Methods and apparatuses for wireless communication

Abstract

Provided are methods and apparatuses for wireless communication. A method comprises: a first device receiving a first reference signal, the first reference signal being sent by means of a first beam set; and the first device performing performance monitoring on a first model on the basis of a measurement result of the first beam set, wherein the first reference signal corresponds to a first monitoring instance, and the first beam set is determined on the basis of an inference beam set of the first model; and the first beam set comprises all beams in the inference beam set, or the first beam set comprises some beams in the inference beam set.

Description

Methods and apparatus for wireless communication Technical Field

[0001] This application relates to the field of communication technology, and more specifically, to a method and apparatus for wireless communication. Background Technology

[0002] In certain beam management scenarios, terminal devices can predict downlink transmission beams using a model and report this prediction to the network device. This model needs to be trained before it can make inference predictions. However, there is a time interval between the training and inference phases. If the beam transmission conditions change, the model may make errors during the inference phase. Therefore, how to monitor the model's performance is a crucial technical problem that needs to be solved. Summary of the Invention

[0003] This application provides a method and apparatus for wireless communication. The various aspects related to the embodiments of this application are described below.

[0004] In a first aspect, a method for wireless communication is provided, comprising: a first device receiving a first reference signal, the first reference signal being transmitted through a first beam set; the first device monitoring the performance of a first model based on measurement results from the first beam set; wherein the first reference signal corresponds to a first monitoring instance, and the first beam set is determined based on an inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes a portion of the beams in the inference beam set.

[0005] In a second aspect, a method for wireless communication is provided, comprising: a second device transmitting a first reference signal, the first reference signal being transmitted through a first beam set; wherein the first reference signal corresponds to a first monitoring instance, the measurement results of the first beam set are used to monitor the performance of a first model, and the first beam set is determined based on the inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes a portion of the beams in the inference beam set.

[0006] Thirdly, an apparatus for wireless communication is provided, the apparatus being a first device, comprising: a transceiver unit for receiving a first reference signal, the first reference signal being transmitted through a first beam set; and a processing unit for monitoring the performance of a first model based on measurement results from the first beam set; wherein the first reference signal corresponds to a first monitoring instance, and the first beam set is determined based on an inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes a portion of the beams in the inference beam set.

[0007] Fourthly, an apparatus for wireless communication is provided, the apparatus being a second device, comprising: a transceiver unit for transmitting a first reference signal, the first reference signal being transmitted through a first beam set; wherein the first reference signal corresponds to a first monitoring instance, the measurement results of the first beam set are used for performance monitoring of a first model, and the first beam set is determined based on the inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes a portion of the beams in the inference beam set.

[0008] Fifthly, a communication device is provided, including a memory and a processor, the memory for storing a program, and the processor for calling the program in the memory to perform the method as described in the first or second aspect.

[0009] A sixth aspect provides an apparatus including a processor for calling a program from memory to perform the method as described in the first or second aspect.

[0010] A seventh aspect provides a chip including a processor for calling a program from memory, causing a device having the chip mounted to perform the method as described in the first or second aspect.

[0011] Eighthly, a computer-readable storage medium is provided having a program stored thereon that causes a computer to perform the method as described in the first or second aspect.

[0012] Ninth aspect, a computer program product is provided, including a program that causes a computer to perform the method as described in the first or second aspect.

[0013] In a tenth aspect, a computer program is provided that causes a computer to perform the method as described in the first or second aspect.

[0014] In this embodiment, the first device (e.g., a terminal device) can determine a first beam set based on the inference beam set of the first model, and then monitor the performance of the first model based on the measurement results of the first beam set. The first beam set can be an inference beam set or a subset of the inference beam set, thereby improving the accuracy of the first model through real-time performance detection of the inference beams. Attached Figure Description

[0015] Figure 1 shows the wireless communication system used in an embodiment of this application.

[0016] Figure 2 is a schematic diagram of the model processing procedure applied in the embodiments of this application.

[0017] Figure 3 is a flowchart illustrating a method for wireless communication provided in an embodiment of this application.

[0018] Figure 4 is a schematic diagram of one possible implementation of the method shown in Figure 3.

[0019] Figure 5 is a schematic diagram of another possible implementation of the method shown in Figure 3.

[0020] Figure 6 is a flowchart illustrating one possible implementation of the method shown in Figure 3.

