Communication method and device

Abstract

A communication method and device, which are beneficial to reducing resource overhead caused by model training and management, and improving model management efficiency. The method comprises: a second communication device performs statistical analysis on model information of at least one first model associated with a second model, generates first information that includes the model information of the at least one first model and / or statistical information corresponding to the model information, and sends the first information to a first communication device; on the basis of the received first information, the first communication device uses the model information of the at least one first model or the statistical information corresponding to the model information as a reference experience to determine a management strategy for maintaining the second model.

Description

Communication method and apparatus

[0001] The present application claims priority to the Chinese Patent Application No. 202411399460.6, filed on September 30, 2024, and entitled "Communication method and apparatus", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] Embodiments of the present application relate to the field of communication, and in particular to a communication method and apparatus. BACKGROUND

[0003] In a wireless communication network, as the diversification of service requirements and the enhancement of network functions, service implementation, network planning, configuration, and resource scheduling also become increasingly complex. For example, service implementation, network planning, configuration, and resource scheduling may involve modulation, coding, transmitters, receivers, multi-antenna technology, or positioning technology in a wireless communication system. Network devices (such as base stations, terminals, etc.) in a wireless communication network can implement, for example, signal modulation and demodulation, information encoding and decoding, channel state information (CSI) feedback, or beam management (BM) schemes by performing related operations.

[0004] Currently, in the process of implementing the above schemes through a wireless communication network, a terminal can implement related operations in the above schemes by training a model (which can also be referred to as a characteristic, function, or algorithm) through collection of a large amount of training data, but this approach can result in a large resource overhead. SUMMARY

[0005] The present application provides a communication method and apparatus, which can reduce the resource overhead caused by model training and management, and improve the efficiency of model management.

[0006] In a first aspect, a communication method is provided. The method can be performed by a first communication apparatus, or by a component (such as a processor, circuit, chip, or chip system, etc.) applied to the first communication apparatus, or by a logic node, logic module, or software that can implement all or part of the functions of the first communication apparatus. The method includes: receiving first information from a second communication apparatus, the first information including model information of at least one first model and / or statistical information corresponding to the model information of the at least one first model; and determining a management strategy of a second model according to the first information, the first model and the second model having an association relationship, and the second model being a model maintained by the first communication apparatus.

[0007] Based on the scheme, in the process of determining the management policy of the second model to be maintained, the first communication device receives first information from the second communication device, the first information including model information of at least one first model having an association relationship with the second model and / or statistical information corresponding to the model information of the at least one first model, and then the first communication device determines the management policy adopted in the process of maintaining the second model according to the first information. That is, in the process of determining the management policy of the second model, the first communication device takes the model information of the first model having an association relationship with the second model or the statistical information corresponding to the model information of the first model as prior information, and determines the management policy of the model according to the prior information. In the process of determining the management policy of the model, a large amount of training data does not need to be collected and the data statistical characteristics of the training data do not need to be analyzed, which is beneficial to reduce the resource overhead required by the first communication device in the process of determining the management policy of the second model to be maintained, and improve the efficiency of the first communication device in maintaining the second model.

[0008] In a second aspect, a communication method is provided. The method can be executed by a second communication device, or by a component (such as a processor, a circuit, a chip, or a chip system, etc.) applied to the second communication device, or by a logic node, a logic module, or software that can realize all or part of the functions of the second communication device. The method includes determining first information, the first information including model information of at least one first model and / or statistical information corresponding to the model information of the at least one first model; and sending the first information to a first communication device; wherein the first information is used to determine a management policy of a second model, and the first model has an association relationship with the second model. The technical effects brought by the second aspect can refer to the technical effects brought by the first aspect, which will not be repeated here.

[0009] In combination with the first aspect and the second aspect, in a possible design, the model information of the first model indicates at least one of the following: performance indicator information of the first model; a state of the first model; whether the first model is changed; a time at which a first data set starts to be collected; a time at which the collection of the first data set is completed; a size of the first data set; generalization range information of the first model; model complexity information of the first model; and wherein the first data set is used to train or change the first model.

[0010] Based on the scheme, the first communication device can accurately obtain dynamic management information and static properties of the first model in the process of model management according to the model information of each first model, so as to facilitate the first communication device to take the association information of other models as prior experience to guide the model management and the management policy decision of the model.

[0011] In combination with the first aspect and the second aspect, in a possible design, the generalization range includes at least one cell to which the first model is applicable, and / or at least one network configuration to which the first model is applicable.

[0012] With reference to the first aspect and the second aspect, in a possible design, whether the first model is changed comprises whether the first model is changed after the first time point.

[0013] Based on this scheme, the first communication apparatus can acquire whether the first model is changed after the reference time point (the first time point), which is beneficial for the first communication apparatus to determine whether to change the second model according to the change trend of the first model after the first time point.

[0014] With reference to the first aspect and the second aspect, in a possible design, in the case where the first model is changed, the generalization range information of the first model indicates a generalization range before the first model is changed and / or a generalization range after the first model is changed; in the case where the first model is not changed, the generalization range information of the first model indicates a current generalization range of the first model.

[0015] Based on this scheme, the first communication apparatus can accurately acquire the generalization range change trend of the first model that is changed in generalization range and the generalization range of the first model that is not changed in generalization range, which is beneficial for determining the target generalization range of the second model, thereby determining the generalization range after model updating or retraining, or selecting the second model with a suitable generalization range as the model to be activated.

[0016] With reference to the first aspect and the second aspect, in a possible design, in the case where the first model is changed, the size of the first data set indicates a size of a data set used for changing the first model; in the case where the first model is not changed, the size of the first data set indicates a size of a data set used for training the first model.

[0017] Based on this scheme, the first communication apparatus can accurately acquire the target size of the first data set used in the process of changing or training the first model, thereby effectively controlling the data amount of training data collected by the first communication apparatus in the process of changing or training the second model, and reducing the probability of collecting meaningless training data.

[0018] With reference to the first aspect and the second aspect, in a possible design, in the case where the first model is changed, the model complexity information of the first model indicates a model complexity before the first model is changed and / or a model complexity after the first model is changed; in the case where the first model is not changed, the model complexity information of the first model indicates a model complexity of the first model.

[0019] Based on the scheme, the first communication device accurately obtains the model complexity change trend of the first model that has changed in model complexity and the model complexity of the first model that has not changed in model complexity, which is beneficial to determine the target model complexity of the second model, so as to determine the model complexity of the model after model updating or retraining, or select the second model with appropriate model complexity as the model to be activated.

[0020] In combination with the first aspect and the second aspect, in a possible design, the statistical information corresponding to the model information of the at least one first model indicates at least one of the following: a proportion of the first model that has changed in the at least one first model; a proportion of the first model whose performance indicator is greater than or equal to the first performance indicator in the at least one first model; a proportion of the first model whose performance indicator is less than the first performance indicator in the at least one first model; a proportion of the first model whose generalization range has changed in the at least one first model; a proportion of the first model whose applicable network configuration changes from the first network configuration to the second network configuration in the at least one first model; a proportion of the first model whose applicable network configuration changes from the second network configuration to the first network configuration in the at least one first model; a proportion of the first model whose performance indicator has changed in the at least one first model; a proportion of the first model whose performance indicator has not changed in the at least one first model; a proportion of the first model whose performance indicator has not changed and whose applicable network configuration is the first network configuration in the at least one first model; a proportion of the first model whose performance indicator has not changed and whose applicable network configuration is the second network configuration in the at least one first model; a proportion of the first model that is in the active state in the at least one first model; a proportion of the first complexity in the model complexity corresponding to the at least one first model, the first complexity being any model complexity in the model complexity corresponding to the at least one first model; a proportion of the first size in the size of the first data set corresponding to the at least one first model, the first size being the size of any first data set in the first data set corresponding to the at least one first model, the first data set being used to change the first model; and wherein each network configuration parameter in the first network configuration has one value, and at least part of the network configuration parameters in the second network configuration have multiple values.

[0021] Based on the scheme, the first communication device can accurately obtain the change trend of the first model according to the first information, which is beneficial to the first communication device to accurately determine the management strategy for maintaining the second model, and in the case of changing or retraining the second model, the target model parameter of the second model can be determined according to the first information, thereby improving the efficiency of maintaining the second model.

[0022] In combination with the first aspect and the second aspect, in a possible design, the proportion of the first model that has changed in the at least one first model includes: a proportion of the first model that has changed after the first time in the at least one first model.

[0023] With reference to the first aspect and the second aspect, in a possible design of the first aspect and the second aspect, the performance indicator information of the first model is used to indicate a performance indicator of the first model.

[0024] With reference to the first aspect and the second aspect, in a possible design of the first aspect and the second aspect, the performance indicator information of the first model is further used to indicate time information that the first model meets the first performance indicator and / or time information that the first model does not meet the first performance indicator; the time information that the first model meets the first performance indicator indicates at least one of a start time, an end time, or a duration that the first model meets the first performance indicator; and the time information that the first model does not meet the first performance indicator indicates at least one of a start time, an end time, or a duration that the first model does not meet the first performance indicator.

[0025] Based on this scheme, the first communication device can obtain a time period and / or a time proportion that the first model meets the first performance indicator, which is beneficial to the first communication device to use different management strategies to maintain the second model in different application time periods according to the statistical information of the model information, and improves the efficiency of maintaining the second model and the accuracy of management strategy decision-making.

[0026] With reference to the first aspect and the second aspect, in a possible design of the first aspect and the second aspect, the state of the first model includes at least one of the following: a support state, an available state, a suitable state, an active state, or a configuration state.

[0027] With reference to the first aspect, the communication method further includes: sending, to the second communication device, second information, where the second information is used to request the first information. Correspondingly, with reference to the second aspect, the communication method further includes: receiving, from the first communication device, second information, where the second information is used to request the first information.

[0028] With reference to the first aspect, the communication method further includes: sending, to the second communication device, model information of the second model. Correspondingly, with reference to the second aspect, the communication method further includes: receiving, from the first communication device, model information of the second model.

[0029] With reference to the first aspect and the second aspect, in a possible design of the first aspect and the second aspect, the model information of the second model is model information of the second model after the second time, or the model information of the second model is model information of the second model after being changed.

[0030] With reference to the first aspect and the second aspect, in a possible design of the first aspect and the second aspect, the at least one first model is a model maintained by at least one third communication device, and the network configuration to which the first model is applicable includes a network configuration to which the second model is applicable.

[0031] Based on the scheme, the application scenarios (corresponding network configurations) between the first model and the second model have a strong correlation, and the model information of the first model can effectively guide the management policy decision of the first communication device in the process of maintaining the second model.

[0032] In a possible design in combination with the first aspect and the second aspect, the management policy includes one of the following: model retraining, model updating, model activation, or model deactivation.

[0033] In a third aspect, a communication device is provided for implementing various methods. The communication device includes modules, units, or means for implementing the corresponding functions of the methods, which can be implemented by hardware, software, or by executing corresponding software with hardware. The hardware or software includes one or more modules or units corresponding to the functions.

[0034] In some possible designs, the communication device can include a processing module and a transceiver module. The processing module can be used to implement the processing functions in any of the aspects and any possible implementation manners thereof. The transceiver module can include a receiving module and a sending module, which are used to implement the receiving functions and the sending functions in any of the aspects and any possible implementation manners thereof.

[0035] In some possible designs, the transceiver module can be composed of a transceiver circuit, a transceiver, a transceiver, or a communication interface.

[0036] In a fourth aspect, a communication device is provided, including a processor and a memory; the memory is used to store computer instructions, when the processor executes the instructions, to make the communication device execute the method in any aspect.

[0037] In a fifth aspect, a communication device is provided, including a processor and a communication interface; the communication interface is used to communicate with modules outside the communication device; the processor is used to execute computer programs or instructions, to make the communication device execute the method in any aspect.

[0038] In a sixth aspect, a communication device is provided, including at least one processor; the processor is used to execute computer programs or instructions stored in a memory, to make the communication device execute the method in any aspect. The memory can be coupled with the processor, or can be independent of the processor.

[0039] In a seventh aspect, a communication device (for example, the communication device can be a chip or a chip system) is provided, including a processor, used to implement the functions involved in any of the first aspect and the second aspect.

[0040] In some possible design, the communication apparatus includes a memory, configured to store necessary program instructions and data.

[0041] In some possible design, the apparatus is a chip system, which can be composed of a chip or include a chip and other discrete devices.

[0042] It can be understood that the communication apparatus provided in the third aspect to the seventh aspect can be the first communication apparatus in the first aspect, or can be a module or unit (for example, a chip, or a chip system, or a circuit) corresponding to the method / operation / step / action described in the first aspect, or can be a module or unit capable of matching the first communication apparatus, or can also be a logic node, a logic module or software capable of realizing all or part of the function of the first communication apparatus; or the communication apparatus can be the second communication apparatus in the second aspect, or can be a module or unit (for example, a chip, or a chip system, or a circuit) corresponding to the method / operation / step / action described in the second aspect, or can be a module or unit capable of matching the second communication apparatus, or can also be a logic node, a logic module or software capable of realizing all or part of the function of the second communication apparatus.

[0043] It can be understood that when the communication apparatus in any one of the third aspect to the seventh aspect is a chip, the sending action / function of the communication apparatus can be understood as outputting information, and the receiving action / function of the communication apparatus can be understood as inputting information.

[0044] The eighth aspect provides a computer readable storage medium, which stores a computer program or instructions, and when the computer program or instructions are executed on the communication apparatus, the communication apparatus can execute the method in any one of the first aspect and the second aspect.

[0045] The ninth aspect provides a computer program product including instructions, and when the computer program product is executed on the communication apparatus, the communication apparatus can execute the method in any one of the first aspect and the second aspect.

[0046] The tenth aspect provides a communication system, which includes the first communication apparatus and the second communication apparatus. The first communication apparatus is configured to execute the method in the first aspect and any possible design thereof, and the second communication apparatus is configured to execute the method in the second aspect and any possible design thereof.

[0047] The technical effects brought by any one of the third aspect to the tenth aspect can be referred to the technical effects brought by different design manners of the first aspect and the second aspect, which will not be repeated here. BRIEF DESCRIPTION OF DRAWINGS

[0048] FIG. 1 is a schematic diagram of data feature randomness provided by the present application;

[0049] FIG. 2 is a schematic diagram of an architecture of a communication system provided by the present application;

[0050] FIG. 3 is a schematic diagram of an architecture of another communication system provided by the present application;

[0051] FIG. 4 is a schematic diagram of a structure of an O-RAN system provided by the present application;

[0052] FIG. 5 is a flowchart of a communication method provided by the present application;

[0053] FIG. 6 is a schematic diagram of an application flow of a communication method provided by the present application;

[0054] FIGS. 7-9 are schematic diagrams of structures of communication apparatuses provided by the present application. DETAILED DESCRIPTION

[0055] In the description of the present application, unless otherwise specified, “ / ” represents that the objects before and after the “ / ” are in an “or” relationship, for example, A / B can represent A or B; “and / or” in the present application is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural.

[0056] In the description of the present application, unless otherwise specified, “multiple” means two or more than two. “At least one of the following” or the like means any combination of the items, including any combination of single item or multiple items. For example, at least one of a, b, or c can represent: a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0057] In addition, in order to facilitate the clear description of the technical solutions of the embodiments of the present application, in the embodiments of the present application, the same items or similar items with basically the same functions and effects are distinguished by using “first”, “second”, etc. The skilled in the art can understand that “first”, “second”, etc. do not limit the quantity and execution order, and “first”, “second”, etc. also do not necessarily mean different.

[0058] In the present embodiments, the word "exemplary" or "for example" is used to mean "an example of" rather than "an ideal”. Any embodiment or design described herein as "exemplary" or "for example" is not necessarily to be construed as preferred or advantageous over other embodiments or designs. Rather, the exemplary or for example embodiments are to be seen as examples—only—of how the related concept could be implemented, and how it can be used in practice.

[0059] It can be understood that, the "embodiments" mentioned in the specification throughout mean that the specific features, structures or characteristics related to the embodiments are included in at least one embodiment of the present application. Therefore, the various embodiments throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It can be understood that, in various embodiments of the present application, the size of the sequence number of each process does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0060] It can be understood that, in the present application, "…", "if" and "when" all refer to the corresponding processing under certain objective circumstances, not the time limit, and do not require judgment action when implementing, nor mean that there are other limitations.

[0061] It can be understood that, in some scenarios, some optional features in the embodiments of the present application can be implemented independently without relying on other features, such as the scheme currently based on, to solve the corresponding technical problems and achieve the corresponding effects. In some scenarios, it can also be combined with other features according to demand. Correspondingly, the device given in the embodiments of the present application can also realize these features or functions, which will not be described here.

[0062] In the present application, except for special description, the same or similar parts of each embodiment can be mutually referred. In various embodiments of the present application, if there is no special description and no logical conflict, the terms and / or descriptions of different embodiments are consistent and can be mutually referred. The technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationship. The following description of the embodiments of the present application does not constitute a limitation on the protection scope of the present application.

[0063] In order to facilitate the understanding of the technical scheme of the embodiments of the present application, first, the brief introduction of the related technology of the present application is as follows.

