Wireless communication method and device

Abstract

The embodiment of the invention provides a wireless communication method and device, and a first communication device can monitor the prediction performance of a first network model based on a measurement result corresponding to a first monitoring signal set or a confidence coefficient corresponding to a predicted spatial filter in a first prediction data set. Measurement overhead for model performance monitoring can be reduced or avoided, so that the model monitoring performance is improved.

Description

Wireless communication method and device Technical Field

[0001] The present invention relates to the field of communications, and more specifically, to a method and device for wireless communications. Background Art

[0002] In the New Radio (NR) system, artificial intelligence (AI) / machine learning (ML) can be introduced to improve system performance. For example, the introduction of AI / ML models for beam prediction, that is, beam prediction through the trained AI / ML model, improves the performance of the beam management system. However, when the beam prediction accuracy (BAP) of the AI / ML model is too low, it will inevitably affect the performance of the beam management system. It can be considered that this AI / ML model is not applicable. The AI / ML model can be adjusted through the life circle management (LCM) mechanism of the AI / ML model. How to monitor the prediction performance of the AI / ML model and how to adjust the AI / ML model based on the LCM mechanism are issues that need to be solved.

[0003] Summary of the Invention

[0004] An embodiment of the present application provides a method and device for wireless communication, in which a first communication device can monitor the prediction performance of a first network model based on the measurement results corresponding to a first monitoring signal set or the confidence corresponding to a spatial filter predicted in a first prediction data set, thereby reducing or avoiding the measurement overhead for model performance monitoring, thereby improving the performance of model monitoring.

[0005] In a first aspect, a wireless communication method is provided, the method comprising:

[0006] The first communication device inputs a first measurement data set into a first network model and outputs a first prediction data set; wherein the first measurement data set includes at least one of the following: identification information of F spatial filters and link quality information corresponding to the F spatial filters; and the first prediction data set includes at least one of the following: identification information of K spatial filters predicted from the W spatial filters and link quality information corresponding to the K spatial filters predicted from the W spatial filters; wherein F, W, and K are all positive integers, and K<W;

[0007] The first communications device monitors the prediction performance of the first network model according to a measurement result corresponding to a first monitoring signal set; wherein the first monitoring signal set includes M reference signals, each reference signal of the M reference signals satisfies a spatial QCL relationship with a reference signal corresponding to at least one spatial filter of the W spatial filters, and M is a positive integer; or

[0008] The first communication device monitors the prediction performance of the first network model according to the confidence levels corresponding to the K spatial filters output by the first network model.

[0009] In a second aspect, a communication device is provided for executing the method in the first aspect.

[0010] Specifically, the communication device includes a functional module for executing the method in the above-mentioned first aspect.

[0011] In a third aspect, a communication device is provided, comprising a processor and a memory; the memory is used to store a computer program, and the processor is used to call and run the computer program stored in the memory, so that the communication device executes the method in the above-mentioned first aspect.

[0012] In a fourth aspect, a device is provided for implementing the method in the first aspect.

[0013] Specifically, the apparatus includes: a processor, configured to call and run a computer program from a memory, so that a device equipped with the apparatus executes the method in the first aspect described above.

[0014] In a fifth aspect, a computer-readable storage medium is provided for storing a computer program, which enables a computer to execute the method in the first aspect.

[0015] In a sixth aspect, a computer program product is provided, comprising computer program instructions, wherein the computer program instructions enable a computer to execute the method in the first aspect.

[0016] In a seventh aspect, a computer program is provided, which, when executed on a computer, enables the computer to execute the method in the first aspect.

[0017] Through the above technical solution, the first communication device can monitor the prediction performance of the first network model based on the measurement results corresponding to the first monitoring signal set or the confidence corresponding to the spatial filter predicted in the first prediction data set, which can reduce or avoid the measurement overhead for model performance monitoring, thereby improving the performance of model monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] FIG1 is a schematic diagram of a communication system architecture applied in an embodiment of the present application.

[0019] FIG2 is a schematic diagram of the connection of neurons in a neural network provided by the present application.

[0020] FIG3 is a schematic structural diagram of a neural network provided in this application.

[0021] FIG4 is a schematic diagram of a convolutional neural network provided in this application.

[0022] FIG5 is a schematic structural diagram of an LSTM unit provided in this application.

[0023] FIG6 is a schematic diagram of a downlink beam scanning process provided in the present application.

[0024] FIG7 is a schematic diagram of another downlink beam scanning process provided in the present application.

[0025] FIG8 is a schematic diagram of another downlink beam scanning process provided in the present application.

[0026] FIG9 is a schematic diagram of a spatial domain beam prediction model provided in this application.

[0027] FIG10 is a schematic diagram of another spatial domain beam prediction model provided in this application.

[0028] FIG11 is a schematic diagram of a time domain beam prediction model provided in an embodiment of the present application.

[0029] FIG12 is a schematic flowchart of a wireless communication method provided according to an embodiment of the present application.

[0030] FIG13 is a schematic diagram of a QCL type D relationship provided according to an embodiment of the present application.

[0031] Figure 14 is a flowchart of UE-side model monitoring and UE-side decision LCM operation according to an embodiment of the present application.

[0032] FIG15 is a flowchart of a UE-side model monitoring and an NW-side LCM operation decision according to an embodiment of the present application.

[0033] FIG16 is a flowchart of an NW-side model monitoring and NW-side decision LCM operation according to an embodiment of the present application.

[0034] Figure 17 is a schematic block diagram of a communication device provided according to an embodiment of the present application.

[0035] Figure 18 is a schematic block diagram of a communication device provided according to an embodiment of the present application.

[0036] FIG19 is a schematic block diagram of a device provided according to an embodiment of the present application.

[0037] Figure 20 is a schematic block diagram of a communication system provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0038] The following will describe the technical solutions in the embodiments of this application in conjunction with the drawings in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. With respect to the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0039] The technical solutions of the embodiments of the present application can be applied to various communication systems, such as: Global System of Mobile communication (GSM) system, Code Division Multiple Access (CDMA) system, Wideband Code Division Multiple Access (WCDMA) system, General Packet Radio Service (GPRS), Long Term Evolution (LTE) system, Advanced Long Term Evolution (LTE-A) system, New Radio (NR) system, NR system evolution system, LTE-based access to unlicensed spectrum (LTE-U) system on unlicensed spectrum, NR-based access to unlicensed spectrum (NR-U) system on unlicensed spectrum, Non-Terrestrial Networks (NTN) system, Universal Mobile Telecommunication System (UMTS), Wireless Local Area Networks (WLAN), Internet of Things (IoT), Wireless Fidelity (WFI) system. Fidelity, WiFi), fifth-generation communication (5th-Generation, 5G) system, sixth-generation communication (6G) system or other communication systems.

[0040] Generally speaking, traditional communication systems support a limited number of connections and are easy to implement. However, with the development of communication technology, mobile communication systems will not only support traditional communications, but will also support, for example, device-to-device (D2D) communication, machine-to-machine (M2M) communication, machine type communication (MTC), vehicle-to-vehicle (V2V) communication, sidelink (SL) communication, vehicle-to-everything (V2X) communication, etc. The embodiments of the present application can also be applied to these communication systems.

[0041] In some embodiments, the communication system in the embodiments of the present application can be applied to a carrier aggregation (CA) scenario, a dual connectivity (DC) scenario, an independent (SA) networking scenario, or a non-standalone (NSA) networking scenario.

[0042] In some embodiments, the communication system in the embodiments of the present application can be applied to an unlicensed spectrum, where the unlicensed spectrum can also be considered as a shared spectrum; or, the communication system in the embodiments of the present application can also be applied to an authorized spectrum, where the authorized spectrum can also be considered as an unshared spectrum.

[0043] In some embodiments, the communication system in the embodiments of the present application can be applied to the FR1 frequency band (corresponding to the frequency band range of 410MHz to 7.125GHz), can also be applied to the FR2 frequency band (corresponding to the frequency band range of 24.25GHz to 52.6GHz), and can also be applied to new frequency bands such as high-frequency bands corresponding to the frequency band range of 52.6GHz to 71GHz or the frequency band range of 71GHz to 114.25GHz.

[0044] The embodiments of the present application describe various embodiments in conjunction with network devices and terminal devices, wherein the terminal device may also be referred to as user equipment (UE), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent or user device, etc.

[0045] The terminal device can be a station (ST) in a WLAN, a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA) device, a handheld device with wireless communication capabilities, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device, a wearable device, a terminal device in a next-generation communication system such as an NR network, or a terminal device in a future evolved Public Land Mobile Network (PLMN) network, etc.

[0046] In an embodiment of the present application, the terminal device can be deployed on land, including indoors or outdoors, handheld, wearable or vehicle-mounted; it can also be deployed on the water surface (such as ships, etc.); it can also be deployed in the air (such as airplanes, balloons and satellites, etc.).

[0047] In an embodiment of the present application, the terminal device may be a mobile phone, a tablet computer, a computer with wireless transceiver function, a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal device in industrial control, a wireless terminal device in self-driving, a wireless terminal device in remote medical, a wireless terminal device in a smart grid, a wireless terminal device in transportation safety, a wireless terminal device in a smart city or a wireless terminal device in a smart home, an in-vehicle communication device, a wireless communication chip / application specific integrated circuit (ASIC) / system on chip (SoC), etc.

[0048] As an example and not a limitation, in the embodiment of the present application, the terminal device may also be a wearable device. Wearable devices may also be called wearable smart devices, which are a general term for wearable devices that are intelligently designed and developed using wearable technology for daily wear, such as glasses, gloves, watches, clothing, and shoes. A wearable device is a portable device that is worn directly on the body or integrated into the user's clothes or accessories. Wearable devices are not only hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are fully functional, large in size, and can achieve complete or partial functions without relying on smartphones, such as smart watches or smart glasses, as well as those that only focus on a certain type of application function and need to be used in conjunction with other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0049] In an embodiment of the present application, the network device may be a device for communicating with a mobile device. The network device may be an access point (AP) in WLAN, a base station (BTS) in GSM or CDMA, a base station (NodeB, NB) in WCDMA, an evolved base station (eNB or eNodeB) in LTE, or a relay station or access point, or a network device or base station (gNB) or a transmission reception point (TRP) in a vehicle-mounted device, a wearable device, and an NR network, or a network device in a future evolved PLMN network or a network device in an NTN network, etc.

[0050] As an example and not a limitation, in an embodiment of the present application, the network device may have a mobile feature, for example, the network device may be a mobile device. In some embodiments, the network device may be a satellite or a balloon station. For example, the satellite may be a low earth orbit (LEO) satellite, a medium earth orbit (MEO) satellite, a geostationary earth orbit (GEO) satellite, a high elliptical orbit (HEO) satellite, etc. In some embodiments, the network device may also be a base station set up in a location such as land or water.

[0051] In an embodiment of the present application, the network device can provide services for a cell, and the terminal device communicates with the network device through the transmission resources used by the cell (for example, frequency domain resources, or spectrum resources). The cell can be a cell corresponding to the network device (for example, a base station). The cell can belong to a macro base station or a base station corresponding to a small cell. The small cells here may include: metro cells, micro cells, pico cells, femto cells, etc. These small cells have the characteristics of small coverage and low transmission power, and are suitable for providing high-speed data transmission services.

[0052] For example, a communication system 100 used in an embodiment of the present application is shown in FIG1 . The communication system 100 may include a network device 110, which may be a device that communicates with a terminal device 120 (or a communication terminal or terminal). The network device 110 may provide communication coverage for a specific geographic area and may communicate with terminal devices within the coverage area.

[0053] FIG1 exemplarily shows a network device and two terminal devices. In some embodiments, the communication system 100 may include multiple network devices and each network device may include other number of terminal devices within its coverage area, which is not limited in this application.

[0054] In some embodiments, the communication system 100 may further include other network entities such as a network controller and a mobility management entity, which is not limited in the embodiments of the present application.

[0055] It should be understood that in the embodiments of the present application, a device having a communication function in a network / system may be referred to as a communication device. Taking the communication system 100 shown in FIG1 as an example, the communication device may include a network device 110 and a terminal device 120 having a communication function. The network device 110 and the terminal device 120 may be the specific devices described above and will not be described in detail here. The communication device may also include other devices in the communication system 100, such as a network controller, a mobility management entity, and other network entities, which are not limited in the embodiments of the present application.

[0056] It should be understood that the terms "system" and "network" are often used interchangeably herein. The term "and / or" is simply a description of an association between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " generally indicates that the related objects are in an "or" relationship.

[0057] It should be understood that this document relates to a first communication device and a second communication device. The first communication device can be a terminal device, such as a mobile phone, machine facilities, customer premises equipment (CPE), industrial equipment, vehicles, etc.; the second communication device can be a peer communication device of the first communication device, such as a network device, mobile phone, industrial equipment, vehicles, etc. In the embodiments of the present application, the first communication device can be a terminal device, and the second communication device can be a network device (i.e., uplink communication or downlink communication); alternatively, the first communication device can be a first terminal, and the second communication device can be a second terminal (i.e., sideline communication).

[0058] The terms used in the embodiments of this application are intended solely to explain the specific embodiments of this application and are not intended to limit this application. The terms "first," "second," "third," and "fourth," etc., in the specification, claims, and accompanying drawings of this application, are used to distinguish between different objects, not to describe a specific order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions.

[0059] It should be understood that the "indication" mentioned in the embodiments of this application can be a direct indication, an indirect indication, or an indication of an association. For example, "A indicates B" can mean that A directly indicates B, for example, B can be obtained through A; it can also mean that A indirectly indicates B, for example, A indicates C, and B can be obtained through C; it can also mean that there is an association between A and B.

[0060] In the description of the embodiments of the present application, the term "corresponding" may indicate a direct or indirect correspondence between the two, or an association relationship between the two, or a relationship between indication and being indicated, configuration and being configured, etc.