[0021] Figure 7 is a schematic diagram of one possible configuration of the first beam set in Figure 3.

[0022] Figure 8 is a schematic diagram of another possible configuration of the first beam set in Figure 3.

[0023] Figure 9 is a schematic diagram of another possible configuration of the first beam set in Figure 3.

[0024] Figure 10 is a schematic diagram of another possible configuration of the first beam set in Figure 3.

[0025] Figure 11 is a schematic diagram of another possible configuration of the first beam set in Figure 3.

[0026] Figure 12 is a schematic diagram of a device for wireless communication provided in an embodiment of this application.

[0027] Figure 13 is a schematic diagram of another device for wireless communication provided in an embodiment of this application.

[0028] Figure 14 is a schematic diagram of the structure of a wireless communication device provided in an embodiment of this application. Detailed Implementation

[0029] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0030] Figure 1 is a schematic diagram of the architecture of the wireless communication system 100 used in an embodiment of this application. As shown in Figure 1, the wireless communication system 100 may include a network device 110 and a terminal device 120. The network device 110 may be a device that communicates with the terminal device 120. The network device 110 may provide communication coverage for a specific geographical area and may communicate with terminal devices located within that coverage area.

[0031] Figure 1 exemplarily illustrates a network device and two terminal devices. Optionally, the wireless communication system 100 may include multiple network devices, and each network device may include an additional number of terminal devices within its coverage area; this is not limited. In other words, the wireless communication system may include one or more network devices, and each network device may support wireless communication for one or more terminal devices.

[0032] In the embodiments of this application, the communication system shown in FIG1 may also include other network entities such as a mobility management entity (MME), an access and mobility management function (AMF), and a network controller. The embodiments of this application do not limit this.

[0033] It should be understood that the embodiments of this application can be applied to various communication systems. For example, the embodiments of this application can be applied to Global System for Mobile Communication (GSM), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), General Packet Radio Service (GPRS), Long Term Evolution (LTE), Advanced Long Term Evolution (LTE-A), New Radio (NR), evolution systems of NR, LTE-based access to unlicensed spectrum (LTE-U), NR-based access to unlicensed spectrum (NR-U), Universal Mobile Telecommunications System (UMTS), Wireless Local Area Networks (WLAN), Wireless Fidelity (WiFi), and 5th-generation (5G) systems. The embodiments of this application can also be applied to other communication systems, such as 6th-generation (6G) mobile communication systems, or future communication systems such as satellite communication systems.

[0034] Traditional communication systems support a limited number of connections and are easy to implement. However, with the development of communication technology, communication systems can support not only traditional cellular communication but also one or more other types of communication. For example, a communication system can support one or more of the following communication methods: device-to-device (D2D) communication, machine-to-machine (M2M) communication, machine-type communication (MTC), enhanced machine-type communication (eMTC), vehicle-to-vehicle (V2V) communication, and vehicle-to-everything (V2X) communication. The embodiments of this application can also be applied to communication systems that support the above-mentioned communication methods.

[0035] The communication system in this application embodiment can be applied to carrier aggregation (CA) scenarios, dual connectivity (DC) scenarios, and standalone (SA) network deployment scenarios.

[0036] The communication system in this application embodiment can be applied to unlicensed spectrum. This unlicensed spectrum can also be considered a shared spectrum. Alternatively, the communication system in this application embodiment can also be applied to licensed spectrum. This licensed spectrum can also be considered a dedicated spectrum.

[0037] The embodiments of this application can be applied to non-terrestrial network (NTN) systems. As an example, the NTN system can be a 4G-based NTN system, an NR-based NTN system, an Internet of Things (IoT)-based NTN system, or a narrowband Internet of Things (NB-IoT)-based NTN system.

[0038] The wireless communication system in this application embodiment can utilize the following resources to support wireless communication with one or more communication devices: time resources (e.g., symbols, sub-slots, time slots, subframes, frames, etc.) or frequency resources (e.g., subcarriers, carriers). Additionally, the wireless communication system can support wireless communication across various radio access technologies (RATs), including third-generation (3G), fourth-generation (4G), fifth-generation (5G), and other suitable RATs beyond 5G.