[0064] 1. Artificial intelligence / machine learning model:

[0065] In a wireless communication network, as the diversification of service requirements and the enhancement of network functions, service implementation, network planning, configuration, and resource scheduling also become increasingly complex. For example, service implementation, network planning, configuration, and resource scheduling can involve modulation, coding, transmitters, receivers, multi-antenna technology, or positioning technology in a wireless communication system. Network devices (such as base stations, terminals, etc.) in a wireless communication network can implement one or more wireless communication functions / operations, such as signal modulation and demodulation, information encoding and decoding, CSI feedback, channel equalization, channel estimation, pilot generation, precoding, resource mapping, resource demapping, interference suppression, interference estimation, interference prediction, BM, or mobility management, by performing related operations.

[0066] For example, a communication device can perform the technical solutions in the above-mentioned various service scenarios through traditional algorithms, or can apply artificial intelligence (AI) / machine learning (ML) technology to the technical solutions in the above-mentioned service scenarios. AI / ML technology refers to training a model through related data, and then using the trained model to achieve a specific purpose (such as a wireless communication function / operation). The purpose that the model can achieve is related to the data used during training.

[0067] The training data includes data obtained in a communication network, such as signal processing information, channel information, and radio frequency information. The signal processing information includes information generated in a baseband signal processing process, such as at least one of information generated in a signal sampling, modulation, demodulation, coding, decoding, precoding, resource mapping, or digital filtering process. The channel information includes information corresponding to a channel environment, such as at least one of power information, amplitude information, phase information, time delay information, multipath information, signal propagation time information, distance information, speed information, large-scale channel information, small-scale channel information, channel scattering information, or line-of-sight (LOS) / non-line-of-sight (NLOS) information. For example, the data of the CSI use case and the BM use case is embodied as channel information CSI, and the data in the positioning use case is embodied as channel information and / or position information. The radio frequency information includes information generated in an analog processing process, such as at least one of information generated in a digital-to-analog conversion, analog-to-digital conversion, digital pre-distortion, frequency conversion, radio frequency modulation, radio frequency demodulation, power amplification, low-noise amplification, analog filtering, or duplex processing.

[0068] For example, taking the service scenario of AI / ML-based CSI feedback as an example, the service scenario of AI / ML-based CSI feedback includes AI / ML-based CSI compression and AI / ML-based CSI prediction. AI / ML-based CSI compression refers to that a terminal compresses downlink CSI (measured by the terminal) through a model trained based on AI / ML technology. Then, the terminal can send the compressed CSI to a network device through an air interface. The network device restores (decompresses) the CSI through a model trained based on AI / ML technology. Compared with a traditional compression algorithm, the AI / ML-based compression algorithm has a higher compression rate and better CSI restoration capability. Therefore, the terminal can feed back more CSI through a smaller air interface overhead, so that the network device can more accurately perform downlink precoding. AI / ML-based CSI prediction refers to that a network device predicts downlink CSI at a future time through a model trained based on AI / ML technology and downlink CSI at a current / historical time, and then performs precoding according to the predicted CSI. The CSI predicted by this scheme is more matched to the channel state when downlink data is scheduled, so it can overcome the problem of channel aging and achieve more accurate downlink precoding. The training data of the model involved in the service scenario of AI / ML-based CSI feedback includes CSI.

[0069] For example, taking the service scenario of AI / ML-based BM as an example, a network device and a terminal device can predict a transmission beam and / or a reception beam through AI / ML technology, for example, infer a small number of beam scanning results through a model trained based on AI / ML technology to obtain an optimal beam. Compared with a traditional scheme in which a large number of beams need to be scanned to obtain an optimal beam, AI / ML-based beam prediction can reduce the processing overhead of beam scanning. For example, a terminal can scan a small number of beams, and then predict an optimal beam from a large number of candidate beams through a model trained based on AI / ML technology, so it is not necessary to scan all candidate beams, reducing the overhead. The small number of beams scanned by the terminal can be sparse beams or wide beams, and the candidate beams can be dense beams or narrow beams. The terminal can input the beam information scanned at a current / historical time into the model to predict an optimal beam at a future time, so it is not necessary to perform beam scanning again at the future time, thereby improving the efficiency of beam scanning. The training data of the model involved in the service scenario of AI / ML-based BM includes beam information, for example, beam identity document (ID) and / or reference signal receiving power (RSRP) corresponding to the beam, and other beam-related information.

[0070] Taking a business scenario of AI / ML-based positioning as an example, a communication apparatus inputs channel information into a model trained based on AL / ML technology, and obtains intermediate parameters required for positioning through reasoning, or directly obtains position coordinate values. Compared with a traditional positioning algorithm, the intermediate parameters or the positioning coordinate values obtained based on AL / ML are more accurate. In the business scenario of AL / ML-based positioning, training data of the model involved includes channel information and / or position information, and the channel information includes power information, phase information, time delay information, distance information, speed information, channel scattering information, and LOS / NLOS information and other channel-related information.

[0071] 2. Model management

[0072] Since the model performance on the terminal side is affected by various factors, for example, network configuration on the base station side, external environment, and terminal internal conditions, etc., the terminal needs to determine a suitable management strategy to maintain the model according to the actual performance of the model and the current data characteristics, so that the performance indicators of the model can meet the preset performance indicator requirements.

[0073] Among them, the network configuration on the base station side mainly includes private configuration and radio resource control (RRC) configuration.

[0074] For example, the RRC configuration includes reference signal configuration, data characteristic configuration, quasi-collocation (QCL) assumption, and beam configuration. The private configuration includes base station height, inter-site distance (ISD), base station antenna form, antenna downtilt angle, and mapping relationship between beam and radio frequency antenna port, etc.

[0075] For example, the reference signal configuration is configuration information related to channel state information reference signal (CSI-RS), such as CSI-RS resource configuration and CSI reporting configuration, etc.; the data characteristic configuration is a data characteristic identifier; and the beam configuration is the mapping relationship between beams, beam set, and beam list.

[0076] For example, the external environment mainly refers to the channel condition in which the terminal is located. For example, the Doppler spread of the channel in which the terminal is located, the multi-path transmission delay, the channel interference intensity, and the received signal strength, etc. The internal conditions of the terminal mainly include the power, the remaining storage space, and the computing power of the terminal, etc. In the case that the terminal is in a situation of insufficient power, insufficient storage space, or insufficient computing power, it is easy to appear the problem that part of the management strategies cannot be supported or there is a conflict between part of the management strategies.

[0077] It should be understood that the data feature configuration in the embodiments of the present application can also be referred to as a data feature or a data classification feature, wherein the "feature" can also be described as a condition, a situation, an environment or a circumstance, a type, a status, or a data distribution. The feature in other terms "xx feature" (for example, a reference feature, a physical feature, or a feature of the xth data, etc.) in the embodiments of the present application can also be replaced similarly.

[0078] For example, the data feature identifier can also be referred to as an associated ID, a data ID, a dataset ID, a data categorization ID, or a property ID.

[0079] The data feature identifier can be used by the terminal to classify the measurement data. The measurement data is used to determine the model input, or to determine the training data (for example, the model output, the label) used for model training, and can also be used to implement related operations of model inference. For example, the terminal can assume that the downlink transmission beam set / list has the same or similar data features under the same data feature identifier.

[0080] Currently, the commonly used model performance monitoring is achieved by comparing the predicted value / inferred value of the model for a certain parameter with the true value. For example, in the business scenario of the AI / ML-based BM, the Top-K accuracy predicted by the model can be compared with the true measurement value. The Top-K can be understood as the proportion of the correct label contained in the top K results with the largest probability in the prediction result. According to this monitoring method, it can only be determined whether the model performance changes, but the reason for the change of the model performance cannot be determined, and thus the model performance cannot be used by the terminal to determine the correct management strategy.

[0081] As a possible implementation, the terminal can trigger data collection and data analysis when the model performance changes (for example, the model performance is lower than a preset threshold), and determine the relationship between the data statistical features of the training data currently collected and the data statistical features of the training data set used in the model generation process to determine the management strategy used by the model for subsequent maintenance.

[0082] For example, the data statistical features can include correlation, value range, statistical distribution, speed of change over time, periodicity, value probability, mean, or variance, and other statistical information.

[0083] That is, the current commonly used model management process includes the following steps:

[0084] Step 1, collect training data.

[0085] In the case that the performance of the maintained current model changes, the terminal triggers the process of collecting a large amount of training data, and the collection and storage of training data are performed through signaling interaction with the base station.

[0086] Step 2, analyze the data statistical characteristics of the training data and determine the management strategy.

[0087] After the terminal completes the collection of a large amount of training data, it performs data analysis on the collected training data, obtains the data statistical characteristics of the collected training data, and then compares and analyzes the data statistical characteristics of the collected training data with the data statistical characteristics of the training data set used for training the current model, to determine whether the data statistical characteristics of the collected training data change compared with the data statistical characteristics of the training data set. In the case that the data statistical characteristics change, the degree of change and the duration of change of the data statistical characteristics are obtained. Then the terminal determines the management strategy according to the change of the data statistical characteristics.

[0088] This way relies on the collection of a large amount of training data, so it will increase the overhead for collecting training data (for example, the overhead of reference signal signaling / resources in the data collection process, the overhead of terminal side data measurement / storage, the power consumption overhead of base station side data transmission or terminal side data reception, etc.).

[0089] In addition, in the case of model retraining or model updating of the current model, it can also be determined whether to store the configuration parameters of the current model according to whether the data characteristics will return to the data characteristics corresponding to the training data set. For example, in the case that the preset indication information from the base station indicates that the data characteristics will return to the data characteristics corresponding to the training data set after a certain time, the configuration parameters of the current model can be stored.

[0090] 3. Model management strategy:

[0091] Currently, the management strategy of the terminal side model mainly includes: model training, model retraining, model updating, model selection, model activation, model deactivation, model switching, model monitoring, model inference, data collection and data analysis, etc.

[0092] For example, model training can be understood as generating a model for implementing a specific communication function based on training data; model retraining can be understood as regenerating a model for implementing a specific communication function based on training data; model updating can be understood as further training a current model based on training data, updating model parameters and / or architecture; model selection can be understood as selecting a model to be activated from a model for implementing a specific communication function before model activation; model activation can be understood as enabling the selected model and executing a specific communication function through the model; model deactivation can be understood as ending the enablement of a current model and no longer continuing to execute a specific communication function through the current model; model switching can be understood as replacing a model executing a current communication function from a current model to another model; model monitoring can be understood as monitoring performance indicators and other data of a current model; data collection can be understood as obtaining training data in the process of training, updating or retraining a model; and data analysis can be understood as analyzing and / or comparing data features of collected training data.

[0093] As a possible implementation, in the process of maintaining a current model in an activated state by a terminal, a terminal with a degraded model performance can trigger a data collection and data analysis process. In the case of detecting that a data statistical feature change is a transient disturbance, the terminal can not perform model updating or model retraining for a short period of time, and model deactivation can be used as a model management strategy. In the case of detecting that a data statistical feature change is a quasi-static statistical feature change, model retraining can be used as a model management strategy, or model updating can be used as a model management strategy according to the degree of change in the model statistical feature.

[0094] For example, in the case where the similarity between the data statistical feature of the collected training data and the data statistical feature of the training data set is greater than or equal to a given threshold, the current model is updated according to the collected training data to obtain a model that meets the performance indicator requirement under the current data, without the need to retrain a new model. In the case where the similarity between the data statistical feature of the collected training data and the data statistical feature of the training data set is less than a given threshold, a new model is retrained according to the collected training data.

[0095] Here, the data statistical feature change being transient can be understood as the duration of the data statistical feature change being less than or equal to a preset time length, or can also be understood as a change in a specific data statistical feature. The given threshold corresponding to the data statistical feature and the preset time length can be pre-defined by a protocol, or can be pre-determined between a terminal and a base station or a model management device.

[0096] For example, taking the training data as the channel information as an example, with reference to FIG. 1, the information related to the channel mainly includes: path loss, shadow fading, angle of arrival (AOA) and direction of arrival (DOA) of direct and reflected paths, phase delay power, multipath fading, beam tracking, CSI caused by Doppler, and reflection, diffuse reflection and occlusion caused by randomly appearing objects.

[0097] Among them, the randomness of path loss, shadow fading, angle of arrival of direct and reflected paths, and phase delay power is weak, which can be regarded as quasi-static data, that is, the channel disturbance caused by path loss and other factors belongs to the change of quasi-static statistical characteristics; multipath fading, beam tracking and CSI caused by Doppler have moderate randomness, which belong to the change of dynamic characteristics; the change of channel amplitude caused by reflection, diffuse reflection and occlusion caused by randomly appearing objects has strong randomness, that is, the change of channel amplitude caused by random objects belongs to instantaneous disturbance.

[0098] It should be understood that the model involved in the embodiments of the present application can be described as a function (such as an artificial intelligence (AI) function or a machine learning (ML) function), a characteristic or an algorithm, etc. The model includes, for example, an AI model or an ML model, etc. In order to simplify the description, the model / function / characteristic / algorithm in the present application is referred to as a model, that is, the model in the present application can be replaced by an algorithm / function / characteristic.

[0099] That is, in the model maintenance process on the terminal side, in the case of performance degradation of the model, the terminal needs to collect a large amount of training data, and determine the management strategy for model maintenance according to the relationship between the data statistical characteristics of the collected training data and the data statistical characteristics of the training data set of the training model. The determination of the management strategy of the model or the model maintenance will bring great resource overhead, and the model maintenance efficiency is low.

[0100] Based on this, the embodiment of the present application provides a communication method, in the process of determining the management policy of the second model maintained by the first communication device, the first communication device receives the first information from the second communication device, the first information includes the model information of at least one first model associated with the second model and / or the statistical information corresponding to the model information of the at least one first model, and then the first communication device determines the management policy used in the process of maintaining the second model according to the first information. That is, in the process of determining the management policy of the second model, the first communication device takes the model information of the first model associated with the second model or the statistical information corresponding to the model information of the first model as prior information, and determines the management policy of the model according to the prior information. In the process of determining the management policy of the model, it is not necessary to collect a large amount of training data and analyze the data statistical characteristics of the training data, which is beneficial to reduce the resource overhead required by the first communication device in the process of determining the management policy of the second model, and improve the efficiency of the first communication device in maintaining the second model.

[0101] The technical scheme of the embodiment of the present application can be applied to various communication systems, which can be a third generation partnership project (3GPP) communication system, for example, a long term evolution (LTE) system, a fourth generation (4G) system, a new radio (NR) system, a fifth generation (5G) system, a system of mixed networking of LTE and 5G, a communication and perception integrated system, a non-terrestrial network (NTN), a device-to-device (D2D) communication system, a vehicle to everything (V2X) communication system, a machine-type communication (MTC) system, an internet of things (IoT) system, or other future communication systems. The communication system can also be a non-3GPP communication system, which is not limited.

[0102] Among the above, the communication system applicable to the present application is only an example, and the communication system applicable to the present application is not limited thereto, and the communication system provided by the present application does not cause any limitation to the scheme of the present application. It is uniformly explained here that the following will not be described in detail.

[0103] FIG. 2 shows a possible, non-limiting system diagram. As shown in FIG. 2, the communication system 10 includes a radio access network (RAN) 100 and a core network (CN) 200. The RAN 100 includes at least one RAN node (e.g., 110a and 110b in FIG. 2, collectively referred to as 110) and at least one terminal (e.g., 120a-120j in FIG. 2, collectively referred to as 120). Other RAN nodes, such as wireless relay devices and / or wireless backhaul devices (not shown in FIG. 2), etc., can also be included in the RAN 100. The terminal 120 is connected to the RAN node 110 in a wireless manner. The RAN node 110 is connected to the core network 200 in a wireless or wired manner. The core network node in the core network 200 and the RAN node 110 in the RAN 100 can be different physical devices respectively, or can be the same physical device integrated with the logical functions of the core network and the logical functions of the radio access network.

[0104] In a possible implementation, the core network node can refer to a device in the core network 200 that provides service support for the terminal 120. In the embodiments of the present application, the core network node in the core network 200 includes a sensing function (SF) network element, which is mainly used to implement sensing functions, such as sensing control functions and / or sensing calculation functions. Further, the SF network element can also support sensing billing functions when the terminal 120 and / or the RAN node 110 perform sensing. For example, the sensing control function can include determining sensing devices, sensing nodes, etc. The sensing device can be understood as a device that transmits and / or receives sensing signals, and further performs corresponding signal processing on the received echo signals to obtain sensing measurement data. For example, the sensing device can be the RAN node 110 or the terminal 120, etc. The sensing node can refer to a network node participating in the sensing service process in the wireless network. The sensing calculation function can include performing corresponding signal processing on the echo signals received by the sensing device to obtain sensing measurement data, and further processing the sensing measurement data and application information to obtain sensing results, etc.

[0105] For example, the SF network element can also be referred to as a communication device, for example, the SF network element can be understood as a communication device with core network sensing functions. In addition, the SF network element can also be referred to as a sensing server, etc., without limitation.

[0106] In a possible scenario, the functions of the SF network element can be implemented by a network data analysis function (NWDAF) network element, or the SF network element and the NWDAF network element can be combined.