[0061] In the embodiments of the present application, "pre-definition" or "pre-configuration" may be implemented by pre-storing corresponding codes, tables, or other methods that can be used to indicate relevant information in a device (e.g., a terminal device and a network device). The present application does not limit the specific implementation method. For example, pre-definition may refer to information defined in a protocol.

[0062] In the embodiments of the present application, the “protocol” may refer to a standard protocol in the communications field, for example, it may be an evolution of an existing LTE protocol, NR protocol, Wi-Fi protocol, or a protocol related to other communications systems. The present application does not limit the protocol type.

[0063] To facilitate a better understanding of the embodiments of the present application, the neural network and machine learning related to the present application are explained.

[0064] A neural network (NN) is a computational model consisting of multiple interconnected neuron nodes, where the connections between nodes represent weighted values ​​from input signals to output signals, called weights. Each node performs weighted summation (SUM) on different input signals and outputs them through a specific activation function (f). Figure 2 is a schematic diagram of a neuron structure, where a1, a2, …, an represent input signals, w1, w2, …, wn represent weights, f represents the activation function, and t represents the output.

[0065] A simple neural network, shown in Figure 3, consists of an input layer, hidden layers, and an output layer. By using different connections, weights, and activation functions among multiple neurons, different outputs can be generated, thereby fitting the mapping relationship from input to output. Each node in the previous level is connected to all nodes in the next level. This neural network is a fully connected neural network, also known as a deep neural network (DNN).

[0066] The basic structure of a convolutional neural network (CNN) consists of an input layer, multiple convolutional layers, multiple pooling layers, a fully connected layer, and an output layer, as shown in Figure 4. Each neuron in the convolutional kernel of a convolutional layer is locally connected to its input. The introduction of a pooling layer extracts the local maximum or average features of a layer, effectively reducing network parameters and exploiting local features, enabling the CNN to converge quickly and achieve excellent performance.

[0067] Deep learning utilizes deep neural networks with multiple hidden layers, significantly improving the network's ability to learn features and fitting complex, nonlinear mappings from input to output. Consequently, it has found widespread application in speech and image processing. In addition to deep neural networks, deep learning also includes other commonly used basic structures such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) for different tasks.

[0068] The basic structure of a convolutional neural network includes an input layer, multiple convolutional layers, multiple pooling layers, a fully connected layer, and an output layer, as shown in Figure 4. Each neuron in the convolution kernel of the convolutional layer is locally connected to its input, and the introduction of the pooling layer extracts the local maximum or average features of a certain layer, effectively reducing the network parameters and mining local features, enabling the convolutional neural network to converge quickly and achieve excellent performance.

[0069] RNNs are neural networks that model sequential data and have achieved remarkable success in natural language processing applications such as machine translation and speech recognition. Specifically, network devices memorize information from past moments and use it in the calculation of current outputs. This means that nodes in hidden layers are no longer disconnected but connected, and the input to a hidden layer includes not only the input layer but also the output of the previous hidden layer. Common RNN structures include long short-term memory (LSTM) and gated recurrent unit (GRU). Figure 5 shows a basic LSTM cell structure, which can include a tanh activation function. Unlike RNNs, which only consider the most recent state, the LSTM cell state determines which states should be retained and which should be forgotten, addressing the long-term memory limitations of traditional RNNs.

[0070] To facilitate a better understanding of the embodiments of the present application, the NR beam management related to the present application is explained.

[0071] NR systems introduce millimeter-wave frequency band communications and corresponding beam management mechanisms, including both uplink and downlink beam management. Downlink beam management includes downlink beam sweeping, UE beam measurement and reporting, and network (NW) downlink beam indication.

[0072] The downlink beam scanning process may include three processes, namely P1, P2 and P3 processes. The P1 process refers to the network device scanning different transmit beams and the UE scanning different receive beams; the P2 process refers to the network device scanning different transmit beams and the UE using the same receive beam; the P3 process refers to the network device using the same transmit beam and the UE scanning different receive beams. Generally, the network device completes the above beam scanning process by sending a downlink reference signal. Optionally, the downlink reference signal may include but is not limited to a synchronization signal block (SSB) and / or a channel state information reference signal (CSI-RS).

[0073] FIG6 is a schematic diagram of the P1 process (or called the downlink full scan process), FIG7 is a schematic diagram of the P2 process, and FIG8 is a schematic diagram of the P3 process.

[0074] As shown in FIG6 , in the P1 process, the network device traverses all transmit beams to send downlink reference signals, and the UE side traverses all receive beams to perform measurements and determine corresponding measurement results.

[0075] As shown in Figure 7, in the P2 process, the network device traverses all transmit beams to send downlink reference signals, and the UE side uses a specific receive beam to perform measurements to determine the corresponding measurement results.

[0076] As shown in Figure 8, in the P3 process, the network device can use a specific transmit beam to send a downlink reference signal, and the UE side traverses all receive beams to perform measurements and determine the corresponding measurement results.

[0077] Traditional beam reporting in NR means that the UE measures the Layer 1 Reference Signal Receiving Power (L1-RSRP) values ​​of different beams (pairs), selects the K transmit beams with the highest L1-RSRP, and reports them to the NW in the form of uplink control information (UCI). Here, L1-RSRP can also be replaced by other beam link indicators, such as Layer 1 Signal to Interference plus Noise Ratio (L1-SINR) and Layer 1 Reference Signal Received Quality (L1-RSRQ).

[0078] After the network device learns the optimal beam reported by the terminal device, it can carry the Transmission Configuration Indicator (TCI) status (which contains the transmit beam with the downlink reference signal as a reference) through Media Access Control (MAC) or Downlink Control Information (DCI) signaling to complete the beam indication to the UE. The UE uses the receive beam corresponding to the transmit beam for downlink reception.

[0079] To facilitate a better understanding of the embodiments of the present application, the AI / ML-based beam management related to the present application is explained.

[0080] AI / ML-based beam management can predict downlink beams in both spatial and temporal domains.

[0081] Spatial-domain beam prediction (also known as Beam Management Case 1 (BM-Case 1)): The downlink beams in Dataset A (Set A) are predicted in the spatial domain by measuring the beams in Dataset B (Set B). Set B is either a subset of Set A or different sets of beams. Set B can be considered a subset of the beams (pairs); Set A can be considered the full set of beams (pairs).

[0082] Figure 9 schematically shows the input and output relationship of the beam prediction model. It can be considered that the model solves a multi-classification problem, that is, the relationship between the L1-RSRP input of a partial subset (i.e., Set B) and the L1-RSRP of the optimal K beams, where the partial beam measurement set (i.e., Set B, which is part of the L1-RSRP measured by the full set Set A) is used as the input of the model. The output is the optimal K beam indices selected from the full set Set A, that is, the K beams with the highest L1-RSRP. The labels used by the model are the K optimal (i.e., highest L1-RSRP) beam indices measured in the full set Set A. Specifically, as shown in Figure 9, the measurement data set B (Set B) includes the L1-RSRP corresponding to T beam indices, the prediction data set A (Set A) includes S beam indices, and the AI / ML model 1 predicts the optimal K beam indices (beam index #2 in Figure 9). It should be noted that the beam in Figure 9 can also be replaced by a beam pair. The specific description is similar to the beam and will not be repeated here.

[0083] Figure 10 schematically shows the optimal beam quality prediction model, which can be understood as a linear regression problem. The input and output relationship of the model is the relationship from the input L1-RSRP of a partial subset (i.e., Set B) to the L1-RSRP of the optimal K beams. The input part is the same as the beam prediction model in Figure 9, but the difference is that the output of this model is K (K>=1) optimal L1-RSRPs. The label is the optimal K L1-RSRPs measured in the full set (i.e., Set A), and the corresponding K beam indices. Specifically, as shown in Figure 10, the measurement data set B (Set B) includes L1-RSRPs corresponding to T beam indices, the prediction data set A (Set A) includes L1-RSRPs corresponding to S beam indices, and the AI / ML model 2 predicts K (K>=1) optimal L1-RSRPs. It should be noted that the beam in Figure 10 can also be replaced by a beam pair. The specific description is similar to the beam and will not be repeated here.

[0084] Time-domain beam prediction (also known as Beam Management Case 2 (BM-Case 2)): The time-domain prediction of the downlink beams in Dataset A (Set A) is performed using the beams in Historical Measurement Dataset B (Set B). Set B is either a subset of Set A, the same as Set A, or a subset of Set A. Set B can be considered a partial subset of the beams (pairs); Set A can be considered the full set of beams (pairs).

[0085] For the prediction of time-domain beam pairs and their performance, the LSTM model used is shown in Figure 11. This LSTM model can be understood as extending M instances as input in time series, equivalent to a cascade of M LSTM units. Each LSTM unit receives the L1-RSRP of the beam pair for instance m (Set Bm) in dataset B, where 1 ≤ m ≤ M.

[0086] It should be noted that the beam (pair) index of Set Bm can be implicitly input through the fixed ordering of L1-RSRP. After completing the performance input of M instances, the LSTM model can predict the optimal beam (pair) for the next F instances, the performance of the optimal beam (pair) (i.e., link quality information), and the dwelling time of the optimal beam (pair).

[0087] To facilitate a better understanding of the embodiments of the present application, the performance monitoring of the AI / ML model related to the present application is explained.

[0088] Compared with traditional beam measurement, the predictive performance of AI / ML-based beam prediction can be expressed by the beam prediction accuracy (BAP). BAP is defined as the probability, or ratio (based on statistical implementation), that the optimal beam in the actual measurement is included in the top K (top-K, K>=1) beams predicted by the AI / ML model.

[0089] If the AI / ML model's prediction success rate is too low, it will inevitably affect the performance of the beam management system, and the AI / ML model can be considered unsuitable. The AI / ML model's LCM (Life Circle Management) mechanism is used to adjust the AI / ML model.

[0090] In order to facilitate a better understanding of the embodiments of the present application, the problems solved by the present application are explained.

[0091] For the NR beam scanning process, if it is a periodic full-beam scanning process for the downlink, that is, the periodic P1 process (as shown in Figure 6), the UE needs to periodically traverse all combinations of transmit beams and receive beams, which will bring a lot of overhead and delay. For example, assuming that the NW deploys 64 different downlink transmission directions in FR2 (carried by up to 64 SSBs), the UE uses multiple antenna panels (including only one receive beam panel) to perform receive beam scanning simultaneously when receiving, and each antenna panel has 4 receive beams. The UE needs to measure at least 64*4=256 beam pairs, which corresponds to a downlink resource overhead of 256 resources. Therefore, the NR system defines beam prediction in the spatial domain and time domain.

[0092] However, the success rate of beam prediction is affected by many factors. For example, the model is trained in the scenario of cell A, but is not suitable for the beam (pair) configuration in the scenario of cell B. Therefore, when the beam prediction capability of the model drops to a certain level, the LCM mechanism will intervene to complete operations such as model replacement, activation, deactivation, and update, so that the beam prediction performance of the model returns to the level that meets the basic requirements of beam management.

[0093] Specifically, for beam (pair) prediction using a model deployed on the UE side, the UE uses measurements of Set B to predict the optimal K beams (pairs) in Set A. For Tx beam prediction, the model outputs the top-K transmit beams; for Tx-Rx beam pair prediction, the top-K beam pairs output by the model also include the top-K transmit beams. To compare the accuracy of beam (pair) prediction, the NW configures a periodically or dynamically triggered Set A, allowing the UE to measure all beams (pairs) in Set A to identify the K beams (pairs) with the best measured beam (pair) performance. The performance of the beam prediction is determined by comparing the model-predicted top-K beams (pairs) to see if they include the optimal (top-1) measured beam (pair). If so, the beam (pair) is considered accurately predicted; otherwise, it is considered inaccurate. This allows for the generation of a key performance indicator (KPI) for beam prediction accuracy (BPA). However, the most significant disadvantage of this approach is that the measurement overhead of Set A is large, which goes against the original intention of using beam (pair) prediction to reduce beam measurement overhead.

[0094] Based on the above problems, the present application proposes a spatial filter prediction solution, in which the first communication device can monitor the prediction performance of the first network model based on the measurement results corresponding to the first monitoring signal set or the confidence level corresponding to the spatial filter predicted in the first prediction data set, thereby reducing or avoiding the measurement overhead for model performance monitoring, thereby improving the performance of model monitoring.

[0095] It should be noted that the word "beam (pair)" means "beam" or "beam pair". Specifically, in the embodiment of this application, a beam may refer to a transmit beam or a receive beam, and a beam pair refers to a pair of transmit beams and receive beams. In addition, a spatial filter can be used instead of a beam (pair). For AI / ML models, their output can be understood as inference or prediction. In this application, inference and prediction have the same meaning and can be interchanged.

[0096] To facilitate understanding of the technical solutions of the embodiments of the present application, the technical solutions of the present application are described in detail below through specific embodiments. The following related technologies can be combined with the technical solutions of the embodiments of the present application as optional solutions, and they all fall within the scope of protection of the embodiments of the present application. The embodiments of the present application include at least part of the following contents.

[0097] FIG12 is a schematic flowchart of a wireless communication method 200 according to an embodiment of the present application. As shown in FIG12 , the wireless communication method 200 may include at least part of the following contents:

[0098] S210. The first communication device inputs a first measurement data set into a first network model and outputs a first prediction data set; wherein the first measurement data set includes at least one of the following: identification information of F spatial filters and link quality information corresponding to the F spatial filters; and the first prediction data set includes at least one of the following: identification information of K spatial filters predicted from the W spatial filters and link quality information corresponding to the K spatial filters predicted from the W spatial filters; wherein F, W, and K are all positive integers and K<W.