[0039] The terminal equipment in this application embodiment may also be referred to as user equipment (UE), access terminal, user unit, user station, mobile station, mobile station (MS), mobile terminal (MT), remote station, remote terminal, mobile device, user terminal, terminal, user communication equipment, wireless communication equipment, user agent, or user device, etc.

[0040] In some embodiments, the terminal device in this application can be a device that provides voice and / or data connectivity to a user, and can be used to connect people, objects, and machines, such as a handheld device with wireless connectivity, an in-vehicle device, etc. The terminal device in the embodiments of this application can be a mobile phone, tablet computer, laptop computer, PDA, mobile internet device (MID), wearable device, virtual reality (VR) device, augmented reality (AR) device, wireless terminal in industrial control, wireless terminal in self-driving, wireless terminal in remote medical surgery, wireless terminal in smart grid, wireless terminal in transportation safety, wireless terminal in smart city, wireless terminal in smart home, etc. Optionally, the UE can be used to act as a base station. For example, the UE can act as a scheduling entity, providing sidelink signals between UEs in V2X or D2D, etc. For example, cellular phones and cars communicate with each other using sidelink signals. Cellular phones and smart home devices can communicate without relaying communication signals through base stations.

[0041] In some embodiments, the terminal device may be a station (ST) in a WLAN. In some embodiments, the terminal device may be a cellular phone, cordless phone, session initiation protocol (SIP) phone, wireless local loop (WLL) station, personal digital assistant (PDA) device, handheld device with wireless communication capabilities, computing device or other processing device connected to a wireless modem, in-vehicle device, wearable device, terminal device in a next-generation communication system (e.g., NR system), or terminal device in a future public land mobile network (PLMN) network, etc.

[0042] The network device in this application embodiment can be a device for communicating with a terminal device, and can also be referred to as an access network device or a radio access network device. For example, the network device can be a base station. In this application embodiment, the network device can refer to a radio access network (RAN) node (or device) that connects the terminal device to the wireless network. A base station can broadly encompass, or be replaced by, various names including: NodeB, evolved NodeB (eNB), next-generation NodeB (gNB), relay station, access point, transmitting and receiving point (TRP), transmitting point (TP), master station (MeNB), secondary station (SeNB), multi-mode radio (MSR) node, home base station, network controller, access node, wireless node, access point (AP), transmission node, transceiver node, baseband unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), central unit (CU), distributed unit (DU), location node, network communication equipment, etc. A base station can be a macro base station, micro base station, relay node, donor node, or similar entities, or combinations thereof. A base station can also refer to a communication module, modem, or chip installed within the aforementioned equipment or apparatus. A base station can also be a mobile switching center, a device that performs base station functions in D2D, V2X, and M2M communications, a network-side device in a 6G network, or a device that performs base station functions in future communication systems. A base station can support networks using the same or different access technologies. The embodiments of this application do not limit the specific technologies or device forms used in the network equipment.

[0043] Base stations can be fixed or mobile. For example, a helicopter or drone can be configured to act as a mobile base station, and one or more cells can move depending on the location of the mobile base station. In other examples, a helicopter or drone can be configured as a device to communicate with another base station.

[0044] In some deployments, the network device in this application embodiment may refer to a CU or a DU, or the network device may include both a CU and a DU. The gNB may also include an AAU.

[0045] Network devices and terminal devices can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted; they can also be deployed on water; and they can also be deployed in the air on airplanes, balloons, and satellites. This application does not limit the scenario in which the network devices and terminal devices are located.

[0046] It should be understood that all or part of the functions of the communication device in this application can also be implemented by software functions running on hardware, or by virtualization functions instantiated on a platform (e.g., a cloud platform).

[0047] In this embodiment, the network device can provide services to a cell. The terminal device communicates with the network device through the transmission resources (e.g., frequency domain resources, or spectrum resources) used by the cell. The cell can be the cell corresponding to the network device (e.g., a base station). The cell can belong to a macro base station or to a base station corresponding to a small cell. The small cell can include: metro cell, micro cell, pico cell, femto cell, etc. These small cells have the characteristics of small coverage area and low transmission power, and are suitable for providing high-speed data transmission services.

[0048] It should be understood that devices with communication functions in the network / system of this application embodiment can be referred to as communication devices. Taking the wireless communication system 100 shown in FIG1 as an example, the communication device may include network device 110 and terminal device 120 with communication functions, and may also include other devices in the wireless communication system 100, such as network controllers, mobility management entities and other network entities. This application embodiment does not limit this.