[0107] Optionally, in addition to the SF network element, the core network nodes in the core network 200 can also include at least one of the following: an access and mobility management function (AMF) network element, a session management function (SMF) network element, a user plane function (UPF) network element, a policy control function (PCF) network element, a unified data management (UDM) network element, an application function (AF) network element, a network exposure function (NEF) network element, a network slice selection function (NSSF) network element, or a location management function (LMF) network element, etc. Of course, the core network 200 can also include other core network nodes, which are not limited.

[0108] The AMF network element is a network element deployed in the core network 200, which provides mobility management and connection management for the network, such as user location update, user registration network, user handover, etc. The AMF network element can be used as an intermediate route of the LMF, SMF and RAN 100. The SMF network element is mainly responsible for session management in the mobile network, such as session establishment, modification, release, etc. The UPF network element is a functional network element of the user plane, which is mainly responsible for connecting external networks and processing user packets, such as forwarding, charging, etc. The PCF network element is mainly responsible for providing policies to the AMF and SMF, such as quality of service (QoS) policies, slice selection policies, etc. The UDM network element is used to store user data, such as subscription information, authentication / authorization information, etc. The AF network element is responsible for providing services to the 3GPP network. The NEF network element is mainly used to open the capabilities of each network function and is responsible for converting internal and external information. The LMF network element is a device or component deployed in the core network 200, which provides positioning functions for the terminal 120, for example, the LMF network element can initiate a positioning process and perform positioning on a specific terminal.

[0109] It should be noted that the network element in the present application can also be referred to as an entity or a functional entity, for example, the SF network element can also be referred to as an SF entity or an SF functional entity. In addition, the above-mentioned AMF network element, SMF network element, UPF network element, PCF network element, UDM network element, AF network element, NEF network element, and LMF network element can also have other names in future communication systems, which are not limited in the present application.

[0110] In a possible implementation, the RAN 100 can be a 3rd generation partnership project (3GPP) related cellular system, e.g., a 4G, 5G mobile communication system, or a future

[0111] The RAN nodes 110, which can also be referred to as access network devices, RAN entities or access nodes, etc., form part of the communication system to enable wireless access to the communication system. The RAN nodes 110 in the RAN 100 can be the same type of nodes or different types of nodes. In some scenarios, the roles of the RAN nodes 110 and the terminals 120 are relative, e.g., the network element 120i in Figure 2 can be a helicopter or a drone, which can be configured to be a mobile base station, to the terminals 120j accessing to the RAN 100 through the network element 120i, the network element 120i is a base station; but to the base station 110a, the network element 120i is a terminal. The RAN nodes 110 and the terminals 120 are sometimes referred to as communication apparatuses, e.g., the network elements 110a and 110b in Figure 2 can be understood as communication apparatuses with base station functionalities, and the network elements 120a-120j can be understood as communication apparatuses with terminal functionalities.

[0112] For the RAN node 110, in one possible scenario, the RAN node 110 can be a base station, an evolved Node B (eNodeB, also referred to as eNB), an access point (AP), a transmission reception point (TRP), a next generation NodeB (gNB), a next generation NodeB in a future mobile communications system, or an access node in a WiFi system, etc. The RAN node 110 can be a macro base station (e.g., 110a in Figure 2), a micro base station or indoor station (e.g., 110b in Figure 2), a relay node or donor node, or a wireless controller in a CRAN scenario. For example, a satellite base station, a radio network controller (RNC), a base station controller (BSC), a base transceiver station (BTS), a home base station (e.g., home eNodeB, or home NodeB, HNB), a relay station, a balloon station, a drone station, a wireless backhaul node, or a G node in a starlink, etc. It can be understood that the network device can be a device arranged on the ground, or a non-ground device (such as a satellite, a drone, a high-altitude communication device, etc.). In addition, in a communication system using different wireless access technologies, the name of the network device with base station function may be different, which is not limited in the present application. Optionally, the RAN node 110 can also be a server, a wearable device, a vehicle or a vehicle-mounted device, etc. For example, the access network device in the vehicle to everything (V2X) technology can be a road side unit (RSU). The RAN node 110 is also referred to as a next generation-RAN (NG-RAN) node.

[0113] In another possible scenario, a terminal is assisted by multiple RAN nodes 110 to implement wireless access in cooperation, and different RAN nodes 110 respectively implement part of the functions of a base station. For example, a RAN node 110 can be a central unit (CU, also known as a central unit), a distributed unit (DU, also known as a distributed unit), a CU-control plane (CP), a CU-user plane (UP), or a radio unit (RU), etc. The CU and the DU can be separately arranged, or can also be included in the same network element, such as a baseband unit (BBU). The RU can be included in a radio frequency device or a radio frequency unit, such as a remote radio unit (RRU), an active antenna processing unit (AAU), or a remote radio head (RRH).

[0114] In different systems, the CU (or CU-CP and CU-UP), DU or RU can also have different names, but those skilled in the art can understand their meanings. For example, in an ORAN system, the CU can also be referred to as an O-CU (open CU), the DU can also be referred to as an O-DU, the CU-CP can also be referred to as an O-CU-CP, the CU-UP can also be referred to as an O-CU-UP, and the RU can also be referred to as an O-RU. For the convenience of description, the CU, CU-CP, CU-UP, DU and RU are taken as examples for description in this application. Any one of the CU (or CU-CP, CU-UP), DU and RU in this application can be implemented by a software module, a hardware module, or a combination of a software module and a hardware module.

[0115] For the terminal 120, in a possible scenario, the terminal 120 can be a device for implementing a wireless communication function, for example, a terminal or a chip or circuit used in a terminal, or an entity associated with the terminal, etc. Among them, the terminal 120 can be a user equipment (UE), an access terminal, a terminal unit, a terminal station, a mobile station (MS), a mobile station, a remote station, a remote terminal, a mobile device, a wireless communication device, a terminal agent or a terminal device, a subscriber unit, a smart phone, a wireless data card, a tablet computer, a wireless modem, a laptop computer, a machine type communication (MTC) terminal, a tag, etc. in a 5G network or a future evolved public land mobile network (PLMN). The access terminal can be a cellular phone, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (WLL) station, a personal digital assistant (PDA), a handset with wireless communication function, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device or a wearable device, a virtual reality (VR) terminal, an augmented reality (AR) terminal, a wireless terminal in industrial control, a wireless terminal in self driving, a wireless terminal in remote medical, a wireless terminal in smart grid, a wireless terminal in transportation safety, a wireless terminal in smart city, a wireless terminal in smart home, or a terminal node (T node) in starlink, etc. In a possible implementation, the terminal 120 can be mobile or fixed. It can be understood that the terminal and the mobile user can be completely independent. All information related to the user can be stored in a subscriber identity module (SIM) card, which can be used on a terminal device.The terminal can send and / or receive signals through the air interface to complete interaction with the network side device.

[0116] The chip or circuit in the terminal includes at least one of components inside the terminal, such as a chip, a central processing unit (CPU), a network processing unit (NPU), and a terminal radio frequency module.

[0117] The entity associated with the terminal includes a server on the terminal side, a computing / processing node, a computing / processing entity, a computing / processing unit, a server, such as an over the top (OTT) server, etc. OTT refers to services provided by a third party other than a network operator to users based on an operator network, such as OTT voice communication services, OTT multimedia services, and OTT data processing services, etc. The terminal interacts with relevant information (such as data) through communication with the associated network entity. For example, the associated network entity and the terminal belong to the same manufacturer. Due to model training, model selection, etc., it can not be performed on the terminal, but on the OTT server on the terminal side, so the "terminal" in this embodiment also includes the OTT server on the terminal side.

[0118] It should be understood that the terminal in this embodiment can also be referred to as "terminal side" (UE side) or "terminal part" (UE part).

[0119] For example, as shown in FIG. 3, an exemplary implementation of the system shown in FIG. 2 is provided. The communication system can include an AI / ML node, a first communication device, and a second communication device. The second communication device can provide services for the first communication device.

[0120] Optionally, the second communication device can be any device deployed in an access network that can communicate wirelessly with the first communication device (e.g., a terminal), and can also be a chip or chip system that can be provided in the above-mentioned device, and can also be a logical node or a logical module or a function implemented in software, mainly responsible for wireless physical control functions, resource scheduling, wireless resource management, quality of service management, data compression and encryption, wireless access control, and mobility management, etc. on the air interface. Specifically, the second communication device can be a device supporting wired access, or a device supporting wireless access.

[0121] Optionally, the communication system shown in FIG. 3 can include a network device. For example, the network device can be a server, which can be a single server or a server cluster composed of multiple servers. In some embodiments, the server cluster can also be a distributed cluster. The server can provide services for a chip, and thus can also be referred to as a chip server. Alternatively, the network device can be a first network element in a core network. The network device can deliver a model, model information, or statistical information of model information to the first communication apparatus or the second communication apparatus.

[0122] The AI / ML node in FIG. 3 is used to support the use of AI / ML technology in an AI / ML scenario.

[0123] Optionally, the AI / ML node can be deployed in one or more of the following positions in the communication system shown in FIG. 3: the network device, the first communication apparatus, or the second communication apparatus, or the AI / ML node can also be deployed separately, for example, in a position other than any of the above-mentioned devices.

[0124] For example, the AI / ML node can be deployed in a host or a cloud server of an OTT system. When the device in which the AI / ML node is deployed communicates with the second communication apparatus, the device can also act as a terminal in the communication system. When the device in which the AI / ML node is deployed communicates with the terminal, the device can also act as a network device in the communication system.

[0125] It can be understood that the number of AI / ML nodes is not limited in the present application. For example, when there are multiple AI / ML nodes, the multiple AI / ML nodes can be divided based on functions, for example, different AI / ML nodes are responsible for different functions.

[0126] It can also be understood that the AI / ML node can be a separate device, can be integrated into the same device to implement different functions, or can be a network element in a hardware device, or can be a software function running on a dedicated hardware, or a virtualized function instantiated on a platform (e.g., a cloud platform), and the specific form of the AI / ML node is not limited in the present application.

[0127] The AI / ML node can be an AI / ML network element or an AI / ML module.

[0128] It can be understood that the above-mentioned FIG. 3 is only a schematic diagram and does not constitute a limitation on the applicable scenarios of the technical solutions provided in the present application. It should be understood by those skilled in the art that in the specific implementation process, the communication system shown in FIG. 3 can also include fewer devices than those shown in FIG. 3, or the communication system shown in FIG. 3 can also include other devices, and the number of devices in the communication system shown in FIG. 3 can also be determined according to specific needs and is not limited.

[0129] Optionally, each device in FIG. 3, such as the first communication apparatus, the second communication apparatus, and the network device, can be a general device or a special device, and embodiments of the present application do not make a specific limitation.

[0130] Optionally, the related functions of each device in FIG. 3 can be implemented by one device, or implemented by multiple devices together, or implemented by one or more functional modules in one device, and embodiments of the present application do not make a specific limitation. It can be understood that the above functions can be network elements in a hardware device, or software functions running on a special hardware, or a combination of hardware and software, or virtualized functions instantiated on a platform (for example, a cloud platform).

[0131] In a possible implementation, the network device (for example, an access node or a core network node) in embodiments of the present application and the terminal 120, which can also be referred to as a communication apparatus, can be a general device or a special device, and the network device can include an access node (RAN node), an operation administration and maintenance (OAM) device, or a core network node. For the OAM device, it can include a device in an element management system (EMS), or a device in a network management system (NMS). It should be understood that the network device in the embodiments can also be referred to as a "network side" or a "network part". Embodiments of the present application do not make a specific limitation.

[0132] In a possible implementation, the related functions of the terminal 120 or the network device in embodiments of the present application can be implemented by one device, or implemented by multiple devices together, or implemented by one or more functional modules in one device, and embodiments of the present application do not make a specific limitation. It can be understood that the above functions can be network elements in a hardware device, or software functions running on a special hardware, or a combination of hardware and software, or virtualized functions instantiated on a platform (for example, a cloud platform).

[0133] It should be noted that the RAN node can be a device or a component in the device in the above NG-RAN, for example, it can be a ng-eNB node, a gNB node, or a transmission point (TP) in the ng-eNB node and the gNB node, a transmission and reception point (TRP), or a central unit (CU) integrated on the NG-RAN. The RAN node can also be a network element with transmission function, such as a transmission measurement function (TMF) network element. In some embodiments, the RAN node can also be an access node in the O-RAN system. The RAN is usually composed of a series of modules, such as antenna, RRU, and BBU modules. The traditional RAN architecture defines the overall reception and output of the RAN node, and does not limit the transmission and contact between internal modules. The O-RAN architecture defines the architecture contact and standardized interface between each module in the RAN, so that the RAN can be decoupled into multiple standard modules, thereby realizing the combination and replacement of modules.

[0134] For example, as shown in FIG. 4, it is a possible, non-limiting structure diagram of an O-RAN system. Among them, the service management and orchestration framework (SMO) is used as the network management device in the O-RAN, which is used to manage the devices in the O-RAN. The non-real time RAN intelligent controller (Non-RT RIC) is located in the SMO module, which is used to realize the non-real time intelligent management of the RAN function, for example, it can realize the AI / ML workflow including model training and model updating, and guide the application / function in the Near-RT RIC based on the policy. The near-real time RAN intelligent controller (Near-RT RIC) is used to realize the near-real time intelligent management of the RAN. Through data collection and related operations on the E2 interface, the near-real time control and optimization of the modules and resources of the O-RAN are realized.

[0135] An O-RAN central unit (O-CU) includes an O-RAN central unit control plane (O-CU-CP) and an O-RAN central unit user plane (O-CU-UP). The O-CU is configured to implement a radio resource control (RRC) layer, a packet data convergence protocol (PDCP) layer, and a service data adaptation protocol (SDAP) layer and other control functions. The O-CU-CP is configured to implement functions of the RRC layer and control plane functions of the PDCP layer. The O-CU-UP is configured to implement functions of the SDAP layer and user plane functions of the PDCP layer.

[0136] An O-RAN distributed unit (O-DU) is configured to implement a radio link control (RLC) layer, a media access control (MAC) layer, and a higher physical layer (Higher PHY). The Higher PHY functions include one or more of forward error correction (FEC) encoding / decoding, scrambling / descrambling, or modulation / demodulation.

[0137] An O-RAN radio unit (O-RU) is configured to implement lower physical layer (Lower PHY) functions and radio frequency functions. The Lower PHY functions include one or more of fast Fourier transform (FFT) transform / inverse fast Fourier transformation (iFFT) transform, digital beamforming, or extraction and filtering of a physical random access channel (PRACH). That is, the O-RU has functions of a radio frequency device such as a TRP or a RRH and Lower PHY processing functions. In addition, the O-RU, the O-CU, and the O-DU can be collectively configured as an O-eNB / gNB to implement the above functions.

[0138] As a cloud computing platform, the O-RAN cloud (O-Cloud) includes physical infrastructure nodes for hosting O-RAN functions such as RIC, O-DU, etc. The O-Cloud supports software components (such as operating systems, virtual machine monitors, container runtimes), management and orchestration functions.

[0139] In a possible scenario, a sensing unit (SU) is further included in the O-RAN system. The SU is mainly used to implement sensing-related functions, such as transmitting a sensing signal and / or receiving an echo signal of the sensing signal, performing corresponding signal processing on the received echo signal to obtain sensing measurement data, and performing sensing-related processing.

[0140] As a possible implementation, the RAN node can include at least one of a CU, a DU, a SU, and a RU. There is a communication interface between the CU and the SU. There can or can not be a communication interface between the SU and the DU. In the case where there is no communication interface between the SU and the DU, the SU and the DU can communicate through the CU.

[0141] Under the O-RAN architecture, the module that receives the difference reporting of the twin channel and the measurement channel can be a CU, a RT RIC, a Non-RT RIC, etc. The DU is responsible for receiving signals, signal processing, multipath measurement, and channel difference calculation.

[0142] For example, the O-RAN system includes communication interfaces between newly added internal components and other communication interfaces. For example, the A1 interface is an interface between the Non-RT RIC and the Near-RT RIC, which is used for intelligent and dynamic control of O-RAN internal wireless resources. The Non-RT RIC can provide policies, rich information, and ML model updates to the Near-RT RIC through the A1 interface, and the Near-RT RIC can provide policy feedback to the Non-RT RIC through the A1 interface.

[0143] The E2 interface is an open interface between two endpoints, used to connect the Near-RT RIC and the RAN node, which includes the CU, the DU in 5G, the O-RAN compatible eNB in 4G, the O-CU (O-CU-CP and / or O-CU-UP) and / or the O-DU in O-RAN, etc. The Near-RT RIC can obtain RAN node data collection and feedback through the E2 node, and the RAN node can obtain control feedback of the Near-RT RIC through the E2 node.

[0144] The O1 interface is an interface between a management entity in the SMO and an O-RAN module, used for operation management, through which network management (for example, fault management, configuration management, billing management, performance management, security management, also referred to as FCAPS management), software management, and file management are implemented. The O2 interface is an interface between the SMO and an infrastructure management framework supporting O-RAN virtual network functions.