[0099] S220, the first communication device monitors the prediction performance of the first network model according to the measurement results corresponding to the first monitoring signal set; wherein, the first monitoring signal set includes M reference signals, and each reference signal of the M reference signals satisfies the spatial QCL relationship with the reference signal corresponding to at least one spatial filter of the W spatial filters, and M is a positive integer; or, the first communication device monitors the prediction performance of the first network model according to the confidence corresponding to the K spatial filters output by the first network model.

[0100] In an embodiment of the present application, the first communication device can monitor the prediction performance of the first network model based on the measurement results corresponding to the first monitoring signal set or the confidence corresponding to the spatial filter predicted in the first prediction data set, which can reduce or avoid the measurement overhead for model performance monitoring, thereby improving the performance of model monitoring.

[0101] In an embodiment of the present application, the first network model is an AI / ML model. Optionally, the first network model may be an AI / ML model for beam prediction in the spatial domain, as specifically implemented as shown in FIG9 or FIG10 ; or the first network model may be an AI / ML model for beam prediction in the temporal domain, as specifically implemented as shown in FIG11 .

[0102] In some embodiments of the present application, a spatial filter may also be referred to as a beam, a beam pair, a spatial relation, a spatial setting, a spatial domain filter, or a reference signal.

[0103] In some embodiments, the spatial filter includes a transmit spatial filter. Optionally, the transmit spatial filter may also be referred to as a transmit beam (Tx beam) or a transmitting-end spatial domain filter, and the above terms are interchangeable.

[0104] In some embodiments, the spatial filter includes a receive spatial filter. Optionally, the receive spatial filter may also be referred to as a transmit beam (Rx beam) or a receive-end spatial filter, and the above terms are interchangeable.

[0105] In some embodiments, the spatial filter includes a transmit spatial filter and a receive spatial filter. Optionally, the combination of the transmit spatial filter and the receive spatial filter can also be called a beam pair, a spatial filter pair, or a spatial filter bank, and the above terms can be used interchangeably.

[0106] In some embodiments, the identification information of the spatial filter may be an index or an identification of the spatial filter.

[0107] For example, the identification information of the transmit spatial filter may be an index or an identification of the transmit spatial filter.

[0108] For another example, the identification information of the receiving spatial filter may be an index or an identification of the receiving spatial filter.

[0109] For another example, the identification information of the combination of the transmit spatial filter and the receive spatial filter may be a combination index.

[0110] In this embodiment of the present application, the first prediction dataset can be represented by Set A, the first measurement dataset can be represented by Set B, and the first monitoring dataset can be represented by Set C. Set B is either a subset of Set A, or Set B and Set A are two different beam sets. For example, Set B can be understood as a partial subset of the beam (pair), and Set A can be understood as the full set of the beam (pair).

[0111] For example, the first communications device predicts the optimal K beams (pairs) in Set A by measuring Set B. If the prediction is for transmit beams, the model outputs the top-K transmit beams. If the prediction is for transmit and receive beam (Tx-Rx) beam pairs, the model outputs the top-K beam pairs, which include the top-K transmit beams and receive beams.

[0112] In some embodiments, the link quality information may include but is not limited to at least one of the following:

[0113] Layer 1 Reference Signal Receiving Power (L1-RSRP), Layer 1 Reference Signal Receiving Quality (L1-RSRQ), Layer 1 Signal to Interference plus Noise Ratio (L1-SINR).

[0114] In some embodiments, the first communication device is a terminal device, or the first communication device is a network device.

[0115] That is, in an embodiment of the present application, the first network model can be deployed on a terminal device, and the terminal device is responsible for monitoring the predictive performance of the first network model; or, the first network model can be deployed on a network device, and the network device is responsible for monitoring the predictive performance of the first network model.

[0116] In some embodiments, in the above S220, the first communications device monitors the prediction performance of the first network model according to the measurement result corresponding to the first monitoring signal set, including:

[0117] In a case where there is a spatial filter among the K spatial filters that satisfies a quasi-co-located (QCL) relationship with the first spatial filter, the first communications device determines that a prediction result of the first network model is accurate; and / or,

[0118] If there is no spatial filter in the K spatial filters that satisfies the QCL relationship with the first spatial filter, the first communication device determines that the prediction result of the first network model is inaccurate;

[0119] The first spatial filter is an optimal spatial filter included in the measurement result corresponding to the first monitoring signal set, or the first spatial filter is an optimal spatial filter determined based on the measurement result corresponding to the first monitoring signal set.

[0120] In some embodiments, the measurement results corresponding to the first monitoring signal set include at least one of the following: identification information of the spatial filter corresponding to the M reference signals, and link quality information corresponding to the M reference signals. In this case, the first spatial filter is an optimal spatial filter determined based on the measurement results corresponding to the first monitoring signal set.

[0121] In some embodiments, the measurement result corresponding to the first monitoring signal set includes identification information of an optimal spatial filter. In this case, the first spatial filter is the optimal spatial filter included in the measurement result corresponding to the first monitoring signal set.

[0122] In some embodiments, the measurement results corresponding to the first monitoring signal set are obtained by the first communication device through measurement, or the measurement results corresponding to the first monitoring signal set are obtained by the first communication device from other devices.

[0123] For example, the first communication device is a terminal device. The terminal device performs measurement on the first monitoring signal set to obtain a measurement result corresponding to the first monitoring signal set; or the network device performs measurement on the first monitoring signal set to obtain a measurement result corresponding to the first monitoring signal set, and the terminal device obtains the measurement result corresponding to the first monitoring signal set from the network device; or another terminal performs measurement on the first monitoring signal set to obtain a measurement result corresponding to the first monitoring signal set, and the terminal device obtains the measurement result corresponding to the first monitoring signal set from the other terminal; or a third-party device performs measurement on the first monitoring signal set to obtain a measurement result corresponding to the first monitoring signal set, and the terminal device obtains the measurement result corresponding to the first monitoring signal set from the third-party device.

[0124] For example, the first communications device is a network device. The network device performs measurements on a first monitoring signal set to obtain measurement results corresponding to the first monitoring signal set; or the terminal device performs measurements on the first monitoring signal set to obtain measurement results corresponding to the first monitoring signal set, and the network device obtains the measurement results corresponding to the first monitoring signal set from the terminal device; or a third-party device performs measurements on the first monitoring signal set to obtain measurement results corresponding to the first monitoring signal set, and the network device obtains the measurement results corresponding to the first monitoring signal set from the third-party device.

[0125] In some embodiments, the reference signal among the M reference signals is a downlink reference signal.

[0126] In some embodiments, the M reference signals include at least one of the following reference signals: a synchronization signal block (SSB), a channel state information reference signal (CSI-RS).

[0127] Of course, the reference signals in the M reference signals may also be other reference signals, which is not limited in the embodiment of the present application.

[0128] In some embodiments, the spatial filters (i.e., beams) included in the first measurement dataset and the first prediction dataset may be narrow beams from the perspective of beamforming implementation, and the spatial filters (i.e., beams) corresponding to the downlink reference signals in the first monitoring dataset may be wide beams from the perspective of beamforming implementation. That is, each of the M reference signals and the reference signal corresponding to at least one of the W spatial filters satisfy a spatial QCL relationship.

[0129] For example, if the transmit beams included in the first prediction data set (Set A) are 32 narrow beams based on CSI-RS, then the first monitoring data set (Set C) can include 8 SSB wide beams, and each SSB wide beam corresponds to 4 CSI-RS narrow beams. The spatial relationship is shown in Figure 13. In this way, a spatial quasi-co-location relationship is formed between the model monitoring reference signal (RS) (any SSB) in the first monitoring data set (Set C) and the reference signal corresponding to the beam in the first prediction data set (Set A) (multiple CSI-RS), that is, a QCL-Type D relationship. The UE can understand that the SSB and CSI-RS come from roughly the same transmission direction in space and can use the same receive beam for reception. For example, as shown in Figure 13, for the four CSI-RS-based narrow beams in the first prediction data set (Set A), the network device can configure and / or activate a transmission configuration indicator (TCI) state for the four CSI-RS-based narrow beams, where the TCI state includes a wide beam with a QCL-TypeD SSB for the four CSI-RS narrow beams. Next, if the Top-K Tx beams inferred from Set A by the first network model and the optimal Tx beam in Set C are in a QCL relationship, then it can be said that the model prediction is accurate; otherwise, it is considered inaccurate.

[0130] In some embodiments, within a first time window or a first duration, if the number of inaccurate prediction results of the first network model is greater than or equal to a first threshold, the first communication device determines that a first type of model failure has occurred in the first network model.

[0131] That is, within the first time window or the first duration, the accuracy of the prediction results of the first network model is determined multiple times, and the number of inaccurate prediction results of the first network model is accumulated to determine whether the first type of model failure has occurred in the first network model.

[0132] Specifically, after the first communication device determines that the first network model has a first type of model failure, it can trigger the first communication device or other devices to complete model replacement, activation, deactivation, update and other operations based on the LCM mechanism, so that the beam prediction performance of the model returns to the level that meets the basic requirements of beam management.

[0133] Optionally, the first time window may be agreed upon by a protocol, or may be configured or indicated by a network device. Optionally, the first time window may correspond to one or more time units, wherein the time unit may include, but is not limited to, one of the following: a time slot, a mini-time slot, a frame, a subframe, a symbol, a millisecond, or a microsecond.

[0134] Optionally, the first duration may be agreed upon by a protocol, or may be configured or indicated by a network device. Optionally, the first duration may correspond to one or more time units, wherein the time unit may include, but is not limited to, one of the following: a time slot, a mini-time slot, a frame, a subframe, a symbol, a millisecond, or a microsecond.

[0135] Optionally, the first threshold is agreed upon by a protocol, or the first threshold is configured or indicated by a network device.

[0136] In some embodiments, within the second time window or the second duration, if the prediction accuracy of the first network model is lower than a second threshold, the first communications device determines that a second type of model failure has occurred in the first network model.

[0137] That is, within the second time window or the second duration, it is determined multiple times whether the prediction result of the first network model is accurate, and the prediction accuracy of the first network model is calculated, so as to determine whether the first network model has a second type of model failure.

[0138] Specifically, after the first communication device determines that the first network model has a second type of model failure, it can trigger the first communication device or other devices to complete model replacement, activation, deactivation, update and other operations based on the LCM mechanism, so that the beam prediction performance of the model returns to the level that meets the basic requirements of beam management.

[0139] Optionally, the second time window may be agreed upon by a protocol, or may be configured or indicated by a network device. Optionally, the second time window may correspond to one or more time units, wherein the time unit may include, but is not limited to, one of the following: a time slot, a mini-time slot, a frame, a subframe, a symbol, a millisecond, or a microsecond.

[0140] Optionally, the second duration may be agreed upon by a protocol, or may be configured or indicated by a network device. Optionally, the second duration may correspond to one or more time units, wherein the time unit may include, but is not limited to, one of the following: a time slot, a mini-time slot, a frame, a subframe, a symbol, a millisecond, or a microsecond.

[0141] Optionally, the second threshold is agreed upon by a protocol, or the second threshold is configured or indicated by a network device.

[0142] Therefore, in this embodiment, model monitoring based on the first monitoring data set can reduce the measurement overhead for model performance monitoring, thereby improving the performance of model monitoring.

[0143] In some embodiments, in the above S220, the first communication device monitors the prediction performance of the first network model according to the confidence levels corresponding to the K spatial filters output by the first network model, including:

[0144] In a case where the highest confidence level among the confidence levels corresponding to the K spatial filters is equal to or higher than a third threshold, the first communication device determines that the prediction result of the first network model is accurate; and / or,

[0145] When the highest confidence level among the confidence levels corresponding to the K spatial filters is lower than a third threshold, the first communication device determines that the prediction result of the first network model is inaccurate.

[0146] Optionally, the third threshold is agreed upon by a protocol, or the third threshold is configured or indicated by a network device.

[0147] In some embodiments, the K spatial filters included in the first prediction data set are spatial filters corresponding to the first K confidences in the confidence vector output by the first network model, sorted from high to low;

[0148] The length of the confidence vector is W, and each element v of the confidence vector is j Indicates the confidence that the jth spatial filter among the W spatial filters is the optimal spatial filter, 1≤j≤W, v j ∈(0,1).

[0149] It should be noted that v j The closer it is to 1, the greater the possibility that the jth spatial filter is the optimal spatial filter; j The closer it is to 0, the less likely it is that the jth spatial filter is the optimal spatial filter.

[0150] In some embodiments, the output layer of the first network model uses a fully connected layer with W neurons. Optionally, the output layer of the first network model uses Softmax or other activation functions, so that the final output of the first network model is a confidence vector of length W.

[0151] In some embodiments, within a third time window or a third time length, if the number of times the highest confidence level among the confidence levels corresponding to the K spatial filters is lower than the third threshold is greater than or equal to a fourth threshold, the first communication device determines that a third type of model failure has occurred in the first network model.

[0152] Specifically, after the first communication device determines that the first network model has a third type of model failure, it can trigger the first communication device or other devices to complete model replacement, activation, deactivation, update and other operations based on the LCM mechanism, so that the beam prediction performance of the model returns to the level that meets the basic requirements of beam management.

[0153] Optionally, the third time window may be agreed upon by a protocol, or may be configured or indicated by a network device. Optionally, the third time window may correspond to one or more time units, wherein the time unit may include, but is not limited to, one of the following: a time slot, a mini-time slot, a frame, a subframe, a symbol, a millisecond, or a microsecond.

[0154] Optionally, the third duration may be agreed upon by a protocol, or may be configured or indicated by a network device. Optionally, the third duration may correspond to one or more time units, wherein the time unit may include, but is not limited to, one of the following: a time slot, a mini-time slot, a frame, a subframe, a symbol, a millisecond, or a microsecond.

[0155] Optionally, the fourth threshold is agreed upon by a protocol, or the fourth threshold is configured or indicated by a network device.