[0049] For example, a wireless communication system may include one or more network communication devices, such as the base station described above. Each network communication device, such as a base station, may support wireless communication for one or more user communication devices (e.g., terminal devices).

[0050] To facilitate understanding, some related technical knowledge involved in the embodiments of this application is first introduced. The following related technologies are optional solutions and can be arbitrarily combined with the technical solutions of the embodiments of this application, all of which fall within the protection scope of the embodiments of this application. The embodiments of this application include at least some of the following contents.

[0051] In wireless communication systems, terminal devices can acquire beams (also known as spatial beams) to establish a wireless connection to a wireless network. For example, the terminal device can perform beam scanning against available beams transmitted by the wireless network and measure beam properties such as signal strength and signal quality. For example, after performing beam scanning, the terminal device can also perform beam thinning to achieve a potentially narrower set of beams for wireless connection to the wireless network. Beaming not only enables wireless connectivity between the terminal device and the wireless network but also achieves high directional accuracy and high signal quality for wireless signal transmission between the terminal device and the wireless network.

[0052] With the development of communication technology, research on artificial intelligence (AI) / machine learning (ML) technologies based on the air interface of communication systems (e.g., NR systems) has become a research direction. The goals of this research include exploring how to enhance the advantages of the air interface. For example, enhancing support for AI / ML algorithms can improve the performance of the air interface. Similarly, enhancing support for AI / ML algorithms can reduce the complexity and / or overhead of the air interface.

[0053] Research into AI / ML technologies can also enhance beam management (BM) capabilities. As an example, AI / ML enhancements related to beam management can help reduce overhead and lower beam measurement and reporting latency. As another example, applying AI / ML models can predict beams to improve transmission efficiency at the air interface.

[0054] The entire process of augmenting using AI / ML models includes model training, model inference, and model monitoring. During this process, after training, the AI / ML model can generate a set of outputs based on a set of inputs. The input can be a set of beam measurements, while the output can be a set of beams that are different from or larger than the inputs.

[0055] During the model inference process, AI / ML models can predict the optimal beam in a set of different / larger beam groups using a set of beam measurements.

[0056] In some embodiments, the AI / ML model may be located on the terminal device side or the terminal device may perform model training and / or model inference. This model may be referred to as a UE-side model. For example, the AI ​​model may be located on the terminal device, and the AI ​​model may be trained and / or its inference may be performed on or by the terminal device to generate the optimal beam.

[0057] As an example, a terminal device can use beams from set B (also called beam group B) as input to an ML model. This ML model can predict the optimal beam in set A (also called beam group A), which is not fully measured by the terminal device.

[0058] In the example above, set B can be the beam group initially measured by the terminal device, also known as the training beam set. Set B can be multiple beams transmitted by the base station (e.g., gNB). Each beam can correspond to a different direction or angle to cover multiple spatial directions. Each beam can also correspond to a measurement signal used to obtain measurement values, such as reference signal received power (RSRP). The role of set B is to provide the model with preliminary environmental information and channel conditions.

[0059] In the example above, set A can be the beam group that needs to be predicted during the inference phase, also known as the inference beam set. Set A typically has a larger number of beams than set B, or the beam directions may be more concentrated. AI / ML models can predict the optimal beam in set A by measuring set B, thereby improving data transmission efficiency.

[0060] Optionally, the beams of set A and set B can be in the same frequency range. The selection of set B can be given by the base station or determined by the terminal equipment itself. The relationship between set A and set B can be: set A and set B are different (set B is not a subset of set A), or set B is a subset of set A (set A and set B are different), or set A and set B are the same. For the first two cases, set B can be transmitted simultaneously in the measurement window and the prediction window, or it can be transmitted only in the measurement window. The last case can save the reference signal (RS) transmission overhead, as set B, as a measurement resource, can be transmitted only in the measurement window.

[0061] Alternatively, 64 or more beams can be used as the size of the beam set A. For future-oriented networks, network devices will be able to transmit 64 more highly directional narrow beams. More narrow beams can also scan a larger set A; for example, the number of beams in set A could be as high as 256.