[0145] The open front-haul (FH) CUS-Plane interface includes a control plane C-Plane, a user plane U-Plane, and a synchronization plane S-Plane interface. The control plane is used for real-time control between the O-DU and the O-RU, for example, for the O-DU to transmit the weight value for beamforming to the O-RU, or for the O-DU to perform power control on the O-RU, etc. The user plane is used to transmit communication data between the access network device and the terminal between the DU and the RU. The synchronization plane is used for the O-DU to provide clock synchronization to the O-RU. The Open FH M-Plane interface is a management plane interface, used for connection between the O-RU and the O-DU and the SMO, and can implement management, monitoring, and configuration functions, etc.

[0146] In addition, the NG interface is an interface between a RAN node (for example, a base station, a CU, a CU-CP, and a CU-UP) and a core network, NG-u is a user plane NG interface, and NG-c is a control plane NG interface. The Xn interface is an interface between NR RAN nodes, Xn-u is a user plane Xn interface, and Xn-c is a control plane Xn interface. The X2 interface is an interface between LTE RAN nodes, X2-u is a user plane X2 interface, and X2-c is a control plane X2 interface. In the NR system, the X2 interface is mainly used in the E-UTRA-NR dual connectivity (EN-DC) scenario, in which the master base station is an LTE RAN node, and the master base station is connected to the LTE core network through the X2 interface. The E1 interface is an interface between the CU-CP and the CU-UP, the F1-c interface is an interface between the CU-CP and the DU, and the F1-u interface is an interface between the CU-UP and the DU.

[0147] It should be noted that the communication system described in the embodiments of the present application is used to more clearly illustrate the technical solutions of the embodiments of the present application, and does not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art can know that, with the evolution of network architecture and the appearance of new business scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0148] The communication method provided by the embodiments of the present application is described below by taking the first communication device and the second communication device as an example, and by taking the communication system shown in FIG. 2. It should be noted that the names of messages, the names of parameters, or the names of information between the first communication device and the second communication device in the embodiments described below are only examples, and other names can also be used in other embodiments, and the method provided by the present application does not make a specific limitation on this.

[0149] It can be understood that in the embodiments of the present application, the first communication device or the second communication device can perform part or all of the steps in the embodiments of the present application, and these steps or operations are only examples, and the embodiments of the present application can also perform other operations or variations of various operations. In addition, each step can be performed in a different order as presented in the embodiments of the present application, and it is possible that not all operations in the embodiments of the present application are performed.

[0150] It can be understood that the first communication device and the second communication device are taken as an example of the execution subject of the interaction in the present application, but the present application does not limit the execution subject of the interaction. For example, the method performed by the first communication device in the present application can also be performed by a module (such as a chip, a chip system, or a processor) applied to the first communication device, and can also be implemented by a logical node, a logical module, or software that can implement all or part of the functions of the first communication device; the method performed by the second communication device in the present application can also be performed by a module (such as a chip, a chip system, or a processor) applied to the second communication device, and can also be implemented by a logical node, a logical module, or software that can implement all or part of the functions of the second communication device.

[0151] In addition, "sending information" in the present application can be understood as a device sending information to another device, or can also be understood as a logical module in a device sending information to another logical module. For example, "the second communication device sending information" can be understood as the second communication device sending information to another device (such as the first communication device), or can be understood as a logical module 1 (such as a processing module) in the second communication device sending information to a logical module 2 (such as a transceiver module) in the second communication device.

[0152] "Receiving information" in the present application can be understood as a device receiving information from another device, or can also be understood as a logical module in a device receiving information from another logical module. For example, "the first communication device receiving information" can be understood as the first communication device receiving information from another device (such as the second communication device), or can be understood as a logical module 1 (such as a processing module) in the first communication device receiving information from a logical module 2 (such as a transceiver module) in the first communication device.

[0153] In the present application, "sending information to (e.g. the first communication device)" or the related illustration in the drawings can be understood as that the destination of the information is the first communication device. It can include directly or indirectly sending information to the first communication device. "Receiving information from (e.g. the second communication device)" or "receiving information from (e.g. the second communication device)" or "receiving information sent by (e.g. the second communication device)", or the related illustration in the drawings can be understood as that the source of the information is the second communication device, which can include directly or indirectly receiving information from the second communication device. The information between the source and the destination of the information sending can be processed as necessary, such as format change, etc., but the destination can understand the effective information from the source. Similar expressions in the present application can be similarly understood, which will not be repeated here.

[0154] Referring to FIG. 5, a flowchart of a communication method provided by an embodiment of the present application is shown, which can include the following steps:

[0155] S501, the second communication device determines the first information.

[0156] The first information includes model information of at least one first model and / or statistical information corresponding to the model information of the at least one first model.

[0157] For example, the first information including the model information of the at least one first model can be understood as that the first information includes the model information of one or more first models in the at least one first model. The model information can include the generalization range of the first model, the model complexity of the first model, the state of the first model, the performance index of the first model, the size of the data set used for training / altering the first model, the model alteration record of the first model, the collection time of the data set used for altering the first model, etc. Similarly, the first information including the statistical information corresponding to the model information of the at least one first model can be understood as that the first information includes the statistical result of the model information of each first model in the at least one first model, for example, the proportion of the first model that has model alteration in the at least one first model, the at least one generalization range with the largest proportion in the multiple generalization ranges corresponding to the at least one first model, or the one or more model complexities with the largest proportion in the multiple model complexities corresponding to the at least one first model, etc.

[0158] In addition, the proportion of the first model in which the model change occurs in the at least one first model can be understood as a ratio between the number of the first model in which the model change occurs and the total number of models of the at least one first model, or can also be understood as a ratio between the number of the communication devices in which the maintained first model changes and the total number of the communication devices maintaining the at least one first model. Similarly, the meaning of the proportion corresponding to a certain model information or a certain value of the model information can be referred to the above understanding, and will not be described herein.

[0159] That is, the second communication device determines the model information of the at least one first model and / or the statistical information corresponding to the model information of the at least one first model as the first information. For example, the second communication device determines the model information of the at least one first model as the first information, or the second communication device determines the statistical information corresponding to the model information of the at least one first model as the first information, or the second communication device determines the model information of the at least one first model and the statistical information corresponding to the model information of the at least one first model as the first information, and the embodiments of the present application are not limited thereto.

[0160] As a possible implementation, the first information is used to determine a management policy of a second model, and the first model and the second model have an association relationship.

[0161] For example, the first information used to determine the management policy of the second model can be understood as auxiliary information or indication information for determining the management policy of the second model, or can also be understood as having a guiding significance in the process of determining the management policy of the second model.

[0162] The association relationship between the first model and the second model can be understood as that the first model and the second model have the same or similar application scenarios, or can also be understood as that the first model and the second model are used to implement the same or similar communication functions.

[0163] For example, at least one network configuration (denoted as configuration set 1) applicable to the second model is a subset of at least one network configuration (denoted as configuration set 2) applicable to the first model, or the network configuration applicable to the second model is the same as the network configuration applicable to the first model, that is, the network configuration applicable to the first model contains the network configuration applicable to the second model, and the configuration set 2 contains all elements in the configuration set 1.

[0164] As a possible implementation, the at least one first model is a model maintained by at least one third communication device.

[0165] For example, the at least one first model maintained by the at least one third communication device can be understood as follows: each of the at least one first model is maintained by one third communication device, and part or all of the at least one first model can be maintained by the same third communication device, or each of the at least one first model is maintained by different third communication devices.

[0166] For example, the at least one first model includes model 1 to model 10, wherein model 1 and model 2 are maintained by the third communication device 1, model 3 to model 5 are maintained by the third communication device 2, model 6 is maintained by the third communication device 3, and model 7 to model 10 are maintained by the third communication device 4.

[0167] For another example, the at least one first model includes model 1 to model 10, and model 1 to model 10 are all maintained by the third communication device 1.

[0168] S502, the second communication device sends first information to the first communication device. Correspondingly, the first communication device receives the first information from the second communication device.

[0169] For example, the first information can be carried in a predefined message. The predefined message can be understood as a message predefined by a protocol, or a message predefined by the first communication device and the second communication device.

[0170] For example, taking the first communication device as a terminal and the second communication device as a base station as an example, the first information can be carried in a system information block (MIB) or downlink control information (DCI).

[0171] S503, the first communication device determines a management strategy of a second model according to the first information. The second model is a model maintained by the first communication device.

[0172] For example, the second model maintained by the first communication device can be understood as a model generated by the first communication device in advance, or can also be understood as a model to be activated or activated by the first communication device, or can also be understood as a model stored by the first communication device.

[0173] As a possible implementation, the management policy includes one of the following: model retraining, model updating, model activation, model switching, or model deactivation. The meanings of model retraining, model updating, model switching, and model activation can be referred to the relevant descriptions in the foregoing embodiments, and model deactivation can be understood as changing the state of the current model in the enabled state to the disabled state, or can also be understood as stopping using the current model to implement a specific communication function.

[0174] In addition, the management policy of the second model can also include model selection, data collection, and data analysis, and the meanings of each management policy can be referred to the relevant descriptions in the foregoing embodiments. In the case where the second model is a model to be generated by the first communication device, the model management policy of the second model can also include model training, that is, the first communication device can also receive the first information from the second communication device and determine whether to train the second model or determine the target parameters of the trained second model, such as the generalization range, model complexity, or target performance indicators, in the process of preparing to train the second model. The embodiments of the present application do not limit the state (to be trained or trained) of the second model and the management policy of the model.

[0175] For example, the first information of the second communication device includes statistical information corresponding to the model information of at least one first model, and the second communication device triggers the second model maintenance process when the performance indicators of the second model are lower than the first threshold. After obtaining the first information, if it is detected that the proportion of the first model that has changed in the at least one first model is higher than the second threshold, the model updating or model retraining can be used as the management policy of the second model; or if it is detected that the proportion of the first model whose performance indicators are lower than the first threshold in the at least one first model is higher than the third threshold, the model deactivation can be used as the management policy of the second model. Alternatively, the second model is one of a plurality of alternative models for implementing a specific communication function, and if the proportion of the first model whose performance indicators are higher than the first threshold in the at least one first model is higher than the fourth threshold, the model activation can be used as the management policy of the second model, that is, the second model is enabled as the model for implementing the specific communication function.

[0176] In addition, the above is described by taking the statistical information corresponding to the model information of the at least one first model as an example. In the case where the first information includes the model information of the at least one first model, the first communication device can perform statistical analysis on the model information of each first model after receiving the first information, obtain the statistical information corresponding to the model information of the at least one first model, and then determine the management strategy of the second model according to the statistical information corresponding to the model information of the at least one first model obtained through statistical analysis after triggering the maintenance process of the second model. The manner of determining the management strategy of the second model can refer to the related description of the foregoing embodiments, and will not be described in detail.

[0177] In the case where the first information includes the model information of the at least one first model and the statistical information corresponding to the model information of the at least one first model, the first communication device can determine the management strategy of the second model in combination with the model information and the statistical information, or can verify the statistical information by using the model information, and then determine the management strategy of the second model according to the statistical information after the statistical information passes the verification, which is not limited.

[0178] Based on the scheme, in the process of determining the management strategy of the second model to be maintained, the first communication device receives the first information from the second communication device, the first information includes the model information of the at least one first model having an association relationship with the second model and / or the statistical information corresponding to the model information of the at least one first model, and then the first communication device determines the management strategy used in the process of maintaining the second model according to the first information. That is, in the process of determining the management strategy of the second model, the first communication device takes the model information of the first model having an association relationship with the second model or the statistical information corresponding to the model information of the first model as prior information, and determines the management strategy of the model according to the prior information. In the process of determining the management strategy of the model, a large amount of training data does not need to be collected and the data statistical characteristics of the training data do not need to be analyzed, which is beneficial to reduce the resource overhead required by the first communication device in determining the management strategy of the second model in the process of maintaining the second model, and improve the efficiency of the first communication device in maintaining the second model.

[0179] The overall process of the communication method provided by the present application is described above, and the specific implementation of each step is introduced below.

[0180] In a possible implementation, the model information of the first model indicates at least one of the following: performance index information of the first model, a state of the first model, whether the first model is changed, a time of starting to collect the first data set, a collection completion time of the first data set, a size of the first data set, generalization range information of the first model, or model complexity information of the first model.

[0181] The first model whether to change can be understood as whether the first model has undergone model retraining or model update.

[0182] As a possible implementation, the first model whether to change includes whether the first model has changed after the first time. That is, the first model whether to change in the model information of the first model can be used to indicate whether the first model has undergone model retraining or model update after the first time.

[0183] The first time can be a time point that is predefined by the protocol and has a preset interval with the current time, or the first time can be determined by the first communication device and sent to the second communication device, or the first time can be determined by the second communication device in advance and sent to the first communication device.

[0184] For example, the first time can be determined according to the time point when the second model has model performance degradation, for example, the first time is the time point when the second model has performance degradation, or the first time is later than the time point when the second model has performance degradation, and the time interval between the first time and the time point when the second model has performance degradation is less than or equal to a given value, or the first time is earlier than the time point when the second model has performance degradation, and the time interval between the first time and the time point when the second model has performance degradation is less than or equal to a given value.

[0185] Based on this scheme, the first information can accurately contain the information of whether the first model has changed after the first time, so that the first communication device can accurately obtain the model change trend of the at least one first model after the reference time (the first time). In the case of setting the first time as a time point that is appropriately spaced from the current time, it is beneficial for the first communication device to determine the management strategy of the second model according to the recent model change trend of the at least one first model, and to reduce the interference of the history change of the first model on the accuracy of determining the management strategy of the second model according to the first information.

[0186] The first data set is used for training or changing the first model.

[0187] For example, the first data set used for training the first model can be understood as the first data set used in the process of training the first model, or it can also be understood as a subset of the first data set used in the process of training the first model. The first data set used for changing the first model can be understood as the first data set used in the process of updating or retraining the first model, or it can also be understood as a subset of the first data set used in the process of updating or retraining the first model.

[0188] Therefore, the time when the first data set starts to be collected can be understood as the time when the data set for training the first model starts to be collected in the process of training the first model, or can also be understood as the time when the data set for changing the first model starts to be collected in the process of changing the first model; similarly, the collection completion time of the first data set can be understood as the time when the data collection of the data set for training the first model is completed in the process of training the first model, or can also be understood as the time when the data collection of the data set for changing the first model is completed in the process of changing the first model.

[0189] As a possible implementation, in the case where the first model is changed, the first data set is the data set used in the process of changing the first model, and the size of the first data set contained in the model information of the first model indicates the size of the data set used in changing the first model. In the case where the first model is not changed, the first data set is the data set used in the process of training the first model, and the size of the first data set contained in the model information of the first model indicates the size of the data set used in training the first model.

[0190] Optionally, the size of the first data set can be understood as the size of the total storage space occupied by all the data contained in the first data set, or can also be understood as the total number of data contained in the first data set.

[0191] In addition, in the case where whether the first model is changed indicates whether the first model is changed after the first time, in the case where the first model is changed, the first data set is the data set used in the process of changing the first model, and the size of the first data set contained in the model information of the first model indicates the size of the data set used in changing the first model; in the case where the first model is not changed, the first data set is the data set used in the process of training the first model, or the first data set is the data set used in the last change before the first time, and the size of the first data set contained in the model information of the first model indicates the size of the data set used in training the first model or the size of the data set used in the last change before the first time.

[0192] Based on this scheme, the first communication device can obtain the size of the data set used in the training or the last change of the first model, and the time of training or the last change of the first model, which is beneficial to the first communication device to determine whether to train or change the second model, and the target size of the data set used in the process of training or changing the second model.

[0193] The performance index information of the first model is used to indicate the performance index in the running process of the first model, or can also be used to indicate the system performance in the process of the communication device calling the first model.

[0194] Exemplarily, the performance in the running of the first model includes inference performance, prediction performance, or monitoring performance. The performance indicator of the inference performance or the prediction performance can be represented by prediction accuracy, which reflects the similarity between the inference result of the first model and the actual measurement result. For example, for the function of CSI compression feedback, the performance indicator of the first model can be the restoration accuracy of restoring (decompressing) the compressed CSI; for the function of CSI prediction, the performance indicator of the first model can be the accuracy of CSI prediction; for the function of beam prediction, the performance indicator of the first model can be the accuracy of beam prediction; for the function of positioning inference, the performance indicator of the first model can be the accuracy of positioning inference.

[0195] Exemplarily, the system performance in the running of the first model by the communication device includes: throughput, reference signal receiving power (RSRP), signal noise ratio (SNR) / signal to interference plus noise ratio (SINR), or block error rate (BLER), and other parameters for representing system performance.

[0196] As a possible implementation, the performance indicator information of the first model indicates the performance indicator of the first model. Exemplarily, the model information of the first model can include the performance indicator of the first model at the current time, or the model information of the first model can also include the performance indicator of the first model before the first time and the performance indicator of the first model after the first time. The current time can be the time when the first information is determined, or it can also be any time in the process of determining the first information.

[0197] As another possible implementation, the performance indicator information of the first model indicates the performance indicator of the first model, and the time information when the first model meets the first performance indicator and / or the time information when the first model does not meet the first performance indicator.