[0156] Therefore, in this embodiment, confidence-based model monitoring can avoid the measurement overhead for model performance monitoring, thereby improving the performance of model monitoring.

[0157] In some embodiments, the first communication device determines to switch from the first network model to a second network model that implements the same function as the first network model.

[0158] For example, when the first communication device determines that a first type of model failure occurs in the first network model, the first communication device determines to switch from the first network model to a second network model that implements the same function as the first network model.

[0159] For another specific example, when the first communication device determines that a second type of model failure has occurred in the first network model, the first communication device determines to switch from the first network model to a second network model that implements the same function as the first network model.

[0160] For another specific example, when the first communication device determines that the third type of model failure occurs in the first network model, the first communication device determines to switch from the first network model to a second network model that implements the same function as the first network model.

[0161] In some embodiments, the first communication device sends first information;

[0162] The first information includes at least one of the following: a model identifier of the first network model, a model failure type corresponding to the first network model, and a model identifier of the second network model.

[0163] It should be noted that the model failure type corresponding to the first network model may indicate the cause of the model failure event, for example, the first network model made too many inaccurate predictions, or the accuracy of the first network model's predictions was too low, or the confidence level of the first network model's output was too low.

[0164] In some embodiments, the first information is used to request at least one of the following:

[0165] Configuring and / or activating a spatial filter prediction range of the prediction data set corresponding to the second network model;

[0166] Configure and / or activate measurement resources of the measurement data set corresponding to the second network model.

[0167] Optionally, the network device may configure and / or activate at least one of the following based on the request of the terminal device:

[0168] Configuring and / or activating a spatial filter prediction range of the prediction data set corresponding to the second network model;

[0169] Configure and / or activate measurement resources of the measurement data set corresponding to the second network model.

[0170] In some embodiments, when the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the first information is also used to request configuration and / or activation of model monitoring resources of the monitoring signal set corresponding to the second network model.

[0171] Optionally, the network device may configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the second network model based on the request of the terminal device.

[0172] In some embodiments, the first information may be carried via Media Access Control Control Element (MAC CE) signaling, and of course may also be carried via other signaling, which is not limited in the embodiments of the present application.

[0173] In this embodiment, the first communication device is a terminal device. For example, the terminal device sends first information to the network device to request a spatial filter measurement resource adapted to the second network model.

[0174] In some embodiments, the first communication device sends second information; wherein the second information is used to request replacement of a network model that implements the same function as the first network model.

[0175] For example, the terminal device sends the second information to the network device to request to replace a network model that implements the same function as the first network model.

[0176] In some embodiments, the second information includes at least one of the following: a model identifier of the first network model, and a model failure type corresponding to the first network model.

[0177] It should be noted that the model failure type corresponding to the first network model may indicate the cause of the model failure event, for example, the first network model made too many inaccurate predictions, or the accuracy of the first network model's predictions was too low, or the confidence level of the first network model's output was too low.

[0178] For example, when the first communication device determines that a first type of model failure occurs in the first network model, the first communication device sends second information to request replacement of a network model that implements the same function as the first network model.

[0179] For another specific example, when the first communication device determines that the first network model has a second type of model failure, the first communication device sends second information to request a replacement network model that implements the same function as the first network model.

[0180] For another specific example, when the first communication device determines that the third type of model failure has occurred in the first network model, the first communication device sends second information to request to replace a network model that implements the same function as the first network model.

[0181] In some embodiments, the second information may be carried via MAC CE signaling, and of course may also be carried via other signaling, which is not limited in the embodiments of the present application.

[0182] In some embodiments, the first communication device receives third information; wherein the third information is used to instruct deactivation of the first network model and activation of a third network model that implements the same function as the first network model.

[0183] It should be noted that the third information is determined based on the second information, or the third information is a response to the second information.

[0184] For example, after receiving the second information, the network device learns that the terminal device requests to replace a network model with the same function as the first network model. Then, the network device sends a third information to the terminal device to instruct the terminal device to deactivate the first network model and activate the third network model with the same function as the first network model.

[0185] In some embodiments, the third information is used to configure and / or activate at least one of the following:

[0186] The spatial filter prediction range of the prediction data set corresponding to the third network model;

[0187] The measurement resources of the measurement data set corresponding to the third network model.

[0188] In some embodiments, when the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the third information is also used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the third network model.

[0189] In some embodiments, the third information may be carried through MAC CE signaling or downlink control information (DCI), and of course may also be carried through other signaling, which is not limited in the embodiments of the present application.

[0190] In some embodiments, the first communication device sends fourth information, wherein the fourth information is used to request an update of model parameters of the first network model. For example, the terminal device sends the fourth information to the network device to request an update of the model parameters of the first network model.

[0191] For example, when the first communication device determines that a first type of model failure occurs in the first network model, the first communication device sends fourth information to request updating of model parameters of the first network model.

[0192] For another specific example, when the first communication device determines that a second type of model failure has occurred in the first network model, the first communication device sends fourth information to request an update of the model parameters of the first network model.

[0193] For another specific example, when the first communication device determines that the third type of model failure occurs in the first network model, the first communication device sends fourth information to request updating of the model parameters of the first network model.

[0194] In some embodiments, the first communication device receives fifth information;

[0195] The fifth information is used to configure and / or activate at least one of the following:

[0196] updating measurement resources of the prediction data set required for the first network model;

[0197] The measurement resources of the measurement data set required for updating the first network model.

[0198] For example, after receiving the fourth information, the network device learns that the terminal device requests to update the model parameters of the first network model. Then, the network device sends the fifth information to the terminal device to update the measurement resources of the prediction data set required by the first network model and / or update the measurement resources of the measurement data set required by the first network model.

[0199] That is, in this embodiment, the terminal device requests the network device to update the parameters of the old model, that is, through further model training, the model is fine-tuned to enhance the beam (pair) prediction performance of the model.

[0200] In some embodiments, the first communication device implements the functionality implemented by the first network model through other non-network model methods. Specifically, for example, the first communication device falls back to a traditional beam management mechanism that is not AI / ML.

[0201] Optionally, the first communication device may be a terminal device or a network device.

[0202] For example, when the first communication device determines that a first type of model failure occurs in the first network model, the first communication device implements the function implemented by the first network model through other non-network model methods.

[0203] For another specific example, when the first communication device determines that the second type of model failure has occurred in the first network model, the first communication device implements the function implemented by the first network model through other non-network model methods.

[0204] For another specific example, when the first communication device determines that the third type of model failure has occurred in the first network model, the first communication device implements the function implemented by the first network model through other non-network model methods.

[0205] In some embodiments, the first communications device releases measurement resources of the first measurement data set.

[0206] In some embodiments, when the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the first communication device releases the monitoring resources of the first monitoring data set.

[0207] In some embodiments, the first communication device sends model monitoring information; wherein the model monitoring information includes a model identifier of the first network model and a model failure type corresponding to the first network model.

[0208] For example, when the first communication device determines that a first type of model failure occurs in the first network model, the first communication device sends model monitoring information.

[0209] For another specific example, when the first communication device determines that a second type of model failure has occurred in the first network model, the first communication device sends model monitoring information.

[0210] For another specific example, when the first communication device determines that the third type of model failure occurs in the first network model, the first communication device sends model monitoring information.

[0211] It should be noted that the model failure type corresponding to the first network model may indicate the cause of the model failure event, for example, the first network model made too many inaccurate predictions, or the accuracy of the first network model's predictions was too low, or the confidence level of the first network model's output was too low.

[0212] In some embodiments, when the first communication device monitors the prediction performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the model monitoring information also includes the prediction accuracy of the first network model.

[0213] For example, when the prediction accuracy of the first network model in the second time period or the second time window is too low, the terminal device reports the prediction accuracy to the network device, with a value range between 0 and 1.

[0214] In some embodiments, when the first communications device monitors the prediction performance of the first network model based on the confidence scores corresponding to the K spatial filters, the model monitoring information further includes an average value of the maximum values ​​of the confidence scores corresponding to the K spatial filters within a fourth time period. Optionally, the fourth time period is agreed upon by a protocol, or configured or indicated by a network device.

[0215] For example, when the confidence level of the first network model output is too low, the terminal device reports the average value of the highest confidence level of the first network model output within the fourth time period, and the value range is between 0 and 1.

[0216] In some embodiments, the prediction accuracy of the first network model included in the model monitoring information is represented by multiple quantization intervals. For example, four quantization intervals are used, namely: 0-0.25, 0.25-0.5, 0.5-0.75, 0.75-1, which can be quantized into a 2-bit domain, corresponding to 00, 01, 10, and 11 respectively. That is, 00 represents a prediction accuracy in the interval of 0-0.25, 01 represents a prediction accuracy in the interval of 0.25-0.5, 10 represents a prediction accuracy in the interval of 0.5-0.75, and 11 represents a prediction accuracy in the interval of 0.75-1.

[0217] In some embodiments, the average value of the maximum value of the confidence corresponding to the K spatial filters included in the model monitoring information within the fourth time period is represented by multiple quantization intervals. For example, four quantization intervals are used, namely: 0-0.25, 0.25-0.5, 0.5-0.75, 0.75-1, which can be quantized into a 2-bit domain, corresponding to 00, 01, 10, and 11 respectively. That is, 00 represents the average value in the interval of 0-0.25, 01 represents the average value in the interval of 0.25-0.5, 10 represents the average value in the interval of 0.5-0.75, and 11 represents the average value in the interval of 0.75-1.

[0218] In some embodiments, the first communication device receives the sixth information

[0219] The sixth information is used to instruct the first communication device to deactivate the first network model and activate a fourth network model having the same function as that implemented by the first network model.

[0220] For example, the network device receives model monitoring information sent by the terminal device, and then sends sixth information to the terminal device to instruct the terminal device to deactivate the first network model and activate a fourth network model with the same function as the first network model.

[0221] In some embodiments, the sixth information is used to configure and / or activate at least one of the following:

[0222] The spatial filter prediction range of the prediction data set corresponding to the fourth network model;

[0223] The measurement resources of the measurement data set corresponding to the fourth network model.

[0224] In some embodiments, when the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the sixth information is also used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the fourth network model.

[0225] In some embodiments, the first communication device receives seventh information;

[0226] The seventh information is used to indicate resources required to update the model parameters of the first network model.

[0227] For example, the network device receives the model monitoring information sent by the terminal device, and then sends the seventh information to the terminal device to indicate the resources required to update the model parameters of the first network model.

[0228] In some embodiments, the seventh information is used to configure and / or activate at least one of the following:

[0229] updating measurement resources of the prediction data set required for the first network model;

[0230] The measurement resources of the measurement data set required for updating the first network model.

[0231] In some embodiments, the first communication device receives eighth information, wherein the eighth information is used to instruct the first communication device to implement the functionality implemented by the first network model through another non-network model method. Specifically, for example, the eighth information is used to instruct the first communication device to fall back to a traditional beam management mechanism that is not AI / ML.

[0232] For example, the network device receives the model monitoring information sent by the terminal device, and then sends the eighth information to the terminal device to instruct the terminal device to implement the function implemented by the first network model through other non-network models.

[0233] In some embodiments, the eighth information is further used to instruct the first communications device to release measurement resources of the first measurement data set.

[0234] In some embodiments, when the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the eighth information is further used to instruct the first communication device to release the monitoring resources of the first monitoring data set.

[0235] In some embodiments, the first communication device deactivates the first network model and activates a fifth network model that implements the same functionality as the first network model. Optionally, the first communication device is a network device. For example, the network device deactivates the first network model and activates the fifth network model that implements the same functionality as the first network model.

[0236] For example, when the first communication device determines that a first type of model failure occurs in the first network model, the first communication device deactivates the first network model and activates a fifth network model that implements the same function as the first network model.

[0237] For another specific example, when the first communication device determines that the second type of model failure has occurred in the first network model, the first communication device deactivates the first network model and activates a fifth network model that implements the same function as the first network model.

[0238] For another specific example, when the first communication device determines that the third type of model failure occurs in the first network model, the first communication device deactivates the first network model and activates a fifth network model that implements the same function as the first network model.

[0239] In some embodiments, the first communications device configures and / or activates at least one of the following:

[0240] The spatial filter prediction range of the prediction data set corresponding to the fifth network model;

[0241] The measurement resources of the measurement data set corresponding to the fifth network model.

[0242] In some embodiments, when the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the first communication device configures and / or activates the model monitoring resources of the monitoring signal set corresponding to the fifth network model.

[0243] In some embodiments, the first communications device determines resources required to update model parameters of the first network model.

[0244] In some embodiments, the first communications device updates at least one of the following:

[0245] The measurement resources of the prediction data set required by the first network model;

[0246] The measurement resources of the measurement data set required by the first network model.

[0247] Therefore, in an embodiment of the present application, the first communication device can monitor the prediction performance of the first network model based on the measurement results corresponding to the first monitoring signal set or the confidence corresponding to the spatial filter predicted in the first prediction data set, which can reduce or avoid the measurement overhead for model performance monitoring, thereby improving the performance of model monitoring.

[0248] The technical solution of this application is described in detail below through Examples 1 to 3.

[0249] It should be noted that in Examples 1 to 3, the first prediction dataset can be represented by dataset A (Set A), the first measurement dataset can be represented by dataset B (Set B), and the first monitoring dataset can be represented by dataset C (Set C). Set B is either a subset of Set A, or Set B and Set A are two different beam sets. For example, Set B can be understood as a partial subset of the beam (pair); Set A can be understood as the full set of beams (pairs). The model is the first network model described above.

[0250] In embodiment 1, the model is deployed on the UE side, the UE side monitors the prediction performance of the model, and the UE side executes the LCM decision.

[0251] From a process perspective, as shown in Figure 14, this embodiment is based on a UE-side beam (pair) inference model. The UE monitors the model's beam (pair) prediction performance. When the model's beam (pair) prediction performance is poor, the UE performs a responsive model LCM operation.