[0062] Optionally, network devices can transmit various reference signals, such as channel state information (CSI) reference signals (CSI-RS) and synchronization signal blocks (SSBs). It should be noted that SSB can also represent a synchronization signal / physical broadcast channel block (SS / PBCH block). An SSB can include a primary synchronization signal (PSS) and a secondary synchronization signal (SSS).

[0063] Optionally, the terminal device estimates the channel quality of each beam by measuring the RSRP received from the CSI-RS / SSS.

[0064] Optionally, during model training, the AI / ML model can adjust its weights by minimizing the loss function so that the model can accurately predict the optimal beam in set A from the RSRP measurements of set B.

[0065] In some embodiments, AI / ML can be located at the network device (such as a base station) or the network device can perform model training and / or model inference. This model can be referred to as a network-side model (NW-side model). For example, the AI ​​model is located at the base station, and the training of the AI ​​model and / or the inference of the AI ​​model can be performed at the base station or by the base station to generate the optimal beam.

[0066] In some embodiments, the network can have complete control over the data collection process for model training on the terminal device side, including the initiation, termination, and management of data collection and data transmission.

[0067] As an example, when a terminal device uses AI algorithms or ML to process beams detected and measured from a wireless network to infer other beams that may have higher strength and / or higher quality, the terminal device needs to ensure the consistency of the model during the training and inference phases.

[0068] Compared to other beam management technologies, beam management that supports AI / ML technologies enables terminal devices to experience reduced latency, reduced overhead, lower power consumption, and improved signal quality based on beam prediction.

[0069] To facilitate understanding, the entire process of model processing on the terminal device side is described below with reference to Figure 2. Figure 2 illustrates the interaction between the terminal device (e.g., UE) and the network (NW) side. Four beams are used as an example on the network side. As shown in Figure 2, the entire process can include model training, model inference, and reporting. The model training process includes steps S210 and S220, and the model inference process includes steps S230 and S240.

[0070] Referring to Figure 2, in step S210, the terminal device reports training-related information (UEreport training-related information).

[0071] In step S220, the network side performs beam scanning based on four beams.

[0072] In step S230, the terminal device reports inference-related information.

[0073] In step S240, the network side selects two beams from the four beams for beam scanning based on the report from the terminal device.

[0074] In step S250, the terminal device reports the optimal K beams (top-K beam report).

[0075] In step S260, the network side performs beam scanning based on the beams in the beam report. As shown in Figure 2, the two beams used for beam scanning by the network side in step S260 may be different from the two beams used for beam scanning in step S240.

[0076] In step S270, the terminal device sends a beam report so that the network side can determine the optimal beam.

[0077] In step S280, the network side sends a beam indication to the terminal device.

[0078] The above, along with Figure 2, illustrates the process of processing the model when it is located on the terminal device side. To complete this process, the terminal device also needs to collect and analyze data to perform model training. The model inference process is primarily used for beam prediction.

[0079] In relevant scenarios, terminal devices can support beam prediction in the spatial and / or temporal domains, namely BM-Case1 and / or BM-Case2. Based on AI / ML enhancements, beam prediction in the spatial domain (BM-Case1) and beam prediction in the temporal domain (BM-Case2) can reduce the overhead of terminal devices and lower beam measurement and reporting latency.

[0080] BM-Case 1 performs spatial domain downlink (DL) beam prediction on set A based on measurements from set B. For BM-case 1, measurements based on set B of beams are used as model input to predict the Top-1 / Top-K beams in set A.

[0081] BM-Case2 performs time-domain DL beam prediction on set A based on historical measurements from set B. For BM-Case2, measurements based on Set B of beams at historic time instances can be used as model input to predict the time-domain DL beams of set A. Predictions of DL Tx beams and DL Tx / Rx beams can also be used to evaluate prediction performance.

[0082] For BM-Case1 and BM-Case2, the terminal device can report the prediction results to the NW based on the output of the terminal device-side model, or the NW can predict the Top-1 / Top-K beams based on the measurement reports of the NW-side model set B.

[0083] The above section, with reference to Figure 2, introduced model training and model inference. To ensure the correctness of model inference, a model monitoring process is also required.

[0084] AI / ML model monitoring during the model monitoring process serves at least the following purposes: model activation, deactivation, selection, switching, rollback, and updating (including retraining). Model monitoring can also be described as the process of monitoring the inference performance of AI / ML models. There is always a time interval between the model's training and inference processes. When radio parameters / conditions change in the network, the likelihood of errors occurring during the model's inference phase is higher; therefore, it is necessary to continuously correct and train the model based on the results of model monitoring.