[0198] Exemplarily, the first model meeting the first performance indicator can be understood as the performance indicator of the first model being greater than or equal to the first performance indicator. The first performance indicator can be predefined by a protocol, or can be agreed upon by the first communication device and the second communication device in advance.

[0199] That is, the performance indicator information of the first model can not only include the current performance indicator of the first model, but also include the time information when the performance indicator of the first model is greater than or equal to the first performance indicator and / or the time information when the performance indicator of the first model is less than the first performance indicator in the running of the first model.

[0200] For example, the time information that the first model meets the first performance indicator indicates at least one of a start time, an end time or a duration that the first model meets the first performance indicator; and the time information that the first model does not meet the first performance indicator indicates at least one of a start time, an end time or a duration that the first model does not meet the first performance indicator.

[0201] For example, the performance indicator information of the first model can include a current performance indicator of the first model, a start time and an end time that the first model meets the first performance indicator, or the performance indicator information of the first model can include a current performance indicator of the first model, a performance indicator of the first model before a first time and a duration that the first model meets the first performance indicator.

[0202] It is worth mentioning that the start time and the end time that the first model meets the first performance indicator can include multiple time points, for example, the first model starts to meet the first performance indicator at t1, starts to not meet the first performance indicator at t2, re-meets the first performance indicator at t3, and starts to not meet the first performance indicator at t4, then the start time that the first model meets the first performance indicator includes t1 and t3, and the end time includes t2 and t4. Similarly, the start time that the first model does not meet the first performance indicator includes t2 and t4, and the end time that the first model does not meet the first performance indicator includes t1 and t3.

[0203] In addition, the performance indicator information of the first model can also directly indicate a time proportion that the first model meets the first performance indicator, that is, a ratio between a total duration that the first model meets the first performance indicator in a running process and a running duration.

[0204] Optionally, in a case where the model information of the first model includes performance indicator information of the first model, and the at least one first model is a model maintained by the at least one third communication device, the relationship between the at least one third communication device and the first communication device includes the following three possible implementation manners:

[0205] Manner 1, the third communication device and the first communication device are communication devices of the same model.

[0206] Here, the model of the communication device can be understood as the same manufacturer (such as a terminal manufacturer or a chip manufacturer) corresponding to the communication device, or can also be understood as the same chip model included in the communication device.

[0207] That is, in the at least one third communication device maintaining the at least one first model, each third communication device is of the same model as the first communication device, i.e. the proportion of third communication devices in the at least one third communication device that are of the same model as the first communication device is 100%, and the proportion of third communication devices in the at least one third communication device that are of a different model from the first communication device is 0.

[0208] Optionally, the third communication device is of a different model from the first communication device.

[0209] That is, in the at least one third communication device maintaining the at least one first model, there are no third communication devices of the same model as the first communication device, and each third communication device is of a different model from the first communication device. The proportion of third communication devices in the at least one third communication device that are of a different model from the first communication device is 100%, and the proportion of third communication devices in the at least one third communication device that are of the same model as the first communication device is 0.

[0210] Optionally, the third communication device is of a different model from the first communication device.

[0211] That is, in the at least one third communication device maintaining the at least one first model, there are third communication devices of the same model as the first communication device and third communication devices of a different model from the first communication device.

[0212] Optionally, the first information can or can not indicate the proportion of third communication devices in the at least one third communication device that are of the same model as the first communication device and / or the proportion of third communication devices in the at least one third communication device that are of a different model from the first communication device.

[0213] As one possible implementation, the proportion of third communication devices in the at least one third communication device that are of the same model as the first communication device is greater than the proportion of third communication devices in the at least one third communication device that are of a different model from the first communication device.

[0214] For example, the proportion of third communication devices in the at least one third communication device that are of the same model as the first communication device is 10%, and the proportion of third communication devices in the at least one third communication device that are of a different model from the first communication device is 90%; or the proportion of third communication devices in the at least one third communication device that are of the same model as the first communication device is 25%, and the proportion of third communication devices in the at least one third communication device that are of a different model from the first communication device is 75%.

[0215] As another possible implementation, the proportion of the third communication devices of the same model as the first communication device in the at least one third communication device is less than or equal to the proportion of the third communication devices of different models as the first communication device in the at least one third communication device.

[0216] For example, the proportion of the third communication devices of the same model as the first communication device in the at least one third communication device is 10%, and the proportion of the third communication devices of different models as the first communication device in the at least one third communication device is 90%; or the proportion of the third communication devices of the same model as the first communication device in the at least one third communication device is 30%, and the proportion of the third communication devices of different models as the first communication device in the at least one third communication device is 70%.

[0217] Based on the scheme, the first communication device can accurately obtain the performance indicators of the first models, and by setting the models of the third communication devices participating in the statistics, the guiding significance of the performance indicators of the first models to whether to train the second model and the second model training strategy is improved, and the accuracy of determining whether to train the second model and the second model target performance indicator according to the first information is improved.

[0218] The state of the first model can be understood as the state of the first model in the communication device maintaining the model, or can also be understood as the ability state of the communication device maintaining the first model to run the first model.

[0219] As a possible implementation, the state of the first model can include at least one of the following: a supported state, an available state, a suitable state, an activated state, or a configured state. The state of the first model is described below with the second communication device as a terminal as an example.

[0220] The supported state or simply supported is used to indicate whether the terminal maintaining the first model has the ability to execute the first model. In the case that the terminal has the ability to execute the first model, the state of the first model is the supported state. In the case that the terminal does not have the ability to execute the first model, the state of the first model is the not supported state.

[0221] The available state or simply available is used to indicate whether the terminal maintaining the first model has the ability to obtain / load / store the first model. In the case that the terminal has the ability to obtain / load / store the first model, the state of the first model is the available state. In the case that the terminal does not have the ability to obtain / load / store the first model, the state of the first model is the not available state.

[0222] Applicable state or applicable for short is used to indicate whether the terminal maintaining the first model is ready to use the first model to perform a specific communication function or reasoning. In the case that the terminal is ready to use the first model for reasoning, the state of the first model is the applicable state, and in the case that the terminal is not ready to use the first model for reasoning, the state of the first model is the inapplicable state.

[0223] Optionally, the terminal has the condition of executing or running the first model, and / or the terminal has acquired / stored / loaded the first model, and thus can be regarded as the terminal being ready to use the first model for reasoning.

[0224] Configured state is used to indicate whether the terminal maintaining the first model has the configuration of the model management device for executing / running the first model. In the case that the terminal has the configuration of the model management device for executing / running the first model, the state of the first model is the configured state, and in the case that the terminal does not have the configuration of the model management device for executing / running the first model, the state of the first model is the unconfigured state.

[0225] For example, the configuration of the model management device can be configured by the model management device (for example, a base station) to the terminal, for subsequent enabling / activating / instructing the terminal to use / run / execute the first model, that is, the configuration is the configuration related to the use of the first model, such as the configuration related to the input and / or output of the first model. Taking the input of the first model as a reference signal for example, the configuration can be the resource of the reference signal.

[0226] Activated state or activated for short is used to indicate whether the terminal maintaining the first model has activated and executed the first model for reasoning. In the case that the terminal has activated and executed the first model for reasoning, the state of the first model is the activated state, and in the case that the terminal has not activated or has not executed the first model for reasoning, the state of the first model is the deactivated state or the inactivated state.

[0227] It is worth mentioning that the state of the first model can include the current state of the first model, and can also include the state of the first model before the first time and the state of the first model after the first time, without limitation.

[0228] Based on the scheme, the first communication device can accurately acquire whether the first model is supported or in the activated state in the communication device maintaining the first model, which is beneficial for the first communication device to determine that the target management strategy of the second model associated with the first model is model deactivation or model change according to the state of at least one first model.

[0229] The generalization range information of the first model can indicate a generalization range of the first model.

[0230] For example, the generalization range can be understood as at least one application scenario to which the first model is applicable, or can also be understood as at least one application scenario in which the adaptation degree with the first model is higher than a preset value. The preset value can be predefined by a protocol or indicated by the second communication device in advance.

[0231] Optionally, the application scenario includes a condition of the communication device maintaining the first model (an internal condition of the communication device, such as power, computing power, storage space, etc., and / or an external condition of the communication device, such as a channel condition in which the communication device is located) and / or a condition of the network device (a configuration of the network device, such as an RRC configuration, or an internal implementation of the network device, such as an antenna deployment).

[0232] As a possible implementation, the generalization range of the first model includes at least one cell to which the first model is applicable, and / or at least one network configuration to which the first model is applicable.

[0233] For example, the generalization range information of the first model can include a cell identifier of at least one cell to which the first model is applicable, or the generalization range information of the first model can include a network configuration identifier corresponding to at least one network configuration to which the first model is applicable, or the generalization range information of the first model includes a cell identifier of at least one cell to which the first model is applicable, and a network configuration identifier of one or more network configurations corresponding to one or more cell identifiers.

[0234] For example, taking the first model as a cell-specific model, the first model is applicable to three network configurations corresponding to associated ID 1, associated ID 2, and associated ID 3 in a cell corresponding to cell identifier 1. The generalization range information of the first model can include cell identifier 1, or the generalization range information of the first model can include cell identifier 1 and associated ID 1 to associated ID 3.

[0235] For example, the first model is a multi-cell model, the first model is applicable to two network configurations corresponding to associated ID1 and associated ID2 in a cell corresponding to cell identifier 1, and a network configuration corresponding to associated ID1 in cell identifier 2. The generalization range information of the first model can include cell identifier 1 and cell identifier 2, or the generalization range information of the first model can include cell identifier 1, associated ID1 and associated ID2 associated with cell identifier 1, cell identifier 2, and associated ID1 associated with cell identifier 2.

[0236] Based on this scheme, the first communication device can accurately obtain each cell and / or network configuration to which each first model is applicable, thereby determining whether to change the second model according to whether the generalization range of each first model in the at least one first model is changed, and facilitating the first communication device to accurately determine the target generalization range of the second model after the change of the second model according to the generalization range of each first model, thereby guiding the change of the second model.

[0237] As a possible implementation, in the case where the first model is changed, the generalization range information of the first model indicates the generalization range before the change of the first model and / or the generalization range after the change of the first model; in the case where the first model is not changed, the generalization range information of the first model indicates the current generalization range of the first model.

[0238] For example, in the case where the first model is changed after the first time, the generalization range information of the first model can include the generalization range of the first model before the change and the generalization range of the first model after the change; in the case where the first model is not changed after the first time, the generalization range of the first model can include the generalization range of the first model at the first time or the current generalization range of the first model at the current time.

[0239] For example, the generalization range information of the first model includes the current generalization range of the first model, or the generalization range information of the first model can also include the generalization range of the first model before the first time and the generalization range of the first model after the first time.

[0240] Based on this scheme, the first communication device can accurately obtain the change trend of the generalization range of the at least one first model, which is conducive to the first communication device to determine the target generalization range of the second model according to the change trend of the generalization range of the at least one first model, thereby guiding the change of the second model, or the first communication device to determine whether to change the second model according to the similarity between the current generalization range of the second model and the target generalization range.

[0241] The model complexity information of the first model can indicate a model complexity of the first model.

[0242] For example, the model complexity of the first model can include at least one of a storage space size required by the first model, a model order of the first model, or a number of model parameters included in the first model.

[0243] As a possible implementation, in a case where the first model is changed, the model complexity information of the first model indicates a model complexity of the first model before the change and / or a model complexity of the first model after the change; in a case where the first model is not changed, the model complexity information of the first model indicates a current model complexity of the first model.

[0244] For example, in a case where the first model is changed after a first time point, the model complexity information of the first model can include a model complexity of the first model before the first time point and a model complexity of the first model after the first time point; in a case where the first model is not changed after the first time point, the generalization range of the first model can include the model complexity of the first model at the first time point or a current model complexity of the first model at a current time point.

[0245] Based on the scheme, the first communication device can accurately obtain the change of the model complexity of each first model, determine the model complexity change trend of at least one first model, and facilitate the first communication device to determine the target model complexity of the second model, and determine whether to change the second model according to the target model complexity, and guide the change of the second model in the process of changing the second model.

[0246] In a case where the first information is the model information of at least one first model, the first communication device determines the management strategy of the second model according to the first information, including the following two implementation manners:

[0247] Manner one, determining whether to change the second model according to the first information.

[0248] As a first possible implementation, in a case where the model information of the first models includes whether the first models are changed, the first communication device can acquire, according to the first information, a proportion (denoted as D1) of the first models in which model change occurs in the at least one first model and / or a proportion (denoted as D2) of the first models in which model change does not occur in the at least one first model, and in a case where D1 is greater than a first threshold or D2 is less than or equal to the first threshold, it is determined that model change occurs in most of the first models, and then model updating or model retraining can be taken as the management strategy of the second model; in a case where D2 is greater than the first threshold or D1 is less than or equal to the first threshold, it is determined that model change does not occur in most of the first models, and then model deactivation can be taken as the management strategy of the second model. The first threshold can be predetermined by a protocol or can be agreed by the first communication device and the second communication device in advance.

[0249] Alternatively, in a case where model change occurs in most of the at least one first model, the first communication device can take model updating or model retraining as the management strategy of the second model; and in a case where model change does not occur in most of the at least one first model, the first communication device can take model deactivation as the management strategy of the second model.

[0250] As a second possible implementation, in a case where the model information of the first models includes a time (denoted as T1) at which the first data set starts to be collected and / or a collection completion time (denoted as T2) of the first data set, the first communication device can regard T1 or T2 as a change time of the first model, and determine a time interval between the change time and a current time for each first model. Then, in a case where, in the first models in which model change occurs, a proportion of the first models in which the time interval between the change time and the current time is less than a second threshold is greater than a proportion of the first models in which the time interval between the change time and the current time is greater than or equal to the second threshold, if a time interval between a change time of the second model and the current time is greater than the second threshold, model updating or model retraining is taken as the management strategy of the second model; and if the time interval between the change time of the second model and the current time is less than or equal to the second threshold, model deactivation is taken as the management strategy of the second model.

[0251] Alternatively, in a case where a time interval (T3) between a change time of the second model and the current time is greater than a time interval between a change time of most of the first models and the current time, the first communication device can take model updating or model retraining as the management strategy of the second model; and in a case where T3 is less than or equal to the time interval between the change time of most of the first models and the current time, the first communication device can take model deactivation as the management strategy of the second model.

[0252] As a third possible implementation, in a case where the model information of the first model includes model complexity information of the first model, the first communication device can acquire, according to the first information, a plurality of model complexities corresponding to the at least one first model at the current time, and acquire a proportion of the first complexity in the model complexities corresponding to the at least one first model, the first complexity being any one of the plurality of model complexities corresponding to the at least one first model. Then the first communication device detects whether the current model complexity (L1) of the second model is the model complexity (L2) with the highest proportion in the model complexities corresponding to the at least one first model, and in a case where L1 is less than L2, takes model updating or model retraining as the management strategy of the second model; in a case where L1 is greater than or equal to L2, takes model deactivation as the management strategy of the second model.

[0253] That is, in a case where the current model complexity of the second model is lower than the current model complexity of the majority of the at least one first model, the first communication device can take model updating or model retraining as the management strategy of the second model; in a case where the current model complexity of the second model is the same as or higher than the current model complexity of the majority of the at least one first model, the first communication device can take model deactivation as the management strategy of the second model.

[0254] As a fourth possible implementation, in a case where the model information of the first model includes performance indicator information of the first model, the first communication device can determine the management strategy of the second model according to the first information in the following four ways:

[0255] Way 1: The first communication device determines the management strategy of the second model according to a proportion of the first model whose performance indicator changes in the at least one first model.

[0256] For example, the first communication device can acquire, according to the first information, a proportion (denoted as D3) of the first model whose performance indicator changes in the at least one first model and / or a proportion (denoted as D4) of the first model whose performance indicator does not change in the at least one first model, and in a case where D3 is greater than a third threshold or D4 is less than or equal to the third threshold, it is determined that the performance indicators of the majority of the first model change, and then model deactivation can be taken as the management strategy of the second model; in a case where D3 is less than or equal to the third threshold or D4 is greater than the third threshold, it is determined that the performance indicators of the majority of the first model do not change, and then model updating or model retraining can be taken as the management strategy of the second model. The third threshold can be predetermined by a protocol or agreed by the first communication device and the second communication device in advance.

[0257] The performance index change of the first model can be understood as a change in performance index of the first model after the first time.

[0258] Alternatively, in the case that the performance index of the majority of the at least one first model changes, the first communication device can deactivate the model as the management strategy of the second model; in the case that the performance index of the majority of the at least one first model does not change, the first communication device can update the model or retrain the model as the management strategy of the second model.

[0259] In mode 2, the first communication device determines the management strategy of the second model according to a proportion of the first model whose performance index meets / does not meet the first performance index in the at least one first model.

[0260] For example, the performance index of the second model cannot meet the first performance index, and the first communication device can determine, according to the first information, whether the current performance index of each first model in the at least one first model meets the first performance index, and further determine a proportion (denoted as A1) of the first model whose current performance index is greater than or equal to the first performance index and / or a proportion (denoted as A2) of the first model whose current performance index is less than the first performance index in the at least one first model. In the case that A1 is greater than a fourth threshold or A2 is less than or equal to the fourth threshold, it is determined that the majority of the first model can meet the first performance index, and the first communication device can update the model or retrain the model as the management strategy of the second model; in the case that A1 is less than or equal to the fourth threshold or A2 is greater than the fourth threshold, it is determined that the majority of the first model cannot meet the first performance index, and the first communication device can deactivate the model as the management strategy of the second model. The fourth threshold can be predetermined by a protocol or agreed by the first communication device and the second communication device in advance.