[0252] For example, the model is deployed on the UE side, and the UE performs beam (pair) prediction based on the model. The UE predicts the optimal K beams (pairs) in Set A by measuring Set B. If the prediction is for Tx beams, the model outputs the top-K transmit beams; if the prediction is for Tx-Rx beam pairs, the top-K beam pairs output by the model also include the top-K transmit beams.

[0253] In embodiment 1, the terminal device may monitor the predicted performance of the model based on Set C.

[0254] Set C can include a set of downlink reference signals, such as SSB and / or CSI-RS. Unlike Set A and Set B, which only include narrow beams, the downlink reference signals in Set C can be wide beams based on the beamforming implementation.

[0255] For example, if Set A contains 32 CSI-RS-based narrow beams for transmit beams, then Set C can contain 8 wide SSB beams, with each wide SSB beam corresponding to 4 narrow CSI-RS beams. The spatial relationship is shown in Figure 13. This creates a quasi-co-location relationship between the reference signal in Set C (any SSB) and the reference signals corresponding to multiple beams (pairs) in Set A (multiple corresponding CSI-RSs), namely a QCL-Type D relationship. The UE can understand that the reference signal in Set C (any SSB) and the reference signals corresponding to multiple beams (pairs) in Set A come from roughly the same spatial transmission direction and can use the same receive beam for reception.

[0256] To monitor the performance of the model, the UE measures Set C, which significantly reduces the overhead of measurements based on Set A. For example, for multiple CSI-RS-based narrow beams in Set A, the NW can configure and / or activate a TCI state that includes a wide beam with multiple CSI-RS-based narrow beam QCL Type D (QCL-Type D) SSBs. Next, if the top-K Tx beams inferred from Set A by the model have a QCL relationship with the optimal Tx beam in Set C, then the model prediction is accurate; otherwise, it is considered inaccurate.

[0257] Specifically, when the Top-K beams (pairs) predicted by the terminal device based on the model do not exist in the same location as the (Top-1) optimal beam (pair) obtained based on the Set C measurement, it can be understood that the result of this beam (pair) prediction is inaccurate; otherwise, the result of this beam (pair) prediction is considered to be accurate.

[0258] In a certain time window, the measurement results of Set C are compared with the inferred results of Set A. When the number of inaccurate beam measurements reaches a certain number, N MFE , the terminal device understands that this is a model failure event (Model Failure Event, MFE), such as the terminal device can determine that the model has a first type of model failure.

[0259] Through multiple measurements of the terminal device, the beam prediction accuracy rate (BAP) can be calculated. If the BAP is lower than a certain threshold BAP within a certain measurement time window, th , the terminal device can determine that a second type of model failure has occurred in the model.

[0260] In embodiment 1, the terminal device may monitor the prediction performance of the model based on the confidence level.

[0261] The specific structure of the model can adopt DNN or other neural network architecture, and this case does not impose specific restrictions. The output layer of the model adopts a fully connected layer containing W neurons (the number of neurons should be the same as the number of beams (pairs) contained in Set A), and adopts Softmax or other activation functions, so that the final output of the model is a confidence vector of length W, denoted as v. Each element v of the confidence vector v is j ∈(0,1), 1≤j≤W represents the confidence that the jth beam (pair) in Set A is the optimal beam pair, v j The closer it is to 1, the greater the possibility that the jth beam (pair) is the optimal beam pair. jThe closer it is to 0, the less likely the j-th beam (pair) is to be the optimal beam pair.

[0262] The model can select the K beam (pair) indices with the highest confidence from the confidence vector v as the final prediction result of the model. When K = 1, the optimal beam (pair) index is output; when K > 1, the optimal multiple beam (pair) indices are output.

[0263] When the highest confidence v among the Top-K beams (pairs) output by the model Top1 Below a predefined threshold v th When the model prediction confidence is lower than a certain threshold more than a certain number of times C within a certain time window, it can be understood that the model does not have enough confidence in the prediction of the optimal beam. MFE Afterwards, the terminal device can determine that a third type of model failure has occurred in the model.

[0264] The most obvious benefit of confidence-based model monitoring is that it does not require additional measurements of Set A or Set C. It does not incur the measurement overhead of model monitoring.

[0265] After a model failure occurs in the model, the terminal device can decide on subsequent LCM operations and send a request related to the model change to the network device, as shown in FIG14 .

[0266] In embodiment 1, the UE side determines the LCM, and the NW side cooperates with the LCM.

[0267] After a model failure event occurs, the UE, as the LCM entity responsible for model monitoring, needs to make a decision on model LCM management. The specific operation may be one of the following options 1 to 4.

[0268] Option 1

[0269] The UE decides to change to a new beam (pair) prediction model and requests beam (pair) measurement resources adapted to the new model from the NW. MAC CE is used as the signaling carrier, including: the model identifier of the old model, the model failure type, and the model identifier of the new model.

[0270] The UE deactivates the old model where the model failure event occurs and activates the new model.

[0271] The UE requests the NW for downlink measurement resources adapted to the new model, including at least one of the following:

[0272] Configure and / or activate the beam(pair) prediction range of Set A for the new model;

[0273] Configure and / or activate the measurement resources of Set B of the new model;

[0274] Configure and / or activate the model monitoring resources of Set C for the new model.

[0275] Option 2

[0276] The UE requests the NW to change a new beam (pair) prediction model. If the NW agrees, the NW will respond to the UE's LCM operation request.

[0277] For example, the UE reports the old model in which a model failure event occurs, using MAC CE as the carrier of the signaling, including the model identifier of the old model and the model failure type.

[0278] For example, the NW deactivates the old model where a model failure event occurs and activates the new model.

[0279] Option 3

[0280] The UE requests the NW to update the parameters of the old model, that is, to fine-tune the model through further model training to enhance the beam (pair) prediction performance of the model.

[0281] UE requests the NW for the resources required for fine-tuning, including

[0282] Configure and / or activate the measurement resources of Set A required for model update. This serves as the full set of beam (pair) predictions. The UE can find the optimal beam (pair) information as the label of the data set by measuring Set A.

[0283] Configure and / or activate the measurement resources of Set B required for model update as a subset of beam (pair) measurements.

[0284] Option 4

[0285] The UE falls back to the traditional non-AI / ML beam management mechanism.

[0286] The UE requests the NW to release the measurement resources of Set B and the model monitoring resources of Set C.

[0287] The UE completes traditional beam reporting through traditional beam (pair) measurement.

[0288] In Example 2, the model is deployed on the UE side, the UE side monitors the prediction performance of the model, and the NW side executes the LCM decision.

[0289] From a process perspective, as shown in Figure 15, this embodiment is based on a UE-side beam (pair) inference model. The UE monitors the model's beam (pair) prediction performance. If the model's beam (pair) prediction performance is poor, the UE reports a model failure event (MFE) to the network network (NW), which then makes decisions regarding the model's LCM.

[0290] For example, the model is deployed on the UE side, and the UE performs beam (pair) prediction based on the model. The UE predicts the optimal K beams (pairs) in Set A by measuring Set B. If the prediction is for Tx beams, the model outputs the top-K transmit beams; if the prediction is for Tx-Rx beam pairs, the top-K beam pairs output by the model also include the top-K transmit beams.

[0291] In embodiment 2, the terminal device may monitor the predicted performance of the model based on Set C.

[0292] Set C can include a set of downlink reference signals, such as SSB and / or CSI-RS. Unlike Set A and Set B, which only include narrow beams, the downlink reference signals in Set C can be wide beams based on the beamforming implementation.

[0293] For example, if Set A contains 32 CSI-RS-based narrow beams for transmit beams, then Set C can contain 8 wide SSB beams, with each wide SSB beam corresponding to 4 narrow CSI-RS beams. The spatial relationship is shown in Figure 13. This creates a quasi-co-location relationship between the reference signal in Set C (any SSB) and the reference signals corresponding to multiple beams (pairs) in Set A (multiple corresponding CSI-RSs), namely a QCL-Type D relationship. The UE can understand that the reference signal in Set C (any SSB) and the reference signals corresponding to multiple beams (pairs) in Set A come from roughly the same spatial transmission direction and can use the same receive beam for reception.

[0294] To monitor the performance of the model, the UE measures Set C, which significantly reduces the overhead of measurements based on Set A. For example, for multiple CSI-RS-based narrow beams in Set A, the NW can configure and / or activate a TCI state that includes a wide beam with multiple CSI-RS-based narrow beam QCL Type D (QCL-Type D) SSBs. Next, if the top-K Tx beams inferred from Set A by the model have a QCL relationship with the optimal Tx beam in Set C, then the model prediction is accurate; otherwise, it is considered inaccurate.

[0295] Specifically, when the Top-K beams (pairs) predicted by the terminal device based on the model do not exist in the same location as the (Top-1) optimal beam (pair) obtained based on the Set C measurement, it can be understood that the result of this beam (pair) prediction is inaccurate; otherwise, the result of this beam (pair) prediction is considered to be accurate.

[0296] In a certain time window, the measurement results of Set C are compared with the inferred results of Set A. When the number of inaccurate beam measurements reaches a certain number, N MFE , the terminal device understands that this is a model failure event (Model Failure Event, MFE), such as the terminal device can determine that the model has a first type of model failure.

[0297] Through multiple measurements of the terminal device, the beam prediction accuracy rate (BAP) can be calculated. If the BAP is lower than a certain threshold BAP within a certain measurement time window, th , the terminal device can determine that a second type of model failure has occurred in the model.

[0298] In embodiment 2, the terminal device may monitor the prediction performance of the model based on the confidence level.

[0299] The specific structure of the model can adopt DNN or other neural network architecture, and this case does not impose specific restrictions. The output layer of the model adopts a fully connected layer containing W neurons (the number of neurons should be the same as the number of beams (pairs) contained in Set A), and adopts Softmax or other activation functions, so that the final output of the model is a confidence vector of length W, denoted as v. Each element v of the confidence vector v is j ∈(0,1), 1≤j≤W represents the confidence that the jth beam (pair) in Set A is the optimal beam pair, v j The closer it is to 1, the greater the possibility that the jth beam (pair) is the optimal beam pair. j The closer it is to 0, the less likely the j-th beam (pair) is to be the optimal beam pair.

[0300] The model can select the K beam (pair) indices with the highest confidence from the confidence vector v as the final prediction result of the model. When K = 1, the optimal beam (pair) index is output; when K > 1, the optimal multiple beam (pair) indices are output.

[0301] When the highest confidence v among the Top-K beams (pairs) output by the model Top1 Below a predefined threshold v th When the model prediction confidence is lower than a certain threshold more than a certain number of times C within a certain time window, it can be understood that the model does not have enough confidence in the prediction of the optimal beam. MFE Afterwards, the terminal device can determine that a third type of model failure has occurred in the model.

[0302] The most obvious benefit of confidence-based model monitoring is that it does not require additional measurements of Set A or Set C. It does not incur the measurement overhead of model monitoring.

[0303] After a model failure occurs in the model, the terminal device reports the model monitoring information, and the network device makes a decision on the model LCM based on the model monitoring information, as shown in FIG15 .

[0304] Regardless of whether it is measurement-based model monitoring or confidence-based model monitoring, when the Model Failure Event (MFE) condition is met, the UE reports model monitoring information to the NW, which includes but is not limited to the following information:

[0305] The model has a common identity (ID) between the NW and the UE, namely the model ID.

[0306] The model failure type, i.e., type 1 model failure / type 2 model failure / type 3 model failure, indicates the cause of the model failure event, such as low accuracy of model prediction or low confidence in model output.

[0307] When the BAP of the model is too low within a period of time, the UE reports the BAP value to the NW, which ranges from 0 to 1.

[0308] When the confidence level of the model output is too low, the UE reports the average value of the highest confidence level of the model over a period of time, which also ranges from 0 to 1.

[0309] For the above values ​​between 0 and 1, quantization can be used. For example, quantization can be divided into four intervals: 0-0.25, 0.25-0.5, 0.5-0.75, and 0.75-1, which can be quantized into a 2-bit domain, corresponding to 00, 01, 10, and 11 respectively.

[0310] In Example 2, NW determines LCM.

[0311] After receiving the model monitoring information reported by the UE, the NW makes a decision related to LCM. The specific operation may be one of the following options 5 to 7.

[0312] Option 5

[0313] The NW deactivates the old model for the UE and activates a new model for beam (pair) prediction. Optionally, a MAC CE is used as a signaling carrier, which includes the model ID of the old model and the model ID of the new model.

[0314] The NW configures and / or activates measurement resources adapted to the new model, including at least one of the following:

[0315] The beam (pair) prediction range of Set A of the new model;

[0316] The measurement resources of Set B of the new model;

[0317] Model monitoring resources for Set C of the new model.

[0318] Option 6

[0319] The NW updates the parameters of the model for the UE, that is, through further model training, fine-tuning the model to enhance the beam (pair) prediction performance of the model.

[0320] The NW indicates to the UE the resources required for fine-tuning, including

[0321] Configure and / or activate the measurement resources of Set A required for model update. This serves as the full set of beam (pair) predictions. The UE can find the optimal beam (pair) information as the label of the data set by measuring Set A.

[0322] Configure and / or activate the measurement resources of Set B required for model update as a subset of beam (pair) measurements.

[0323] Option 7

[0324] The NW notifies the UE to fall back to the traditional beam management mechanism without AI / ML.

[0325] Notify the UE to release the measurement resources of Set B (model input) and Set C (model monitoring).

[0326] The UE completes traditional beam reporting through traditional beam (pair) measurement.

[0327] In Example 3, the model is deployed on the NW side, the NW side monitors the prediction performance of the model, and the NW side executes the LCM decision.

