present application also provides a computer program product, including computer program instructions.
[0233] Optionally, the computer program product can be applied to the first device in the embodiments of the present application, and the computer program instructions enable the computer to execute the corresponding processes implemented by the first device in the various methods of the embodiments of the present application. For the sake of brevity, they are not repeated here.
[0234] Optionally, the computer program product can be applied to the second device in the embodiments of the present application, and the computer program instructions enable the computer to execute the corresponding processes implemented by the second device in the various methods of the embodiments of the present application. For the sake of brevity, they are not repeated here.
[0235] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0236] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0237] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0238] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0239] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0240] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0241] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
Claims
1. A prediction method, It is characterized in that Applied to a first device, the method includes: When it is determined that the prediction confidence of the first model is lower than the threshold, a first processing method is used to obtain the target result; the first processing method includes at least one of the following: using the second model to predict and obtain the target result; using multiple models to predict and obtain multiple results, and determining the target result based on the multiple results; using a non-AI algorithm to obtain the target result; using a default value or a conservative value as the target result; A result message is sent to the second device, where the result message includes the target result.
2. The method according to claim 1, It is characterized in that The method further comprises: receiving a configuration request message sent by the second device; A configuration response message is sent to the second device, where the configuration response message includes the threshold.
3. The method according to claim 2, It is characterized in that The configuration response message further includes at least one of the following information corresponding to the threshold: model identifier, model type, and model function identifier.
4. The method according to any one of claims 1 to 3, It is characterized in that The threshold is configured according to at least one of the following dimensions: model type, model, model function.
5. The method according to any one of claims 1 to 3, It is characterized in that The result message also includes at least one of the following information: first information, the first information being used to indicate a processing method for obtaining the target result; second information, where the second information is used to indicate a model identifier corresponding to a model for obtaining the target result; third information, the third information being used to indicate a model function corresponding to the model for obtaining the target result; fourth information, the fourth information being used to indicate a prediction confidence of the target result; The fifth information is used to indicate whether the target result is obtained based on a non-AI algorithm or based on a default value or a conservative value.
6. A prediction method, It is characterized in that Applied to the second device, the method includes: Receiving a result message sent by the first device, wherein the result message includes a target result; When it is determined that the prediction confidence of the target result is lower than a threshold, a second processing method is adopted; the second processing method includes at least one of the following: using a third model to process the target result; sending a notification message to the first device, the notification message is used to notify the first device to obtain the target result through an updated model, and / or notify the first device to obtain the target result using a non-AI algorithm, and / or notify the first device to use a default value or a conservative value as the target result.
7. The method according to claim 6, It is characterized in that The notification message includes at least one of the following information: Sixth information, where the sixth information is used to indicate a model identifier corresponding to the model updated by the first device; seventh information, where the seventh information is used to indicate a model function corresponding to the model updated by the first device; Eighth information, where the eighth information is used to instruct the first device to obtain the target result by adopting a non-AI algorithm or based on a default value or a conservative value; Ninth information, the ninth information is used to indicate the effectiveness condition of the notification message.
8. The method according to claim 6, It is characterized in that The method further comprises: Sending a configuration request message to the first device and / or the third device; A configuration response message sent by the first device and / or the third device is received, where the configuration response message includes the threshold.
9. The method according to claim 8, It is characterized in that The configuration response message further includes at least one of the following information corresponding to the threshold: model identifier, model type, and model function identifier.
10. The method according to any one of claims 6 to 9, It is characterized in that The threshold is configured according to at least one of the following dimensions: model type, model, model function.
11. A prediction device, It is characterized in that Applied to a first device, the apparatus comprises: A first processing unit is used to determine that when the prediction confidence of the first model is lower than a threshold, a first processing method is used to obtain a target result; the first processing method includes at least one of the following: using the second model to predict and obtain the target result; using multiple models to predict and obtain multiple results, and determining the target result based on the multiple results; using a non-AI algorithm to obtain the target result; using a default value or a conservative value as the target result; The first communication unit is used to send a result message to the second device, where the result message includes the target result.
12. A prediction device, It is characterized in that Applied to a second device, the apparatus comprises: A second communication unit, configured to receive a result message sent by the first device, wherein the result message includes a target result; A second processing unit is used to determine that when the prediction confidence of the target result is lower than a threshold, a second processing method is adopted; the second processing method includes at least one of the following: using a third model to process the target result; sending a notification message to the first device, the notification message is used to notify the first device to obtain the target result through an updated model, and / or notify the first device to obtain the target result by using a non-AI algorithm, and / or notify the first device to use a default value or a conservative value as the target result.
13. A communication device, It is characterized in that include: A processor and a memory, the memory being used to store a computer program, the processor being used to call and run the computer program stored in the memory to execute the method according to any one of claims 1 to 10.
14. A computer-readable storage medium, It is characterized in that Used to store a computer program, wherein the computer program causes a computer to execute the method according to any one of claims 1 to 10.
15. A computer program product, It is characterized in that The method comprises computer program instructions which cause a computer to execute the method as claimed in any one of claims 1 to 10.
Citation Information
Cited By
Information transmission method and apparatus, device, readable storage medium, and program product
WO2025195313A1
Method and apparatus for wireless communication
WO2026188413A1