[0261] Alternatively, in the case that the performance index of the second model cannot meet the first performance index, if the majority of the at least one first model cannot meet the first performance index, the first communication device can deactivate the model as the management strategy of the second model; if the majority of the at least one first model can meet the first performance index, the first communication device can update the model or retrain the model as the management strategy of the second model.

[0262] The performance index of the first model and the performance index of the second model can be understood as the performance index of the model at the first time, or can also be understood as the performance index of the model at the current time.

[0263] In mode 3, the first communication device can determine the management strategy of the second model according to a proportion of the first model whose performance index cannot meet the first performance index in the at least one first model.

[0264] Exemplarily, when the performance indicator of the second model fails to meet the first performance indicator, the first communication device can determine, according to the first information, a time (denoted as t5) when each first model starts to fail to meet the first performance indicator, and then determine a proportion (denoted as D5) of the first models in the at least one first model whose corresponding t5 is earlier than or equal to a time when the second model fails to meet the first performance indicator, and / or a proportion (denoted as D6) of the first models in the at least one first model whose corresponding t5 is later than the time when the second model fails to meet the first performance indicator. When D5 is greater than a fifth threshold or D6 is less than or equal to the fifth threshold, it is determined that the second model has a higher probability of needing to be changed, and the model updating or model retraining can be taken as the management strategy of the second model. When D6 is greater than the fifth threshold or D5 is less than or equal to the fifth threshold, it is determined that the change of the performance indicator of the second model is probably caused by network configuration change, and the second model has a lower probability of needing to be changed, and the model deactivation can be taken as the management strategy of the second model.

[0265] Alternatively, when the performance indicator of the second model fails to meet the first performance indicator at a certain time, if the time when a majority of the first models in the at least one first model starts to fail to meet the first performance indicator is at or before the time, the first communication device can take the model updating or model retraining as the management strategy of the second model. If the time when a majority of the first models in the at least one first model starts to fail to meet the first performance indicator is after the time, the first communication device can take the model deactivation as the management strategy of the second model.

[0266] Option 4: The first communication device can determine the management strategy of the second model according to a proportion of the first models in the at least one first model whose total duration of being able to meet the first performance indicator is greater than a total duration of the second model being able to meet the first performance indicator.

[0267] Exemplarily, the first communication device can determine, according to the first information, a total duration (denoted as s1) of each first model being able to meet the first performance indicator, and then obtain a total duration (denoted as s2) of the second model being able to meet the first performance indicator. When a proportion of the first models in the at least one first model whose corresponding s1 is greater than s2 is higher than a proportion of the first models in the at least one first model whose corresponding s1 is less than or equal to s2, it is determined that the second model has a higher probability of needing to be changed, and the model updating or model retraining can be taken as the management strategy of the second model. When the proportion of the first models in the at least one first model whose corresponding s1 is greater than s2 is not higher than the proportion of the first models in the at least one first model whose corresponding s1 is less than or equal to s2, it is determined that the second model has a lower probability of needing to be changed, and the model deactivation can be taken as the management strategy of the second model.

[0268] Or, in a case that the total time length that the majority of the at least one first model can meet the first performance index is greater than the total time length that the second model can meet the first performance index, the first communication device can take the model updating or model retraining as the management strategy of the second model; in a case that the total time length that the majority of the at least one first model can meet the first performance index is less than or equal to the total time length that the second model can meet the first performance index, the first communication device can take the model deactivation as the management strategy of the second model.

[0269] As a fifth possible implementation, in a case that the model information of the first model includes the state of the first model, the first communication device can acquire, according to the first information, a proportion (denoted as D7) of the first model in the at least one first model that is in the activated state and / or a proportion (denoted as D8) of the first model in the at least one first model that is in the deactivated state, and in a case that D7 is greater than a sixth threshold or D8 is less than or equal to the sixth threshold, it is determined that the majority of the first model is in the activated state, and then the model updating or model retraining can be taken as the management strategy of the second model; in a case that D8 is greater than the sixth threshold or D7 is less than or equal to the sixth threshold, it is determined that the majority of the first model is in the deactivated state, and then the model deactivation can be taken as the management strategy of the second model. The sixth threshold can be predetermined by a protocol, or can be agreed by the first communication device and the second communication device in advance.

[0270] Or, in a case that the majority of the at least one first model is still in the activated state, the first communication device can take the model updating or model retraining as the management strategy of the second model; in a case that the majority of the at least one first model is in the deactivated state, the first communication device can take the model deactivation as the management strategy of the second model, and temporarily not perform the change of the second model.

[0271] Optionally, the state of the first model and the state of the second model can be understood as the state of the model after the first time, or can also be understood as the state of the model at the current time.

[0272] Based on the scheme, the first communication device can determine whether the second model needs to be changed according to the state of the first model in the communication device maintaining the model, and improve the efficiency of the first communication device in determining the management strategy of the second model.

[0273] As a sixth possible implementation, in a case where the model information of the first model includes the generalization range information of the first model, in a case where the model information of the first model includes whether the first model is changed, the first communication device can acquire, according to the first information, a proportion of the first model in which the model generalization range is changed (denoted as D9) and / or a proportion of the first model in which the model is not changed (denoted as D10) in the at least one first model, and in a case where D9 is greater than a seventh threshold or D10 is less than or equal to the seventh threshold, it is determined that the generalization range of the majority of the first model is changed, and then model updating or model retraining can be taken as the management strategy of the second model; in a case where D10 is greater than the seventh threshold or D9 is less than or equal to the seventh threshold, it is determined that the generalization range of the majority of the first model is not changed, and then model deactivation can be taken as the management strategy of the second model. The seventh threshold can be predetermined by a protocol, or can be agreed upon by the first communication device and the second communication device in advance.

[0274] That is, in a case where the generalization range of the majority of the at least one first model is changed, the first communication device can take model updating or model retraining as the management strategy of the second model; in a case where the generalization range of the majority of the at least one first model is not changed, the first communication device can take model deactivation as the management strategy of the second model.

[0275] Optionally, the generalization range of the first model can be understood as the generalization range of the first model at the current time, or can also be understood as the generalization range of the first model after the first time.

[0276] It is worth mentioning that the above embodiments are described by taking the model information of the first model including one item of information of the first model as an example, and in the application process, the model information of the first model can include multiple items of information of the first model, and the process of determining the management strategy of the second model can be implemented in combination with the multiple items of information. For example, each item of information is assigned a priority or a weight, the management strategy of the second model is determined according to the priority of each item of model information, or the management strategy of the second model is determined according to each item of model information respectively, then the weight score corresponding to each management strategy is determined by weighted summation according to the weight of each item of information, and the management strategy with the highest weight score is taken as the management strategy of the second model, which is not limited.

[0277] Mode two, determining the target model information of the second model according to the first information.

[0278] That is, in a case where the first communication device determines, according to the first information, that the management strategy of the second model is model updating or model retraining, the first communication device can further determine, according to the first information, target model information of the second model after retraining or updating of the second model, for example, model complexity, model generalization range, or target performance index, etc.; or determine to change model change related information in the process of changing the second model, for example, the size of the data set used to change the second model.

[0279] As a first possible implementation, in a case where the first information includes generalization range information of the first model, the first communication device can determine, according to the first information, a target generalization range of the second model after updating.

[0280] The first communication device determining, according to the first information, a target generalization range of the second model includes the following two possible ways:

[0281] Way 1: The first information includes generalization range information of the first model, and the first communication device determines, according to a change trend of at least one first model generalization range, a target generalization range of the second model.

[0282] For example, the first communication device can acquire, according to the first information, a proportion (denoted as B1) of the first model that changes the applicable network configuration from the first network configuration to the second network configuration and / or a proportion (denoted as B2) of the first model that changes the applicable network configuration from the second network configuration to the first network configuration. In a case where B1 is greater than a first preset value or B2 is less than or equal to the first preset value, the second network configuration is used as the target network configuration applicable after updating of the second model; in a case where B2 is greater than the first preset value or B1 is less than or equal to the first preset value, the first network configuration is used as the target network configuration applicable after updating of the second model. The first preset value is defined in advance by a protocol or agreed in advance by the first communication device and the second communication device.

[0283] Among them, there is one value for each network configuration parameter in the first network configuration, and there are multiple values for at least part of the network configuration parameters in the second network configuration. That is, the first network configuration is a network configuration that only contains a specific network side resource configuration method, and the second network configuration is a network configuration that contains multiple different network side resource configuration methods. The model with the first network configuration as the applicable network configuration can also be called a scenario model, and the model with the second network configuration as the applicable network configuration can also be called a generalization model.

[0284] Or, in the case that the majority of the at least one first model updates the model type from the scenario model to the generalized model, the first communication apparatus can change the second model to the generalized model through model retraining or model updating; in the case that the majority of the at least one first model updates the model type from the generalized model to the scenario model, the first communication apparatus can change the second model to the scenario model through model retraining or model updating.

[0285] Optionally, in the case that the majority of the first models updates the model type from the scenario model to the generalized model, if the second model is the scenario model, the second model can be changed to the generalized model through model retraining, and if the second model is the generalized model, the generalization range of the second model can be updated according to the model updating manner.

[0286] Similarly, in the case that the majority of the first models updates the model type from the generalized model to the scenario model, if the second model is the generalized model, the second model can be changed to the scenario model through model retraining, and if the second model is the scenario model, the generalization range of the second model can be updated according to the model updating manner.

[0287] Optionally, the generalization range of the first model and the generalization range of the second model can be understood as the generalization range of the model at the current time, or can also be understood as the generalization range of the model after the first time.

[0288] Optionally, the generalization range of the first model and the generalization range of the second model can be understood as the generalization range of the model at the current time, or can also be understood as the generalization range of the model after the first time.

[0289] Optionally, the generalization range of the first model and the generalization range of the second model can be understood as the generalization range of the model at the current time, or can also be understood as the generalization range of the model after the first time.

[0290] Or, in at least one first model, most of the scenario-based models can meet the first performance indicator, and the first communication device can change the second model to a scenario-based model through model retraining or model updating; in at least one first model, most of the generalized models can meet the first performance indicator, and the first communication device can change the second model to a generalized model through model retraining or model updating. The specific change of the second model can refer to the related description in the foregoing embodiments, and will not be repeated here.

[0291] Optionally, whether the generalization range of the first model and the second model and the performance indicator meet the first performance indicator can be understood as whether the generalization range of the model at the current moment and the performance indicator at the current moment meet the first performance indicator, or it can also be understood as whether the generalization range of the model after the first moment and the performance indicator after the first moment meet the first performance indicator.

[0292] As a second possible implementation, when the first information includes the size of the first data set of the first model, the first communication device can determine the size of the data set used in the process of changing or training the second model according to the first information.

[0293] The first communication device determines the size of the data set used in the process of changing or training the second model according to the first information (i.e., the target size) includes the following two possible ways:

[0294] Method 1: The first information includes the size of the first data set of the first model, and the first communication device determines the target size of the data set used in the process of changing or training the second model according to the size of the first data set used by most of the first models.

[0295] For example, the first communication device can obtain the size of the first data set corresponding to at least one first model according to the first information, and obtain the proportion of the first size in the size of the first data set corresponding to at least one first model, and the first size is the size of any first data set in the size of the first data set corresponding to at least one first model. Then the first communication device can determine the target size of the data set used in the process of training or changing the second model according to the first data set size with the highest proportion (denoted as M1). For example, K times of M is taken as the target size of the data set used in the process of training or changing the second model by the first communication device, K is greater than 0, such as 0.25, 0.5, 0.75, 1.5 or 2, etc.

[0296] Optionally, the first communication device can further count a proportion (denoted as B5) of the first models in the at least one first model, in which the size of the corresponding first data set is not less than the size of each first data set, and / or a proportion (denoted as B6) of the first models in the at least one first model, in which the size of the corresponding first data set is less than the size of each first data set, and take B5 or B6 as the proportion corresponding to the size of each first data set. Then, the first communication device can take the size of the first data set with the largest size among the sizes of the at least one first data set corresponding to the proportion greater than or equal to a third preset value as the target size of the data set used in the process of training or modifying the second model, or take the size of the first data set with the smallest size among the sizes of the at least one first data set corresponding to the proportion less than the third preset value as the target size of the data set used in the process of training or modifying the second model.

[0297] Alternatively, in the case that the sizes of the data sets used in the training or modification of the majority of the first models in the at least one first model are N, the first communication device can take K times of N as the target size of the data set used in the process of modifying or training the second model.

[0298] Optionally, the size of the first data set can be understood as the size of the data set used in the training of the first model, or can be understood as the size of the data set used in the most recent modification of the first model, or can be understood as the size of the data set used in the modification of the first model after the first time point.

[0299] In mode 2, the first information includes the size of the first data set of the first model and the performance index information of the first model, and the first communication device determines the target size of the data set used in the modification or training of the second model according to the size of the first data set used in the training of the first model satisfying the first performance index.

[0300] For example, the first communication device can obtain the size of the first data set corresponding to each first model and the performance index according to the first information, and obtain the size of each first data set corresponding to the first model satisfying the first performance index and the proportion of the size of each first data set in the at least one first model. Then, the first communication device can determine the target size of the data set used in the training or modification of the second model according to the first data set with the highest proportion (denoted as M2) among the sizes of the first data sets corresponding to the first model satisfying the first performance index. For example, the first communication device can take K times of M2 as the target size of the data set used in the training or modification of the second model, and K is greater than 0.

[0301] Similarly, the first communication device can also take the proportion of the first model whose corresponding first data set size is not less than the size of each first data set and / or the proportion of the first model whose corresponding first data set size is less than the size of each first data set, as the proportion corresponding to each first data set size, and determine the target size of the data set used for changing or training the second model according to the relationship between the proportion corresponding to each first data set size and the given preset value. The way of determining the target size in the embodiments of the present application can refer to the related description in the foregoing embodiments, and the difference is that the first model participating in the statistics is changed from at least one first model to at least one first model capable of meeting the first performance index. The specific determination manner is not described again.

[0302] Optionally, the size of the first data set can be understood as the size of the data set used for training the first model, or can also be understood as the size of the data set used for the last change of the first model, or can also be understood as the size of the data set used for the change of the first model after the first time. The first model meeting the first performance index can be understood as the current performance index of the first model meeting the first performance index, or can also be understood as the performance index of the first model after the first time meeting the first performance index.

[0303] As a third possible implementation, in a case where the first information includes model complexity information of the first model, the first communication device can determine the target model complexity of the second model according to the first information.

[0304] The first communication device determining the target model complexity of the second model according to the first information includes the following two possible ways:

[0305] Way 1, the first information includes model complexity information of the first model, and the first communication device determines the target model complexity of the second model according to the model complexity change trend of the at least one first model.

[0306] For example, the first communication device can obtain a plurality of model complexities corresponding to the at least one first model according to the first information, and determine the proportion of the first model corresponding to each model complexity (which can also be understood as the probability of each model complexity appearing in the plurality of model complexities corresponding to the at least one first model). Then the first communication device can take the model complexity with the highest proportion as the target model complexity of the second model.

[0307] Or, in a case where the model complexity of the majority of the at least one first model is greater than or equal to P, the first communication device can take P as the target model complexity of the second model.

[0308] Optionally, the model complexity of the first model can be understood as the model complexity of the first model at the current time, or can also be understood as the model complexity of the first model after the first time.

[0309] In mode 2, the first information includes the model complexity information of the first model and the performance index information of the first model, and the first communication device can determine the target model complexity of the second model according to the model complexity of the first model satisfying the first performance index.

[0310] For example, the first communication device can obtain, according to the first information, the model complexity corresponding to each first model satisfying the first performance index in the at least one first model and the proportion of each model complexity, and then the first communication device can take the model complexity with the highest proportion as the target model complexity of the second model.

[0311] Alternatively, in the case that the model complexity of the majority of the first models satisfying the first performance index in the at least one first model is greater than or equal to P, the first communication device can take P as the target model complexity of the second model.

[0312] Optionally, the model complexity of the first model can be understood as the model complexity of the first model at the current time, or can also be understood as the model complexity of the first model after the first time.

[0313] As a fourth possible implementation, in the case that the first information includes the performance index information of the first model, the first communication device can determine the target performance index of the second model according to the first information.

[0314] For example, the first communication device can further obtain, according to the first information, the proportion of the first model in which the performance index changes from the first index to the second index in the at least one first model, such as the proportion of the first model in which the prediction accuracy changes from 90% to 80%, the proportion of the first model in which the prediction accuracy changes from 80% to 90%, the proportion of the first model in which the prediction accuracy changes from 80% to 70%, and the proportion of the first model in which the prediction accuracy changes from 70% to 80%, etc. Then, the proportion of the first model corresponding to each possible performance index change is counted, and the performance index of the first model after the change in the performance index change mode with the highest proportion is taken as the target performance index of the second model after the update.