[0328] From a process perspective, as shown in Figure 16, this embodiment is based on the NW-side beam (pair) inference model. The NW monitors the model's beam (pair) prediction performance. If the model's beam (pair) prediction performance is poor, the NW makes a decision regarding the model's LCM.

[0329] For example, the model is deployed on the network network (NW), and the NW predicts beams (pairs) based on the model. The UE measures Set B and reports all measurement results of Set B, including the Tx beam ID and its corresponding L1-RSRP, to the NW. The NW uses the UE's reported Set B measurement results as input to the model, performing inference and selecting the optimal top-K beams (pairs) from Set A.

[0330] In embodiment 2, the network device may monitor the predicted performance of the model based on Set C.

[0331] Set C can include a set of downlink reference signals, such as SSB and / or CSI-RS. Unlike Set A and Set B, which only include narrow beams, the downlink reference signals in Set C can be wide beams based on the beamforming implementation.

[0332] For model monitoring, the UE measures the wide beams in Set C, significantly reducing the overhead of measurements based on Set A. Because the model is not on the UE, the UE cannot perform comparisons. The UE reports the measurement results of the model monitoring reference signal set Set C, including the Tx beam ID and / or corresponding L1-RSRP, to the NW. Traditional beam reporting can be used to report the optimal beam in Set C. Alternatively, traditional beam reporting can be simplified to only report the ID of the optimal Tx beam without reporting its L1-RSRP.

[0333] For example, if Set A contains 32 CSI-RS-based narrow beams for transmit beams, then Set C can contain 8 wide SSB beams, with each wide SSB beam corresponding to 4 narrow CSI-RS beams. The spatial relationship is shown in Figure 13. This creates a quasi-co-location relationship between the reference signal in Set C (any SSB) and the reference signals corresponding to multiple beams (pairs) in Set A (multiple corresponding CSI-RSs), namely a QCL-Type D relationship. The UE can understand that the reference signal in Set C (any SSB) and the reference signals corresponding to multiple beams (pairs) in Set A come from roughly the same spatial transmission direction and can use the same receive beam for reception.

[0334] To monitor the performance of the model, the UE measures Set C. For example, for multiple CSI-RS-based narrow beams in Set A, the NW can configure and / or activate a TCI state that includes a wide beam with multiple CSI-RS-based narrow beams and QCL Type D SSBs. If the model infers that the top-K Tx beams from Set A are in a QCL relationship with the optimal Tx beam in Set C, the model prediction is accurate; otherwise, it is considered inaccurate.

[0335] Specifically, when NW receives the wide beam measurement results in Set C, runs the beam (pair) prediction model, and infers the optimal Top-K beams (pairs) in Set A, if the Top-K beams (pairs) predicted by NW based on the model do not exist in the same location as the (Top-1) optimal beam (pair) obtained based on the Set C measurement, it can be understood that the result of this beam (pair) prediction is inaccurate; otherwise, the result of this beam (pair) prediction is considered accurate.

[0336] In a certain time window, the measurement results of Set C are compared with the inferred results of Set A. When the number of inaccurate beam measurements reaches a certain number, N MFE , the terminal device understands that this is a model failure event (Model Failure Event, MFE), such as the network device can determine that the model has a first type of model failure.

[0337] Through multiple measurements of network equipment, the beam prediction accuracy rate (BAP) can be calculated. If the BAP is lower than a certain threshold BAP within a certain measurement time window, th , the network device can determine that the model has experienced a second type of model failure.

[0338] In embodiment 2, the network device may monitor the prediction performance of the model based on the confidence level.

[0339] The specific structure of the model can adopt DNN or other neural network architecture, and this case does not impose specific restrictions. The output layer of the model adopts a fully connected layer containing W neurons (the number of neurons should be the same as the number of beams (pairs) contained in Set A), and adopts Softmax or other activation functions, so that the final output of the model is a confidence vector of length W, denoted as v. Each element v of the confidence vector v is j ∈(0,1), 1≤j≤W represents the confidence that the jth beam (pair) in Set A is the optimal beam pair, v j The closer it is to 1, the greater the possibility that the jth beam (pair) is the optimal beam pair. jThe closer it is to 0, the less likely the j-th beam (pair) is to be the optimal beam pair.

[0340] The model can select the K beam (pair) indices with the highest confidence from the confidence vector v as the final prediction result of the model. When K = 1, the optimal beam (pair) index is output; when K > 1, the optimal multiple beam (pair) indices are output.

[0341] When the highest confidence v among the Top-K beams (pairs) output by the model Top1 Below a predefined threshold v th When the model prediction confidence is lower than a certain threshold more than a certain number of times C within a certain time window, it can be understood that the model does not have enough confidence in the prediction of the optimal beam. MFE Afterwards, the network device can determine that a third type of model failure has occurred in the model.

[0342] The most obvious benefit of confidence-based model monitoring is that it does not require additional measurements of Set A or Set C. It does not incur the measurement overhead of model monitoring.

[0343] After a model failure occurs in the model, the network device makes a decision regarding the model LCM, as shown in FIG16 .

[0344] In Example 3, NW determines LCM.

[0345] NW makes a decision on LCM, and the specific operation may be one of the following options 8 to 10.

[0346] Option 8

[0347] The NW deactivates the old model for the UE and activates a new model for beam (pair) prediction. Optionally, a MAC CE is used as a signaling carrier, which includes the model ID of the old model and the model ID of the new model.

[0348] The NW configures and / or activates measurement resources adapted to the new model, including at least one of the following:

[0349] The beam (pair) prediction range of Set A of the new model;

[0350] The measurement resources of Set B of the new model;

[0351] Model monitoring resources for Set C of the new model.

[0352] Option 9

[0353] The NW updates the parameters of the model for the UE, that is, through further model training, fine-tuning the model to enhance the beam (pair) prediction performance of the model.

[0354] The NW indicates to the UE the resources required for fine-tuning, including

[0355] Configure and / or activate the measurement resources of Set A required for model update. This serves as the full set of beam (pair) predictions. The UE can find the optimal beam (pair) information as the label of the data set by measuring Set A.

[0356] Configure and / or activate the measurement resources of Set B required for model update as a subset of beam (pair) measurements.

[0357] Option 10

[0358] The NW notifies the UE to fall back to the traditional beam management mechanism without AI / ML.

[0359] Notify the UE to release the measurement resources of Set B (model input) and Set C (model monitoring).

[0360] The UE completes traditional beam reporting through traditional beam (pair) measurement.

[0361] The above text, in combination with Figures 12 to 16, describes in detail the method embodiment of the present application. The following text, in combination with Figures 17 to 20, describes in detail the device embodiment of the present application. It should be understood that the device embodiment and the method embodiment correspond to each other, and similar descriptions can refer to the method embodiment.

[0362] Figure 17 shows a schematic block diagram of a communication device 300 according to an embodiment of the present application. The communication device 300 is a first communication device. As shown in Figure 17, the communication device 300 includes: a processing unit 310;

[0363] The processing unit 310 is configured to input a first measurement data set into a first network model and output a first prediction data set; wherein the first measurement data set includes at least one of the following: identification information of F spatial filters and link quality information corresponding to the F spatial filters; and the first prediction data set includes at least one of the following: identification information of K spatial filters predicted from the W spatial filters and link quality information corresponding to the K spatial filters predicted from the W spatial filters; wherein F, W, and K are all positive integers, and K<W;

[0364] The processing unit 310 is further configured to monitor the prediction performance of the first network model based on the measurement results corresponding to the first monitoring signal set; wherein the first monitoring signal set includes M reference signals, each reference signal of the M reference signals satisfies a spatial quasi-co-location (QCL) relationship with a reference signal corresponding to at least one of the W spatial filters, and M is a positive integer; or

[0365] The processing unit 310 is further configured to monitor the prediction performance of the first network model according to the confidence levels corresponding to the K spatial filters output by the first network model.

[0366] In some embodiments, the processing unit 310 is specifically configured to:

[0367] If there is a spatial filter in the K spatial filters that satisfies the QCL relationship with the first spatial filter, determining that the prediction result of the first network model is accurate; and / or,

[0368] If there is no spatial filter in the K spatial filters that satisfies the QCL relationship with the first spatial filter, determining that the prediction result of the first network model is inaccurate;

[0369] The first spatial filter is an optimal spatial filter included in the measurement result corresponding to the first monitoring signal set, or the first spatial filter is an optimal spatial filter determined based on the measurement result corresponding to the first monitoring signal set.

[0370] In some embodiments, the measurement result corresponding to the first monitoring signal set includes at least one of the following: identification information of the spatial filter corresponding to the M reference signals, link quality information corresponding to the M reference signals; or

[0371] The measurement result corresponding to the first monitoring signal set includes identification information of the optimal spatial filter.

[0372] In some embodiments, the measurement results corresponding to the first monitoring signal set are obtained by the first communication device through measurement, or the measurement results corresponding to the first monitoring signal set are obtained by the first communication device from other devices.

[0373] In some embodiments, the M reference signals include at least one of the following reference signals:

[0374] Synchronization signal block SSB, channel state information reference signal CSI-RS.

[0375] In some embodiments, within a first time window or a first duration, if the number of inaccurate prediction results of the first network model is greater than or equal to a first threshold, the processing unit 310 is further configured to determine that a first type of model failure has occurred in the first network model; and / or,

[0376] Within the second time window or the second duration, if the prediction accuracy of the first network model is lower than a second threshold, the processing unit 310 is further configured to determine that a second type of model failure has occurred in the first network model.

[0377] In some embodiments, the processing unit 310 is specifically configured to:

[0378] When the highest confidence level among the confidence levels corresponding to the K spatial filters is equal to or higher than a third threshold, determining that the prediction result of the first network model is accurate; and / or,

[0379] When the highest confidence level among the confidence levels corresponding to the K spatial filters is lower than a third threshold, it is determined that the prediction result of the first network model is inaccurate.

[0380] In some embodiments, the K spatial filters included in the first prediction data set are spatial filters corresponding to the first K confidences in the confidence vector output by the first network model, sorted from high to low;

[0381] The length of the confidence vector is W, and each element v of the confidence vector is j Indicates the confidence that the jth spatial filter among the W spatial filters is the optimal spatial filter, 1≤j≤W, v j ∈(0,1).

[0382] In some embodiments, the output layer of the first network model adopts a fully connected layer of W neurons.

[0383] In some embodiments, within a third time window or a third time length, if the number of times the highest confidence level among the confidence levels corresponding to the K spatial filters is lower than the third threshold is greater than or equal to a fourth threshold, the processing unit 310 is also used to determine that a third type of model failure has occurred in the first network model.

[0384] In some embodiments, the processing unit 310 is further configured to determine whether to switch from the first network model to a second network model that implements the same function as the first network model.

[0385] In some embodiments, the communication device 300 further includes:

[0386] The communication unit 320 is configured to send the first information;

[0387] The first information includes at least one of the following: a model identifier of the first network model, a model failure type corresponding to the first network model, and a model identifier of the second network model.

[0388] In some embodiments, the first information is used to request at least one of the following:

[0389] Configuring and / or activating a spatial filter prediction range of the prediction data set corresponding to the second network model;

[0390] Configure and / or activate measurement resources of the measurement data set corresponding to the second network model.

[0391] In some embodiments, when the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the first information is also used to request configuration and / or activation of model monitoring resources of the monitoring signal set corresponding to the second network model.

[0392] In some embodiments, the communication device 300 further includes:

[0393] The communication unit 320 is configured to send second information;

[0394] The second information is used to request replacement of a network model that implements the same function as the first network model.

[0395] In some embodiments, the second information includes at least one of the following: a model identifier of the first network model, and a model failure type corresponding to the first network model.

[0396] In some embodiments, the communication unit 320 is further configured to receive third information;

[0397] The third information is used to instruct deactivation of the first network model and activation of a third network model having the same function as that implemented by the first network model.

[0398] In some embodiments, the third information is used to configure and / or activate at least one of the following:

[0399] The spatial filter prediction range of the prediction data set corresponding to the third network model;

[0400] The measurement resources of the measurement data set corresponding to the third network model.

[0401] In some embodiments, when the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the third information is also used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the third network model.

[0402] In some embodiments, the communication device 300 further includes:

[0403] The communication unit 320 is configured to send fourth information;

[0404] The fourth information is used to request updating of the model parameters of the first network model.

[0405] In some embodiments, the communication unit 320 is further configured to receive fifth information;

[0406] The fifth information is used to configure and / or activate at least one of the following:

[0407] updating measurement resources of the prediction data set required for the first network model;

[0408] The measurement resources of the measurement data set required for updating the first network model.

[0409] In some embodiments, the processing unit 310 is further configured to implement the functions implemented by the first network model through other non-network model methods.

[0410] In some embodiments, the processing unit 310 is further configured to release measurement resources of the first measurement data set.

[0411] In some embodiments, when the first communication device monitors the prediction performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the processing unit 310 is further configured to release the monitoring resources of the first monitoring data set.

[0412] In some embodiments, the communication device 300 further includes:

[0413] The communication unit 320 is configured to send model monitoring information, wherein the model monitoring information includes a model identifier of the first network model and a model failure type corresponding to the first network model.

[0414] In some embodiments, when the first communications device monitors the prediction performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the model monitoring information further includes the prediction accuracy of the first network model; or

[0415] When the first communication device monitors the prediction performance of the first network model based on the confidence levels corresponding to the K spatial filters, the model monitoring information also includes an average value of the maximum values ​​of the confidence levels corresponding to the K spatial filters within a fourth time period.

[0416] In some embodiments, the prediction accuracy of the first network model included in the model monitoring information is represented by multiple quantization intervals; or,

[0417] The average value of the maximum value of the confidence levels corresponding to the K spatial filters included in the model monitoring information within the fourth time duration is represented by multiple quantization intervals.

[0418] In some embodiments, the communication unit 320 is further configured to receive a sixth information

[0419] The sixth information is used to instruct the first communication device to deactivate the first network model and activate a fourth network model having the same function as that implemented by the first network model.

[0420] In some embodiments, the sixth information is used to configure and / or activate at least one of the following:

[0421] The spatial filter prediction range of the prediction data set corresponding to the fourth network model;

[0422] The measurement resources of the measurement data set corresponding to the fourth network model.

[0423] In some embodiments, when the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the sixth information is also used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the fourth network model.

[0424] In some embodiments, the communication unit 320 is further configured to receive seventh information;

[0425] The seventh information is used to indicate resources required to update the model parameters of the first network model.

[0426] In some embodiments, the seventh information is used to configure and / or activate at least one of the following:

[0427] updating measurement resources of the prediction data set required for the first network model;

[0428] The measurement resources of the measurement data set required for updating the first network model.

[0429] In some embodiments, the communication unit 320 is further used to receive eighth information; wherein, the eighth information is used to instruct the first communication device to implement the function implemented by the first network model through other non-network model methods.

[0430] In some embodiments, the eighth information is further used to instruct the first communications device to release measurement resources of the first measurement data set.

[0431] In some embodiments, when the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the eighth information is further used to instruct the first communication device to release the monitoring resources of the first monitoring data set.

[0432] In some embodiments, the first communication device is a terminal device.

[0433] In some embodiments, the processing unit 310 is further configured to deactivate the first network model and activate a fifth network model that implements the same function as the first network model.

[0434] In some embodiments, the processing unit 310 is further configured to configure and / or activate at least one of the following:

[0435] The spatial filter prediction range of the prediction data set corresponding to the fifth network model;

[0436] The measurement resources of the measurement data set corresponding to the fifth network model.

[0437] In some embodiments, when the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the processing unit 310 is also used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the fifth network model.

[0438] In some embodiments, the processing unit 310 is further configured to determine resources required to update model parameters of the first network model.

[0439] In some embodiments, the processing unit 310 is further configured to update at least one of the following:

[0440] The measurement resources of the prediction data set required by the first network model;

[0441] The measurement resources of the measurement data set required by the first network model.

[0442] In some embodiments, the first communication device is a network device.

[0443] In some embodiments, the spatial filter comprises a transmit spatial filter; or,

[0444] The spatial filter includes a transmitting spatial filter and a receiving spatial filter.

[0445] In some embodiments, the communication unit may be a communication interface or a transceiver, or an input / output interface of a communication chip or a system on chip. The processing unit may be one or more processors.

[0446] It should be understood that the communication device 300 according to the embodiment of the present application may correspond to the first communication device in the method embodiment of the present application, and the above-mentioned and other operations and / or functions of each unit in the communication device 300 are respectively for implementing the corresponding processes of the first communication device in the method 200 shown in Figure 12. For the sake of brevity, they will not be repeated here.

[0447] Figure 18 is a schematic structural diagram of a communication device 400 provided in an embodiment of the present application. The communication device 400 shown in Figure 18 includes a processor 410, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.

[0448] In some embodiments, as shown in FIG18 , the communication device 400 may further include a memory 420. The processor 410 may call and execute a computer program from the memory 420 to implement the method in the embodiment of the present application.

[0449] The memory 420 may be a separate device independent of the processor 410 , or may be integrated into the processor 410 .

[0450] In some embodiments, as shown in FIG18 , the communication device 400 may further include a transceiver 430 , and the processor 410 may control the transceiver 430 to communicate with other devices. Specifically, the transceiver 430 may send information or data to other devices, or receive information or data sent by other devices.

[0451] The transceiver 430 may include a transmitter and a receiver. The transceiver 430 may further include an antenna, and the number of antennas may be one or more.

[0452] In some embodiments, the processor 410 may implement the functionality of a processing unit in the first communication device, which will not be described in detail here for the sake of brevity.

[0453] In some embodiments, the transceiver 430 may implement the function of the communication unit in the first communication device, which will not be described in detail here for the sake of brevity.

[0454] In some embodiments, the communication device 400 may specifically be the first communication device of the embodiment of the present application, and the communication device 400 may implement the corresponding processes implemented by the first communication device in each method of the embodiment of the present application. For the sake of brevity, they will not be repeated here.

[0455] Figure 19 is a schematic structural diagram of an apparatus according to an embodiment of the present application. The apparatus 500 shown in Figure 19 includes a processor 510, which can call and execute a computer program from a memory to implement the method according to the embodiment of the present application.

[0456] In some embodiments, as shown in FIG19 , the apparatus 500 may further include a memory 520. The processor 510 may call and execute a computer program from the memory 520 to implement the method in the embodiment of the present application.

[0457] The memory 520 may be a separate device independent of the processor 510 , or may be integrated into the processor 510 .

[0458] In some embodiments, the processor 510 may implement the functionality of a processing unit in the first communication device, which will not be described in detail here for the sake of brevity.

[0459] In some embodiments, the apparatus 500 may further include an input interface 530. The processor 510 may control the input interface 530 to communicate with other devices or chips, and specifically, may obtain information or data sent by other devices or chips. Optionally, the processor 510 may be located inside or outside the chip.

[0460] In some embodiments, the input interface 530 may implement the functionality of a communication unit in the first communication device.

[0461] In some embodiments, the apparatus 500 may further include an output interface 540. The processor 510 may control the output interface 540 to communicate with other devices or chips, specifically, to output information or data to other devices or chips. Optionally, the processor 510 may be located inside or outside the chip.

[0462] In some embodiments, the output interface 540 may implement the functionality of a communication unit in the first communication device.

[0463] In some embodiments, the apparatus can be applied to the first communication device in the embodiments of the present application, and the apparatus can implement the corresponding processes implemented by the first communication device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0464] In some embodiments, the device mentioned in the embodiments of the present application may also be a chip, such as a system-on-chip, a system-on-chip, a chip system, or a system-on-chip chip.

[0465] FIG20 is a schematic block diagram of a communication system 600 provided in an embodiment of the present application. As shown in FIG20 , the communication system 600 includes a first communication device 610 and a second communication device 620 .

[0466] The first communication device 610 may be used to implement the corresponding functions implemented by the first communication device in the above method, which will not be described in detail here for the sake of brevity.

[0467] It should be understood that the processor of the embodiments of the present application may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiment can be completed by hardware integrated logic circuits in the processor or software instructions. The above processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The various methods, steps, and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of the present application can be directly embodied as being executed by a hardware decoding processor, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0468] It is understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAM bus random access memory (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0469] It should be understood that the above-mentioned memories are exemplary but not restrictive. For example, the memories in the embodiments of the present application may also be static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAM RAM (DR RAM), etc. In other words, the memories in the embodiments of the present application are intended to include, but are not limited to, these and any other suitable types of memories.

[0470] An embodiment of the present application also provides a computer-readable storage medium for storing a computer program.

[0471] In some embodiments, the computer-readable storage medium can be applied to the first communication device in the embodiments of the present application, and the computer program enables the computer to execute the corresponding processes implemented by the first communication device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0472] An embodiment of the present application also provides a computer program product, including computer program instructions.

[0473] In some embodiments, the computer program product can be applied to the first communication device in the embodiments of the present application, and the computer program instructions enable the computer to execute the corresponding processes implemented by the first communication device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0474] The embodiment of the present application also provides a computer program.

[0475] In some embodiments, the computer program can be applied to the first communication device in the embodiments of the present application. When the computer program runs on a computer, the computer executes the corresponding processes implemented by the first communication device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0476] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0477] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0478] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0479] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of the solution of this embodiment according to actual needs.

[0480] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0481] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. In view of this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0482] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A wireless communication method, characterized in that: include: The first communication device inputs a first measurement data set into a first network model and outputs a first prediction data set; wherein the first measurement data set includes at least one of the following: identification information of F spatial filters and link quality information corresponding to the F spatial filters; and the first prediction data set includes at least one of the following: identification information of K spatial filters predicted from the W spatial filters and link quality information corresponding to the K spatial filters predicted from the W spatial filters; wherein F, W, and K are all positive integers, and K<W; The first communications device monitors the prediction performance of the first network model according to a measurement result corresponding to a first monitoring signal set; wherein the first monitoring signal set includes M reference signals, each of the M reference signals satisfies a spatial quasi-co-location (QCL) relationship with a reference signal corresponding to at least one of the W spatial filters, and M is a positive integer; or The first communication device monitors the prediction performance of the first network model according to the confidence levels corresponding to the K spatial filters output by the first network model.

2. The method according to claim 1, wherein The first communication device monitors the prediction performance of the first network model according to the measurement result corresponding to the first monitoring signal set, including: In a case where there is a spatial filter among the K spatial filters that satisfies a QCL relationship with the first spatial filter, the first communication device determines that the prediction result of the first network model is accurate; and / or, If there is no spatial filter among the K spatial filters that satisfies the QCL relationship with the first spatial filter, the first communication device determines that the prediction result of the first network model is inaccurate; The first spatial filter is an optimal spatial filter included in the measurement result corresponding to the first monitoring signal set, or the first spatial filter is an optimal spatial filter determined based on the measurement result corresponding to the first monitoring signal set.

3. The method according to claim 2, wherein The measurement result corresponding to the first monitoring signal set includes at least one of the following: identification information of the spatial filter corresponding to the M reference signals, link quality information corresponding to the M reference signals; or The measurement result corresponding to the first monitoring signal set includes identification information of the optimal spatial filter.

4. The method according to claim 2 or 3, wherein: The measurement result corresponding to the first monitoring signal set is obtained by measuring by the first communication device, or the measurement result corresponding to the first monitoring signal set is obtained by the first communication device from other devices.

5. The method according to any one of claims 2 to 4, characterized in that The M reference signals include at least one of the following reference signals: Synchronization signal block SSB, channel state information reference signal CSI-RS.

6. The method according to any one of claims 2 to 5, characterized in that The method further comprises: Within a first time window or a first duration, if the number of inaccurate prediction results of the first network model is greater than or equal to a first threshold, the first communications device determines that a first type of model failure has occurred in the first network model; and / or, Within the second time window or the second duration, if the prediction accuracy of the first network model is lower than a second threshold, the first communications device determines that a second type of model failure has occurred in the first network model.

7. The method according to claim 1, wherein The first communication device monitors the prediction performance of the first network model according to the confidence levels corresponding to the K spatial filters output by the first network model, including: In a case where the highest confidence level among the confidence levels corresponding to the K spatial filters is equal to or higher than a third threshold, the first communication device determines that the prediction result of the first network model is accurate; and / or, When the highest confidence level among the confidence levels corresponding to the K spatial filters is lower than a third threshold, the first communication device determines that the prediction result of the first network model is inaccurate.

8. The method according to claim 7, wherein The K spatial filters included in the first prediction data set are spatial filters corresponding to the first K confidences in the confidence vector output by the first network model, sorted from high to low; Wherein, the length of the confidence vector is W, and each element v of the confidence vector is j represents the confidence that the jth spatial filter among the W spatial filters is the optimal spatial filter, 1≤j≤W, v j ∈(0,1).

9. The method according to claim 8, wherein The output layer of the first network model adopts a fully connected layer with W neurons.

10. The method according to any one of claims 7 to 9, characterized in that The method further comprises: Within the third time window or the third time length, if the number of times the highest confidence level among the confidence levels corresponding to the K spatial filters is lower than the third threshold is greater than or equal to the fourth threshold, the first communication device determines that a third type of model failure has occurred in the first network model.

11. The method according to claim 6 or 10, wherein: The method further comprises: The first communication device determines to switch from the first network model to a second network model that implements the same function as the first network model.

12. The method according to claim 11, wherein The method further comprises: The first communication device sends first information; The first information includes at least one of the following: a model identifier of the first network model, a model failure type corresponding to the first network model, and a model identifier of the second network model.

13. The method according to claim 12, wherein: The first information is used to request at least one of the following: Configuring and / or activating a spatial filter prediction range of the prediction data set corresponding to the second network model; Configure and / or activate measurement resources of the measurement data set corresponding to the second network model.

14. The method according to claim 13, wherein When the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the first information is also used to request configuration and / or activation of model monitoring resources of the monitoring signal set corresponding to the second network model.

15. The method according to claim 6 or 10, characterized in that The method further comprises: The first communication device sends second information; The second information is used to request replacement of a network model having the same function as that implemented by the first network model.

16. The method according to claim 15, wherein The second information includes at least one of the following: a model identifier of the first network model, and a model failure type corresponding to the first network model.

17. The method according to claim 15 or 16, wherein: The method further comprises: The first communication device receives third information; The third information is used to instruct deactivation of the first network model and activation of a third network model having the same function as that implemented by the first network model.

18. The method according to claim 17, wherein The third information is used to configure and / or activate at least one of the following: The spatial filter prediction range of the prediction data set corresponding to the third network model; The measurement resources of the measurement data set corresponding to the third network model.

19. The method according to claim 18, wherein When the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the third information is also used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the third network model.

20. The method according to claim 6 or 10, wherein: The method further comprises: The first communication device sends fourth information; The fourth information is used to request updating of the model parameters of the first network model.

21. The method according to claim 20, wherein The first communication device receives fifth information; The fifth information is used to configure and / or activate at least one of the following: updating measurement resources of the prediction data set required for the first network model; The measurement resources of the measurement data set required for updating the first network model.

22. The method according to claim 6 or 10, wherein: The method further comprises: The first communication device implements the functions implemented by the first network model through other non-network model methods.

23. The method according to claim 22, wherein The method further comprises: The first communications device releases measurement resources of the first measurement data set.

24. The method according to claim 23, wherein In a case where the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the method further includes: The first communication device releases monitoring resources of the first monitoring data set.

25. The method according to claim 6 or 10, wherein The method further comprises: The first communication device sends model monitoring information; wherein the model monitoring information includes a model identifier of the first network model and a model failure type corresponding to the first network model.