[0315] Based on the above scheme, in a case that the first information includes model information of the at least one first model, the first communication device can obtain a model change trend of the at least one first model or a model change trend of a first model in the at least one first model that meets one or more specified conditions (such as meeting a first performance indicator and / or being applicable to a first network configuration) by statistical analysis of the model information of each first model, thereby effectively guiding the first communication device to maintain the second model and efficiently deciding whether to change the second model and the specific strategy of changing the second model.

[0316] In a possible implementation, the statistical information corresponding to the model information of the at least one first model indicates at least one of the following: a proportion of the first models in the at least one first model that have changed, a proportion of the first models in the at least one first model that have a performance indicator greater than or equal to the first performance indicator, a proportion of the first models in the at least one first model that have a performance indicator less than the first performance indicator, a proportion of the first models in the at least one first model that have a changed generalization range, a proportion of the first models in the at least one first model that have a changed network configuration from the first network configuration to the second network configuration, a proportion of the first models in the at least one first model that have a changed network configuration from the second network configuration to the first network configuration, a proportion of the first models in the at least one first model that have a changed performance indicator, a proportion of the first models in the at least one first model that have an unchanged performance indicator, a proportion of the first models in the at least one first model that have an unchanged performance indicator and the applicable network configuration being the first network configuration, a proportion of the first models in the at least one first model that have an unchanged performance indicator and the applicable network configuration being the second network configuration, a proportion of the first models in the at least one first model that are in an activated state, a proportion of the first complexity in the model complexity corresponding to the at least one first model, or a proportion of the first size in the size of the first data set corresponding to the at least one first model.

[0317] The meanings of the first data set, the first network configuration, the second network configuration, the first size, and the first complexity can refer to the related descriptions in the foregoing embodiments, and the meanings of the various statistical information can also refer to the related descriptions in the foregoing embodiments.

[0318] In a case that the first information includes statistical information corresponding to the model information of the at least one first model, the first communication device can directly obtain the performance change trend and / or the model change trend of the at least one first model according to the statistical information included in the first information, thereby directly determining the management strategy of the second model, the size of the data set used in the training or changing of the second model, or the target value of at least one model information of the second model after the training or changing according to the statistical information corresponding to the model information of the at least one first model included in the first information.

[0319] The first communication device determines the implementation manner of the management strategy of the second model according to the statistical information corresponding to the model information of the at least one first model, which is similar to the implementation manner of the management strategy of the second model determined by the first communication device according to the model information of the at least one first model in the foregoing embodiment, and the difference is that in the process of determining the implementation manner of the management strategy of the second model according to the model information of the at least one first model, the first communication device needs to perform statistical analysis on the model information of each first model by itself, and in the process of determining the management strategy of the second model according to the statistical information corresponding to the model information of the at least one first model, the first communication device can directly obtain the statistical analysis result of the model information, without the need for additional data statistics and analysis. The specific implementation manner can refer to the related description in the foregoing embodiment, and will not be described here again.

[0320] It is worth mentioning that the statistical information corresponding to the model information of the at least one first model can be statistical information generated according to the model information of the first model at the current time, or can be statistical information generated according to the model information of the first model after the first time, which is not limited.

[0321] In addition, in order to facilitate understanding, the foregoing scheme is described by taking, as an example, that the statistical information corresponding to the model information of the at least one first model is generated according to one or two model information of the change state of the first model, the performance index information of the first model, the generalization range of the first model, the network configuration applicable to the first model, the size of the first data set corresponding to the first model, and the model complexity information of the first model as a statistical condition.

[0322] In the application process, the statistical information can also be determined according to other at least two pieces of model information in the above model information. For example, the statistical information corresponding to the model information of the at least one first model can further include: a proportion of the first model in the at least one first model in which the performance indicator meets the first performance indicator and the applicable network configuration changes from the first network configuration to the second network configuration, a proportion of the first model in the at least one first model in which the performance indicator meets the first performance indicator and the applicable network configuration changes from the second network configuration to the first network configuration, a proportion of the first model in the at least one first model in which the change occurs and the performance indicator does not change, a proportion of the first model in the at least one first model in which the change does not occur and the performance indicator changes, a proportion of the first model in the at least one first model in which the generalization range changes and the performance indicator does not change, a proportion of the first model in the at least one first model in which the generalization range does not change and the performance indicator changes, a proportion of the first model in the at least one first model in which the generalization range changes and the performance indicator does not meet the first performance indicator, or a proportion of the first model in the at least one first model in which the generalization range changes and the performance indicator meets the first performance indicator. Other possible generation of the statistical information corresponding to the model information of the at least one first model according to the statistical results of at least two pieces of model information will not be listed here.

[0323] In a possible implementation, the first information includes first indication information and / or second indication information, the first indication information is used to indicate the management policy of the second model, and the second indication information is used to indicate a target value of at least one piece of model information of the second model.

[0324] For example, the first indication information can be implemented by a bitmap containing a plurality of bits, each bit in the bitmap corresponds to at least one management policy, in the case of setting the bit to 0, it is indicated that the management policy of the second model does not include the at least one management policy corresponding to the bit, and in the case of setting the bit to 1, it is indicated that the management policy of the second model includes the at least one management policy corresponding to the bit. Alternatively, in the case of setting the bit to 0, it is indicated that the management policy of the second model includes the at least one management policy corresponding to the bit, and in the case of setting the bit to 1, it is indicated that the management policy of the second model does not include the at least one management policy corresponding to the bit. The second indication information can be implemented by a plurality of bit sequences or information blocks, each bit sequence or information block is used to indicate a target value of a piece of model information.

[0325] Optionally, the first information can include one or more of the first indication information, the second indication information, the model information of the at least one first model, or the statistical information corresponding to the model information of the at least one first model, without limitation.

[0326] In a possible implementation, before step S502, the first communication device can further send second information to the second communication device. Correspondingly, the second communication device receives the second information from the second communication device. The second information is used to request the first information.

[0327] That is, the first communication device can send the second information used to request the first information to the first communication device before determining the management policy of the second model, and then the second communication device triggers the process of sending the first information from the second communication device to the first communication device after receiving the second information from the first communication device.

[0328] Optionally, after receiving the second information from the first communication device, the second communication device performs step S501, including the following two possible implementation manners:

[0329] Manner 1: The second communication device requests model information of the first model from at least one third communication device, and generates the first information according to the received model information.

[0330] Manner 2: The second communication device determines model information of at least one first model from model information of a plurality of communication devices stored in advance, and generates the first information according to the model information of the at least one first model.

[0331] In addition, the above embodiment is that the first information is generated by the second communication device through statistical analysis on model information of a plurality of third communication devices in communication connection with the second communication device. In the application process, the manner of determining the first information by the second communication device can also be receiving the first information from the core network device, and sending the first information from the core network device to the first communication device.

[0332] That is, the model information of the first communication device and the model information of the third communication device can be reported to the core network device through the second communication device or directly reported to the core network device. In the case that the first communication device requests the first information, the second communication device can send request information of the first information to the core network device, and then the core network device generates the first information through statistical analysis on the maintained model information according to the request information of the second communication device, and sends the first information to the first communication device through the second communication device. This is conducive to reducing the storage space and computing power consumption in the process of determining the first information by the second communication device.

[0333] In a possible implementation, the first communication device sends model information of the second model to the second communication device. Correspondingly, the second communication device receives the model information of the second model from the second communication device.

[0334] That is, the first communication device can report model information of a model (a second model) maintained by the first communication device to a second communication device in communication connection with the first communication device, the second communication device can store the model information of the second model after receiving the model information reported by the first communication device, and in a case where a fourth communication device requests first information for determining a model management strategy, the second communication device can generate the first information requested by the fourth communication device according to the model information of the second model. That is, the model information of the second model reported by the first communication device can be used as model information of a first model in the first information requested by other communication devices, and the model information of the first model included in the first information received by the first communication device can also be model information reported by a third communication device in communication connection with the second communication device.

[0335] Optionally, the model information of the second model is model information of the second model after a second time, or the model information of the second model is model information of the second model after being changed. The second time can be predefined by a protocol, can be indicated by the second communication device to the first communication device, or can be determined by the first communication device and sent to the second communication device.

[0336] That is, the first communication device can send model information of the second model at the second time or after the second time to the second communication device according to the predetermined second time, or the first communication device can directly report model information of the second model after being changed to the second communication device after each change of the second model.

[0337] Based on the above scheme, the first communication device has the ability to report model information of a model maintained by the first communication device according to a specific event (changing the second model or determining the second time), and the second communication device has the ability to receive or query model information of a model maintained by other communication devices in communication connection with the second communication device, which is beneficial to the second communication device to generate first information for determining a management strategy of the second model.

[0338] In a possible implementation, the second communication device broadcasts the first information when a preset condition is met.

[0339] For example, the second communication device broadcasts the first information currently determined when a time interval between a current time and a completion time of the last broadcast of the first information reaches a preset period interval. That is, the second communication device can periodically determine the current first information and broadcast according to the preset period interval.

[0340] For example, the second communication device can dynamically maintain model information from multiple communication devices, and monitor the model performance indicators of the communication devices (the third communication device and the first communication device) in communication connection with the second communication device. In the case that the model performance of the first communication device decreases or the model performance indicators cannot meet the given performance indicators, the first information at the current time is sent to the first communication device.

[0341] Optionally, the second communication device can detect the model performance indicators of the communication devices in communication connection therewith by monitoring the performance of the communication devices, or by receiving performance indicator information from the communication devices. That is, the performance indicators of the models maintained by any communication device can be monitored by the second communication device, or monitored by the communication device and fed back to the second communication device.

[0342] FIG. 6 is a service flowchart of the communication method in the above embodiment. Referring to (a) of FIG. 6, the first communication device is terminal 1, the third communication device is terminal 2, the second communication device is a base station, and the first information is the performance indicator information of at least one first model. After the random access is completed, terminal 1 and terminal 2 can report the model information of part or all of the models maintained by the terminals to the base station according to the model information query request of the base station or preset rules. The base station statistics the model information of the models maintained by the terminals in communication connection therewith. Terminal 1 and terminal 2 determine the current model suitable for the current network configuration of the base station and activate the model according to the current network configuration of the base station, and the base station monitors the performance indicators of the current model of terminal 1 and the performance indicators of the current model of terminal 2. In the case that the performance of the current model of terminal 1 decreases or terminal 1 requests the performance indicator information of the model of other terminals from the base station, the base station can statistics the performance indicator information of other terminals (such as terminal 2) at the current time or after the first time, and sends the performance indicator information or the statistical information of the performance indicator information to terminal 1. After receiving the performance indicator information or the statistical information of the performance indicator information, terminal 1 determines the management strategy of the current model according to the model performance change of other terminals, and collects training data and changes the model in the case that the current model needs to be changed.

[0343] Referring to (b) in FIG. 6, taking the first communication device as terminal 1, the third communication device as terminal 2, the second communication device as a base station, and the first information as the performance index information of at least one first model as an example. After the random access is completed, terminal 1 and terminal 2 can report the model information of part or all of the models maintained by the terminal to the base station according to the model information query request of the base station or a preset rule. Then, the base station can statistically collect the model information of the models maintained by the terminals connected thereto, report the model information of terminal 1 and the model information of terminal 2 to the core network device, and update the model registration information maintained by the core network device. After terminal 1 and terminal 2 determine the model applicable to the current network configuration of the base station and activate the model, the base station can monitor the performance index of the current model of terminal 1 and the performance index of the current model of terminal 2. In the case that the performance of the current model of terminal 1 decreases or terminal 1 requests the model change information of other terminal models from the base station, the base station can generate statistical information (first information) corresponding to the model information of a plurality of models applicable to the same network configuration as the current model of terminal 1 according to the model information of other terminals (such as terminal 2) at the current moment or after the first moment, and send the statistical information to terminal 1. After receiving the first information, terminal 1 can determine the management strategy of the current model according to the first information.

[0344] Based on the above scheme, in the process of determining the management strategy of the second model maintained by the first communication device, the first communication device receives the first information from the second communication device, the first information includes the model information of at least one first model having an association relationship with the second model and / or the statistical information corresponding to the model information of the at least one first model, and then the first communication device determines the management strategy used in the process of maintaining the second model according to the first information. That is, in the process of determining the management strategy of the second model, the first communication device takes the model information of the first model having an association relationship with the second model or the statistical information corresponding to the model information of the first model as prior information, and determines the management strategy of the model according to the prior information. In the process of determining the management strategy of the model, a large amount of training data does not need to be collected and the data statistical characteristics of the training data do not need to be analyzed, which is beneficial to reduce the resource overhead required for the first communication device to determine the management strategy of the second model in the process of maintaining the second model, and improve the efficiency of the first communication device in maintaining the second model.

[0345] The above describes the method provided by the present application, in addition, the present application also provides a communication device for implementing the functions described in the above method embodiments.

[0346] It should be noted that the communication device includes hardware structure and / or software module corresponding to each function in order to realize the above functions. Those skilled in the art can easily understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented in hardware or a combination of hardware and computer software. Whether a certain function is implemented in hardware or computer software driven hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0347] The embodiments of the present application can divide the functional modules of the communication device according to the above-mentioned method embodiments. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. The integrated module can be implemented in the form of hardware or software functional module. It should be noted that the division of modules in the embodiments of the present application is illustrative, and is only a logical functional division. Actual implementation can have another division manner.

[0348] FIG. 7 shows a structural schematic diagram of a communication device 70. The communication device 70 includes a processing module 701 and a transceiver module 702. The communication device 70 can be used to realize the functions of the first communication device or the second communication device.

[0349] In some embodiments, the communication device 70 can further include a storage module (not shown in FIG. 7) for storing program instructions and data.

[0350] In some embodiments, the transceiver module 702, which can also be referred to as a transceiver unit, is used to realize the sending and / or receiving functions. The transceiver module 702 can be composed of a transceiver circuit, a transceiver, a transceiver or a communication interface.

[0351] In some embodiments, the transceiver module 702 can include a receiving module and a sending module, which are respectively used to perform the receiving and sending steps of the first communication device or the second communication device in the above-mentioned method embodiments, and / or are used to support other processes of the technology described herein; the processing module 701 can be used to perform the processing steps of the first communication device or the second communication device in the above-mentioned method embodiments, and / or are used to support other processes of the technology described herein.

[0352] When the communication device 70 is used to realize the functions of the first communication device, in one possible implementation, the transceiver module 702 is used to send second information to the second communication device, and the second information is used to request the first information.

[0353] In a possible implementation, the transceiver module 702 is further configured to send model information of the second model to the second communication apparatus.

[0354] In the case that the communication apparatus 70 is configured to implement the function of the second communication apparatus, in a possible implementation, the transceiver module 702 is configured to receive second information from the second communication apparatus, the second information being used to request the first information.

[0355] In a possible implementation, the transceiver module 702 is configured to receive model information of the second model from the second communication apparatus.

[0356] Wherein, all the related content of each step involved in the above method embodiments can be referred to the function description of the corresponding functional module, which will not be repeated here.

[0357] In the present application, the communication apparatus 70 can be in the form of integrated division of each functional module. The "module" here can refer to a specific application-specific integrated circuit (ASIC), a circuit, a processor and a memory executing one or more software or firmware programs, an integrated logic circuit, and / or other devices that can provide the above functions.

[0358] In some embodiments, when the communication apparatus 70 in FIG. 7 is a chip or a chip system, the function / implementation process of the transceiver module 702 can be implemented through the input / output interface (or communication interface) of the chip or chip system, and the function / implementation process of the processing module 701 can be implemented through the processor (or processing circuit) of the chip or chip system.

[0359] Since the communication apparatus 70 provided by the present embodiment can execute the above method, the technical effects it can obtain can be referred to the above method embodiments, which will not be repeated here.

[0360] As a possible product form, the first communication apparatus or the second communication apparatus described in the embodiments of the present application can be implemented using one or more field programmable gate arrays (FPGAs), programmable logic devices (PLDs), 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 present application.

[0361] As another possible product form, the first communication device or the second communication device described in the embodiments of the present application can be implemented by a general bus architecture. For ease of illustration, refer to FIG. 8, which is a structural schematic diagram of a communication device 800 provided by the embodiments of the present application, the communication device 800 including a processor 801 and a transceiver 802. The communication device 800 can be a first communication device, or a chip or chip system therein; or the communication device 800 can be a second communication device, or a chip or module therein. FIG. 8 only shows the main components of the communication device 800. In addition to the processor 801 and the transceiver 802, the communication device can further include a memory 803, and an input and output device (not shown in the figure).

[0362] Optionally, the processor 801 is mainly used for processing communication protocols and communication data, and controlling the entire communication device, executing software programs, processing data of the software programs, so as to implement the methods provided in the method embodiments described above. The memory 803 is mainly used for storing software programs and data. The transceiver 802 can include a radio frequency circuit and an antenna, the radio frequency circuit being mainly used for conversion between a baseband signal and a radio frequency signal and processing the radio frequency signal. The antenna is mainly used for transceiving radio frequency signals in the form of electromagnetic waves. The input and output device, such as a touch screen, a display screen, a keyboard, etc., is mainly used for receiving data input by a user and outputting data to the user.