26. The method of claim 25, wherein: In a case where the first communication device monitors the prediction performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the model monitoring information further includes a prediction accuracy rate of the first network model; or, In the case where the first communication device monitors the prediction performance of the first network model based on the confidences corresponding to the K spatial filters, the model monitoring information also includes an average value of the maximum values ​​of the confidences corresponding to the K spatial filters within a fourth time period.

27. The method according to claim 26, wherein The prediction accuracy of the first network model included in the model monitoring information is represented by multiple quantization intervals; or, The average value of the maximum value of the confidences corresponding to the K spatial filters included in the model monitoring information within the fourth time length is represented by multiple quantization intervals.

28. The method according to any one of claims 25 to 27, characterized in that The first communication device receives sixth information The sixth information is used to instruct the first communication device to deactivate the first network model and activate a fourth network model having the same function as that implemented by the first network model.

29. The method of claim 28, wherein The sixth information is used to configure and / or activate at least one of the following: The spatial filter prediction range of the prediction data set corresponding to the fourth network model; The measurement resources of the measurement data set corresponding to the fourth network model.

30. The method of claim 29, wherein: When the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the sixth information is also used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the fourth network model.

31. The method according to any one of claims 25 to 27, wherein The method further comprises: The first communication device receives seventh information; The seventh information is used to indicate resources required to update the model parameters of the first network model.

32. The method of claim 31, wherein The seventh information is used to configure and / or activate at least one of the following: updating measurement resources of the prediction data set required for the first network model; The measurement resources of the measurement data set required for updating the first network model.

33. The method according to any one of claims 25 to 27, wherein The method further comprises: The first communication device receives eighth information; wherein, the eighth information is used to instruct the first communication device to implement the function implemented by the first network model through other non-network model methods.

34. The method of claim 33, wherein: The eighth information is further used to instruct the first communications device to release measurement resources of the first measurement data set.

35. The method of claim 34, wherein: In a case where the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the eighth information is further used to instruct the first communication device to release monitoring resources of the first monitoring data set.

36. The method according to any one of claims 1 to 35, wherein The first communication device is a terminal device.

37. The method according to claim 6 or 10, wherein The method further comprises: The first communication device deactivates the first network model and activates a fifth network model having the same function as that implemented by the first network model.

38. The method of claim 37, wherein: The method further comprises: The first communication device configures and / or activates at least one of the following: The spatial filter prediction range of the prediction data set corresponding to the fifth network model; The measurement resources of the measurement data set corresponding to the fifth network model.

39. The method of claim 38, wherein In a case where the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the method further includes: The first communication device configures and / or activates model monitoring resources of the monitoring signal set corresponding to the fifth network model.

40. The method according to claim 6 or 10, wherein The method further comprises: The first communication device determines resources required to update model parameters of the first network model.

41. The method of claim 40, wherein: The method further comprises: The first communication device updates at least one of the following: measurement resources of the prediction data set required by the first network model; The measurement resources of the measurement data set required by the first network model.

42. The method of any one of claims 1 to 10, 22 to 24, 37 to 41, wherein: The first communication device is a network device.

43. The method according to any one of claims 1 to 42, wherein The spatial filter comprises a transmit spatial filter; or, The spatial filter includes a transmitting spatial filter and a receiving spatial filter.

44. A communication device, characterized in that The communication device is a first communication device, and the communication device includes: a processing unit, configured to input a first measurement data set into a first network model and output a first prediction data set; wherein the first measurement data set includes at least one of the following: identification information of F spatial filters and link quality information corresponding to the F spatial filters; and the first prediction data set includes at least one of the following: identification information of K spatial filters predicted from the W spatial filters and link quality information corresponding to the K spatial filters predicted from the W spatial filters; wherein F, W, and K are all positive integers, and K<W; The processing unit is further configured to monitor the prediction performance of the first network model based on the measurement results corresponding to the first monitoring signal set; wherein the first monitoring signal set includes M reference signals, each of the M reference signals satisfies a spatial quasi-co-location (QCL) relationship with a reference signal corresponding to at least one of the W spatial filters, and M is a positive integer; or The processing unit is further configured to monitor the prediction performance of the first network model according to the confidence levels corresponding to the K spatial filters output by the first network model.

45. The apparatus of claim 44, wherein The processing unit is specifically configured to: If there is a spatial filter among the K spatial filters that satisfies the QCL relationship with the first spatial filter, determining that the prediction result of the first network model is accurate; and / or, If there is no spatial filter among the K spatial filters that satisfies the QCL relationship with the first spatial filter, determining that the prediction result of the first network model is inaccurate; The first spatial filter is an optimal spatial filter included in the measurement result corresponding to the first monitoring signal set, or the first spatial filter is an optimal spatial filter determined based on the measurement result corresponding to the first monitoring signal set.

46. ​​The apparatus of claim 45, wherein The measurement result corresponding to the first monitoring signal set includes at least one of the following: identification information of the spatial filter corresponding to the M reference signals, link quality information corresponding to the M reference signals; or The measurement result corresponding to the first monitoring signal set includes identification information of the optimal spatial filter.

47. The apparatus according to claim 45 or 46, wherein The measurement result corresponding to the first monitoring signal set is obtained by measuring by the first communication device, or the measurement result corresponding to the first monitoring signal set is obtained by the first communication device from other devices.

48. Apparatus according to any one of claims 45 to 47, characterised in that The M reference signals include at least one of the following reference signals: Synchronization signal block SSB, channel state information reference signal CSI-RS.

49. The apparatus according to any one of claims 45 to 48, characterized in that Within a first time window or a first duration, if the number of inaccurate prediction results of the first network model is greater than or equal to a first threshold, the processing unit is further configured to determine that a first type of model failure has occurred in the first network model; and / or, Within the second time window or the second duration, if the prediction accuracy of the first network model is lower than a second threshold, the processing unit is further configured to determine that a second type of model failure has occurred in the first network model.

50. The apparatus of claim 44, wherein The processing unit is specifically configured to: When the highest confidence level among the confidence levels corresponding to the K spatial filters is equal to or higher than a third threshold, determining that the prediction result of the first network model is accurate; and / or, When the highest confidence level among the confidence levels corresponding to the K spatial filters is lower than a third threshold, it is determined that the prediction result of the first network model is inaccurate.

51. The apparatus of claim 50, wherein: The K spatial filters included in the first prediction data set are spatial filters corresponding to the first K confidences in the confidence vector output by the first network model, sorted from high to low; Wherein, the length of the confidence vector is W, and each element v of the confidence vector is j represents the confidence that the jth spatial filter among the W spatial filters is the optimal spatial filter, 1≤j≤W, v j ∈(0,1).

52. The apparatus of claim 51, wherein The output layer of the first network model adopts a fully connected layer with W neurons.

53. The apparatus according to any one of claims 50 to 52, characterized in that Within the third time window or the third time length, if the number of times the highest confidence level among the confidence levels corresponding to the K spatial filters is lower than the third threshold is greater than or equal to a fourth threshold, the processing unit is also used to determine that a third type of model failure has occurred in the first network model.

54. The apparatus of claim 49 or 53, wherein: The processing unit is further configured to determine to switch from the first network model to a second network model that implements the same function as the first network model.

55. The apparatus of claim 54, wherein The communication device further includes: a communication unit, configured to send first information; The first information includes at least one of the following: a model identifier of the first network model, a model failure type corresponding to the first network model, and a model identifier of the second network model.

56. The apparatus of claim 55, wherein: The first information is used to request at least one of the following: Configuring and / or activating a spatial filter prediction range of the prediction data set corresponding to the second network model; Configure and / or activate measurement resources of the measurement data set corresponding to the second network model.

57. The apparatus of claim 56, wherein When the first communication device monitors the predicted performance of the first network model based on the measurement results corresponding to the first monitoring signal set, the first information is also used to request configuration and / or activation of model monitoring resources of the monitoring signal set corresponding to the second network model.

58. The apparatus of claim 49 or 53, wherein: The communication device further includes: a communication unit, configured to send second information; The second information is used to request replacement of a network model having the same function as that implemented by the first network model.

59. The apparatus of claim 58, wherein The second information includes at least one of the following: a model identifier of the first network model, and a model failure type corresponding to the first network model.

60. The apparatus according to claim 58 or 59, wherein The communication unit is further configured to receive third information; The third information is used to instruct deactivation of the first network model and activation of a third network model having the same function as that implemented by the first network model.

61. The apparatus of claim 60, wherein: The third information is used to configure and / or activate at least one of the following: The spatial filter prediction range of the prediction data set corresponding to the third network model; The measurement resources of the measurement data set corresponding to the third network model.

62. The apparatus of claim 61, wherein When the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the third information is also used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the third network model.

63. The apparatus of claim 49 or 53, wherein: The communication device further includes: a communication unit, configured to send fourth information; The fourth information is used to request updating of the model parameters of the first network model.

64. The apparatus of claim 63, wherein The communication unit is further configured to receive fifth information; The fifth information is used to configure and / or activate at least one of the following: updating measurement resources of the prediction data set required for the first network model; The measurement resources of the measurement data set required for updating the first network model.

65. The apparatus of claim 49 or 53, wherein: The processing unit is further configured to implement the functions implemented by the first network model through other non-network model methods.

66. The apparatus of claim 65, wherein The processing unit is further configured to release measurement resources of the first measurement data set.

67. The apparatus of claim 66, wherein In a case where the first communication device monitors the prediction performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the processing unit is further configured to release monitoring resources of the first monitoring data set.

68. The apparatus of claim 49 or 53, wherein: The communication device further includes: A communication unit is used to send model monitoring information; wherein the model monitoring information includes the model identifier of the first network model and the model failure type corresponding to the first network model.

69. The apparatus of claim 68, wherein In a case where the first communication device monitors the prediction performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the model monitoring information further includes a prediction accuracy rate of the first network model; or, In the case where the first communication device monitors the prediction performance of the first network model based on the confidences corresponding to the K spatial filters, the model monitoring information also includes an average value of the maximum values ​​of the confidences corresponding to the K spatial filters within a fourth time period.

70. The apparatus of claim 69, wherein The prediction accuracy of the first network model included in the model monitoring information is represented by multiple quantization intervals; or, The average value of the maximum value of the confidences corresponding to the K spatial filters included in the model monitoring information within the fourth time length is represented by multiple quantization intervals.

71. The apparatus of any one of claims 68 to 70, wherein The communication unit is further configured to receive a sixth information The sixth information is used to instruct the first communication device to deactivate the first network model and activate a fourth network model having the same function as that implemented by the first network model.

72. The apparatus of claim 71, wherein The sixth information is used to configure and / or activate at least one of the following: The spatial filter prediction range of the prediction data set corresponding to the fourth network model; The measurement resources of the measurement data set corresponding to the fourth network model.

73. The apparatus of claim 72, wherein: When the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the sixth information is also used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the fourth network model.

74. The apparatus of any one of claims 68 to 70, wherein The communication unit is further configured to receive seventh information; The seventh information is used to indicate resources required to update the model parameters of the first network model.

75. The apparatus of claim 74, wherein: The seventh information is used to configure and / or activate at least one of the following: updating measurement resources of the prediction data set required for the first network model; The measurement resources of the measurement data set required for updating the first network model.

76. The apparatus of any one of claims 68 to 70, wherein The communication unit is further used to receive eighth information; wherein, the eighth information is used to instruct the first communication device to implement the function implemented by the first network model through other non-network model methods.

77. The apparatus of claim 76, wherein The eighth information is further used to instruct the first communications device to release measurement resources of the first measurement data set.

78. The apparatus of claim 77, wherein In a case where the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the eighth information is further used to instruct the first communication device to release monitoring resources of the first monitoring data set.

79. The apparatus of any one of claims 44 to 78, wherein The first communication device is a terminal device.

80. The apparatus of claim 49 or 53, wherein: The processing unit is further configured to deactivate the first network model and activate a fifth network model having the same function as that implemented by the first network model.

81. The apparatus of claim 80, wherein The processing unit is further configured to configure and / or activate at least one of the following: The spatial filter prediction range of the prediction data set corresponding to the fifth network model; The measurement resources of the measurement data set corresponding to the fifth network model.

82. The apparatus of claim 81, wherein When the first communication device monitors the predicted performance of the first network model according to the measurement results corresponding to the first monitoring signal set, the processing unit is further used to configure and / or activate the model monitoring resources of the monitoring signal set corresponding to the fifth network model.

83. The apparatus of claim 44 or 53, wherein: The processing unit is further configured to determine resources required to update model parameters of the first network model.

84. The apparatus of claim 83, wherein The device further comprises: The first communication device updates at least one of the following: measurement resources of the prediction data set required by the first network model; The measurement resources of the measurement data set required by the first network model.

85. The apparatus of any one of claims 44 to 53, 65 to 67, 80 to 84, wherein The first communication device is a network device.

86. The apparatus of any one of claims 44 to 85, wherein The spatial filter comprises a transmit spatial filter; or, The spatial filter includes a transmitting spatial filter and a receiving spatial filter.

87. A communication device, characterized in that include: A processor and a memory, the memory being used to store a computer program, the processor being used to call and run the computer program stored in the memory, so that the communication device executes the method according to any one of claims 1 to 43.

88. A chip, characterized in that include: A processor, configured to call and execute a computer program from a memory, so that a device equipped with the chip executes a method as claimed in any one of claims 1 to 43.

89. A computer-readable storage medium, characterized in that For storing a computer program, when the computer program is executed, the method according to any one of claims 1 to 43 is implemented.

90. A computer program product, characterized in that The method comprises computer program instructions which, when executed, implement the method according to any one of claims 1 to 43.

91. A computer program, characterized in that When the computer program is executed, the method according to any one of claims 1 to 43 is implemented.

Images