[0363] Optionally, the processor 801, the transceiver 802, and the memory 803 can be connected through a communication bus.

[0364] When the communication device is powered on, the processor 801 can read the software programs in the memory 803, interpret and execute instructions of the software programs, and process data of the software programs. When data needs to be transmitted wirelessly, the processor 801 performs baseband processing on the data to be transmitted, and outputs a baseband signal to the radio frequency circuit, the radio frequency circuit performs radio frequency processing on the baseband signal, and transmits the radio frequency signal in the form of electromagnetic waves through the antenna. When data is transmitted to the communication device, the radio frequency circuit receives the radio frequency signal through the antenna, converts the radio frequency signal into a baseband signal, and outputs the baseband signal to the processor 801, the processor 801 converts the baseband signal into data and processes the data.

[0365] In another implementation, the radio frequency circuit and the antenna can be arranged independently of the processor performing baseband processing, for example, in a distributed scenario, the radio frequency circuit and the antenna can be arranged remotely from the communication device.

[0366] In some embodiments, in a hardware implementation, those skilled in the art can conceive that the above-mentioned communication device 70 can adopt the form of the communication device 800 shown in FIG. 8.

[0367] As an example, the function / implementation process of the processing module 701 in FIG. 7 can be implemented by invoking the computer-executed instructions stored in the memory 803 by the processor 801 in the communication apparatus 800 shown in FIG. 8. The function / implementation process of the transceiver module 702 in FIG. 7 can be implemented by the transceiver 802 in the communication apparatus 800 shown in FIG. 8.

[0368] As yet another possible product form, the first communication apparatus or the second communication apparatus in the present application can adopt the constituent structure shown in FIG. 9, or include the components shown in FIG. 9. FIG. 9 is a constituent diagram of a communication apparatus 900 provided in the present application, which can be a chip or a system on chip in the first communication apparatus or the second communication apparatus; or can be a module or a chip or a system on chip in the second communication apparatus.

[0369] As shown in FIG. 9, the communication apparatus 900 includes at least one processor 901, and at least one communication interface (only one communication interface 904 is shown in FIG. 9 by way of example, and the processor 901 is taken as an example for description). Optionally, the communication apparatus 900 can further include a communication bus 902 and a memory 903.

[0370] The processor 901 can be a general central processing unit (CPU), a general processor, a network processor (NP), a digital signal processor (DSP), a microprocessor, a microcontroller, a PLD, or any combination thereof. The processor 901 can also be other apparatuses with processing function, such as a circuit, a device, or a software module, without limitation.

[0371] The communication bus 902 is used to connect different components in the communication apparatus 900, so that different components can communicate. The communication bus 902 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in FIG. 9, but it does not mean that there is only one bus or only one type of bus.

[0372] The communication interface 904 is configured to communicate with other devices or communication networks. For example, the communication interface 904 can be a module, a circuit, a transceiver, or any device capable of realizing communication. Alternatively, the communication interface 904 can also be an input / output interface in the processor 901, configured to realize signal input and signal output of the processor.

[0373] The memory 903 can be a device with a storage function, configured to store instructions and / or data. The instructions can be a computer program.

[0374] For example, the memory 903 can be a read-only memory (ROM) or other type of static storage device that can store static information and / or instructions, or can be a random access memory (RAM) or other type of dynamic storage device that can store information and / or instructions, or can be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disk storage, a magneto-optical disk, a magnetic disk storage medium or other magnetic storage device, and the like, without limitation.

[0375] It should be noted that the memory 903 can exist independently of the processor 901, or can be integrated with the processor 901. The memory 903 can be located in the communication device 900, or can be located outside the communication device 900, without limitation. The processor 901 can be configured to execute instructions stored in the memory 903 to realize the method provided in the embodiments described below.

[0376] As an optional implementation manner, the communication device 900 can further include an output device 905 and an input device 906. The output device 905 communicates with the processor 901, and can display information in various ways. For example, the output device 905 can be a liquid crystal display (LCD), a light emitting diode (LED) display device, a cathode ray tube (CRT) display device, or a projector, etc. The input device 906 communicates with the processor 901, and can receive user input in various ways. For example, the input device 906 can be a mouse, a keyboard, a touch screen device, a sensor device, etc.

[0377] In some embodiments, the communication apparatus 70 shown in Fig. 7 can take the form of the communication apparatus 900 shown in Fig. 9, which can be implemented in hardware.

[0378] As an example, the function / implementation process of the processing module 701 in Fig. 7 can be implemented by invoking the computer-executed instructions stored in the memory 903 by the processor 901 in the communication apparatus 900 shown in Fig. 9. The function / implementation process of the transceiver module 702 in Fig. 7 can be implemented by the communication interface 904 in the communication apparatus 900 shown in Fig. 9.

[0379] It should be noted that the structure shown in Fig. 9 does not constitute a specific limitation on the first communication apparatus or the second communication apparatus. For example, in some other embodiments of the present application, the first communication apparatus or the second communication apparatus can include more or fewer components than those shown, or combine certain components, or split certain components, or different arrangement of components. The components shown can be implemented in hardware, software, or a combination of software and hardware.

[0380] In some embodiments, the present application also provides a communication apparatus, which includes a processor for implementing the method in any of the above method embodiments.

[0381] As a possible implementation, the communication apparatus further includes a memory. The memory is used to save necessary computer programs and data. The computer programs can include instructions, and the processor can invoke the instructions in the computer programs stored in the memory to instruct the communication apparatus to perform the method in any of the above method embodiments. Of course, the memory can also not be in the communication apparatus.

[0382] As another possible implementation, the communication apparatus further includes an interface circuit, which is a code / data read-write interface circuit, and is used to receive computer-executed instructions (computer-executed instructions are stored in the memory, which can be directly read from the memory or can pass through other devices) and transmit them to the processor.

[0383] As yet another possible implementation, the communication apparatus further includes a communication interface, which is used to communicate with modules outside the communication apparatus.

[0384] It can be understood that the communication apparatus can be a chip or a chip system. When the communication apparatus is a chip system, it can be composed of a chip or can include a chip and other discrete devices, and the embodiments of the present application do not make a specific limitation thereon.

[0385] The present application also provides a computer-readable storage medium, which stores a computer program or instructions, and the computer program or instructions are executed by a computer to realize the functions of any of the above method embodiments.

[0386] The application further provides a computer program product, which, when executed by a computer, realizes the functions of any of the method embodiments described above.

[0387] Those skilled in the art can understand that, for the convenience and brevity of description, the specific working processes of the system, device and unit described above can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0388] It can be understood that the system, device and method described in the application can also be implemented in other manners. For example, the device embodiments described above are merely schematic; the division of the units is merely a logical function division; and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0389] The units described as separate components can or can not be physically separate, i.e., can be located in one place, or can be distributed on a plurality of network units. The components shown as units can or can not be physical units. Part or all of the units can be selected according to actual needs to achieve the purposes of the embodiments.

[0390] In addition, each functional unit in each embodiment of the application can be integrated into a processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0391] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software programs, implementation can be, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device containing one or more servers, data centers, etc., that can be integrated with the medium. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive (SSD)). In this embodiment, the computer may include the aforementioned apparatus.

[0392] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0393] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the scope of this application. Accordingly, this specification and drawings are merely illustrative descriptions of the application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of the claims and their equivalents, this application is also intended to include such modifications and modifications.

Claims

A communication method characterized by comprising: The method is applied to a first communication device, and the method comprises: receiving first information from a second communication device, the first information comprising model information of at least one first model and / or statistical information corresponding to the model information of the at least one first model; determining a management policy of a second model according to the first information, the first model being associated with the second model, and the second model being a model maintained by the first communication device. The method of claim 1, wherein The model information of the first model indicates at least one of the following: performance indicator information of the first model; a state of the first model; whether the first model has changed; a time at which a first data set starts to be collected; a time at which the collection of the first data set is completed; a size of the first data set; generalization range information of the first model; model complexity information of the first model; The first data set is used to train or change the first model. The method according to claim 2, characterized in that The generalization range comprises at least one cell to which the first model is applicable, and / or at least one network configuration to which the first model is applicable. The method according to claim 2 or 3, characterized in that Whether the first model has changed comprises whether the first model has changed after a first time. The method according to any one of claims 2 to 4, characterized in that In a case where the first model has changed, the generalization range information of the first model indicates a generalization range before the first model changes and / or a generalization range after the first model changes; In a case where the first model has not changed, the generalization range information of the first model indicates a current generalization range of the first model. The method according to any one of claims 2 to 5, characterized in that In a case where the first model has changed, the size of the first data set indicates a size of a data set used to change the first model; In a case where the first model has not changed, the size of the first data set indicates a size of a data set used to train the first model. The method according to any one of claims 2 to 6, characterized in that In a case where the first model has changed, the model complexity information of the first model indicates a model complexity before the first model changes and / or a model complexity after the first model changes; In a case where the first model has not changed, the model complexity information of the first model indicates a current model complexity of the first model. The method according to any one of claims 3 to 7, characterized in that The statistical information corresponding to the model information of the at least one first model indicates at least one of the following: a proportion of first models in the at least one first model that have changed; a proportion of first models in the at least one first model that have performance indicators greater than or equal to a first performance indicator; a proportion of first models in the at least one first model that have performance indicators less than a first performance indicator; a proportion of first models in the at least one first model whose generalization ranges have changed; a proportion of first models in the at least one first model whose applicable network configurations have changed from a first network configuration to a second network configuration; a proportion of first models in the at least one first model whose applicable network configurations have changed from a second network configuration to a first network configuration; a proportion of first models in the at least one first model whose performance indicators have changed; a proportion of first models in the at least one first model whose performance indicators have not changed; In the at least one first model, the performance indicator does not change, and the proportion of the first model corresponding to the first network configuration is the proportion of the first model in the at least one first model. In the at least one first model, the performance indicator does not change, and the proportion of the first model corresponding to the second network configuration is the proportion of the first model in the at least one first model. In the at least one first model, the proportion of the first model in an active state is the proportion of the first model in the at least one first model. The proportion of the first complexity in the model complexity corresponding to the at least one first model is the proportion of any one of the model complexity corresponding to the at least one first model. The proportion of the first size in the size of the first data set corresponding to the at least one first model is the size of any one of the first data set corresponding to the at least one first model, and the first data set is used to change the first model. In the first network configuration, each network configuration parameter has one value, and in the second network configuration, at least part of the network configuration parameters has multiple values. The method of claim 8, wherein In the at least one first model, the proportion of the first model that changes after the first time is the proportion of the first model in the at least one first model that changes after the first time. The method according to any one of claims 2 to 9, characterized in that The performance indicator information of the first model is used to indicate the performance indicator of the first model. The method of claim 10, wherein The performance indicator information of the first model is also used to indicate time information of the first model satisfying the first performance indicator and / or time information of the first model not satisfying the first performance indicator. The time information of the first model satisfying the first performance indicator indicates at least one of the start time, the end time, or the duration of the first model satisfying the first performance indicator. The time information of the first model not satisfying the first performance indicator indicates at least one of the start time, the end time, or the duration of the first model not satisfying the first performance indicator. The method according to any one of claims 1 to 11, characterized in that The state of the first model includes at least one of the following: support state, available state, applicable state, active state, or configuration state. The method according to any one of claims 1 to 12, characterized in that The method further comprises: sending second information to the second communication device, wherein the second information is used to request the first information. The method according to any one of claims 1 to 13, characterized in that The method further comprises: sending model information of the second model to the second communication device. The method of claim 14, wherein The model information of the second model is the model information of the second model after the second time, or The model information of the second model is the model information of the second model after the change. The method according to any one of claims 1 to 15, characterized in that The at least one first model is a model maintained by at least one third communication device, and the network configuration applicable to the first model includes the network configuration applicable to the second model. The method according to any one of claims 1 to 16, characterized in that The management strategy includes one of the following: model retraining, model updating, model activation, model switching, or model deactivation. A communication method characterized by comprising: The method applied to the second communication device comprises: determining first information, wherein the first information includes model information of at least one first model and / or statistical information corresponding to the model information of the at least one first model; sending the first information to the first communication device; The first information is used to determine a management policy of a second model, and the first model is associated with the second model. The method of claim 18, wherein The model information of the first model indicates at least one of the following: a performance indicator of the first model; a state of the first model; whether the first model is changed; a time when the first data set is started to be collected; a time when the collection of the first data set is completed; a size of the first data set; generalization range information of the first model; model complexity information of the first model. The first data set is used to train or change the first model. The method of claim 19, wherein The generalization range includes at least one cell to which the first model is applicable, and / or at least one network configuration to which the first model is applicable. The method according to claim 19 or 20, characterized in that Whether the first model is changed includes whether the first model is changed after a first time. The method according to any one of claims 19 to 21, characterized in that In a case where the first model is changed, the generalization range information of the first model indicates a generalization range before the first model is changed and / or a generalization range after the first model is changed. In a case where the first model is not changed, the generalization range information of the first model indicates a current generalization range of the first model. The method according to any one of claims 19 to 22, characterized in that In a case where the first model is changed, the size of the first data set indicates a size of a data set used to change the first model. In a case where the first model is not changed, the size of the first data set indicates a size of a data set used to train the first model. The method according to any one of claims 19 to 23, characterized in that In a case where the first model is changed, the model complexity information of the first model indicates a model complexity before the first model is changed and / or a model complexity after the first model is changed. In a case where the first model is not changed, the model complexity information of the first model indicates a model complexity of the first model. The method according to any one of claims 20 to 24, characterized in that The statistical information corresponding to the model information of the at least one first model indicates at least one of the following: a proportion of the first models that are changed in the at least one first model; a proportion of the first models whose performance indicators are greater than or equal to a first performance indicator in the at least one first model; a proportion of the first models whose performance indicators are less than the first performance indicator in the at least one first model; a proportion of the first models whose generalization ranges are changed in the at least one first model; a proportion of the first models whose applicable network configurations are changed from a first network configuration to a second network configuration in the at least one first model; a proportion of the first models whose applicable network configurations are changed from the second network configuration to the first network configuration in the at least one first model; a proportion of the first models whose performance indicators are changed in the at least one first model; a proportion of the first models whose performance indicators are not changed in the at least one first model; a proportion of the first models whose performance indicators are not changed and whose applicable network configurations are the first network configuration in the at least one first model; a proportion of the first models whose performance indicators are not changed and whose applicable network configurations are the second network configuration in the at least one first model; A proportion of the first models in the at least one first model that are in an active state; A proportion of the first complexity in the model complexity corresponding to the at least one first model, the first complexity being any of the model complexity corresponding to the at least one first model; A proportion of the first size in the size of the first data set corresponding to the at least one first model, the first size being the size of any of the first data set corresponding to the at least one first model, the first data set being used to change the first model; The first network configuration has one value for each network configuration parameter, and the second network configuration has multiple values for at least some network configuration parameters. The method of claim 25, wherein A proportion of the first models in the at least one first model that are changed after the first time point. The method according to any one of claims 19 to 26, characterized in that The performance indicator information of the first model is used to indicate a performance indicator of the first model. The method of claim 27, wherein The performance indicator information of the first model is also used to indicate time information of the first model meeting a first performance indicator and / or time information of the first model not meeting the first performance indicator. The time information of the first model meeting the first performance indicator indicates at least one of a start time, an end time, or a duration of the first model meeting the first performance indicator. The time information of the first model not meeting the first performance indicator indicates at least one of a start time, an end time, or a duration of the first model not meeting the first performance indicator. The method according to any one of claims 18 to 28, characterized in that The state of the first model includes at least one of a support state, an available state, a suitable state, an active state, or a configuration state. The method according to any one of claims 18 to 29, characterized in that The method further includes receiving second information from the first communication device, the second information being used to request the first information. The method according to any one of claims 18 to 30, characterized in that The method further includes receiving model information of the second model from the first communication device. The method of claim 31, wherein The model information of the second model is model information of the second model after a second time point, or The model information of the second model is model information of the second model after being changed. The method according to any one of claims 18 to 32, characterized in that The at least one first model is a model maintained by at least one third communication device, and the network configuration applicable to the first model includes a network configuration applicable to the second model. The method according to any one of claims 18 to 33, characterized in that The management strategy includes one of the following multiple items: model retraining, model updating, model activation, model switching, or model deactivation. A communication device characterized by comprising: The communication device includes a processor, and the processor is used to run a computer program or instructions to cause the communication device to perform the method of any one of claims 1-17 or to cause the communication device to perform the method of any one of claims 18-34. A chip or chip system, characterized in that The chip or chip system includes a processor coupled with a memory, and the memory is used to store a program or instructions, when the program or instructions are executed by the processor, the method of any one of claims 1-17 is executed, or the method of any one of claims 18-34 is executed. A computer-readable storage medium, characterized by A computer readable storage medium stores computer instructions or programs that, when run on a computer, cause the method of any of claims 1-17 to be performed, or cause the method of any of claims 18-34 to be performed. A computer program product, characterized by The computer program product comprises computer instructions; when part or all of the computer instructions are run on a computer, cause the method of any of claims 1-17 to be performed, or cause the method of any of claims 18-34 to be performed.

Images