[0085] For example, model monitoring is required in certain situations (such as before using the model in a new radio environment / conditions / parameters).

[0086] During model monitoring, the performance of AI / ML models can be monitored using three methods. Model monitoring can also be called model surveillance or model monitoring. The three performance monitoring methods are: performance monitoring on the terminal device (UE) side, performance monitoring on the network device (gNB) side, and hybrid performance monitoring on both the terminal device side and the network device side.

[0087] In some scenarios, since the terminal device sends uplink beams and the network device sends downlink beams, regardless of whether the model exists on the terminal device side, the network device side, or both sides, the terminal device and the network device may perform beam prediction and reporting based on the model.

[0088] For performance monitoring on the terminal device side, configuration / signaling from the network device to the terminal device can be used for performance monitoring. For network device side and hybrid performance monitoring, configuration / signals from the network device to the terminal device can be used for measurement and / or reporting configuration / signals to support performance testing of the model. In some embodiments, the selection of performance monitoring methods is related to a variety of factors. These factors include, but are not limited to: device requirements, model location, service type, communication scenario, etc.

[0089] As mentioned earlier, there is a certain time interval between the training and inference phases, and changes in radio parameters / conditions may lead to errors in the inference phase. Therefore, in order to promptly identify model problems, how to monitor model performance has become an urgent technical issue to be addressed.

[0090] As an example, the selection of the dataset for model performance monitoring and the timing of model monitoring are both crucial. In other words, data collection for model performance monitoring is critical. This is because model monitoring can continue for an extended period, not ending until AI-based beam management is finalized. The selection of the performance monitoring dataset is essential to ensure efficient data collection and accurate model monitoring. Therefore, how to choose the appropriate performance monitoring dataset for efficient performance monitoring is a key consideration.

[0091] Based on this, embodiments of this application propose a method for wireless communication. In this method, a first beam set for performance monitoring of a first model is determined based on the inference beam set of the first model. Since the first beam set and the inference beam set are associated, a first device (e.g., a terminal device) can promptly compare the prediction results of the inference beam set with the measurement results to ensure the accuracy of the model. It should be understood that the first model in embodiments of this application may be located on the terminal device side and / or the network device side.

[0092] The method for wireless communication proposed in this application will be described in detail below with reference to Figure 3. Figure 3 is presented from the perspective of the interaction between the first device and the second device.

[0093] The first device can be any communication device that supports model performance monitoring. In some embodiments, the first device can be a terminal device. For example, the first device can be a UE with monitoring capabilities. Alternatively, the first device can include a terminal device and any processing device that supports performance monitoring. In some embodiments, the first device can be a network device. For example, the first device can be a base station.

[0094] In some embodiments, the first device supports functional enhancements based on AI / ML operations. For example, the first device has the ability to enhance beam management and / or monitor performance based on AI / ML operations.

[0095] In some embodiments, a first model is deployed on the first device to perform beam prediction. When the second device is a network device, the first device performs DL beam prediction. When the second device is a terminal device, the first device performs side-travel beam prediction.

[0096] As an example, the first model is a model that supports AI algorithms or ML, that is, the first model is an AI / ML model.

[0097] As an example, the beam prediction implemented by the first model can be BM-Case1 as mentioned above, or BM-Case2, or other beam prediction types in the future, which are not limited here.

[0098] In some embodiments, a first model is deployed on the first device side. The first model may not be on the terminal device, but on a server that communicates with the terminal device. For example, the first model is on a server that communicates directly with the terminal device.

[0099] The second device can be any network device communicating with the first device, or it can be a terminal device communicating with the first device. When the first device is a terminal device and is within the coverage area of ​​a network device, the second device can be a network device. When the first device is a network device, the second device can be a terminal device communicating with the network device. In a side-by-side communication system, when the first device communicates with other terminal devices, the second device can also be other terminal devices.

[0100] In some embodiments, the second device can monitor the process by which the first device processes the first model. For example, the second device can determine the current training and inference phases of the first model based on reports sent by the first device, and can also determine the performance monitoring results of the first model based on reports sent by the first device. Furthermore, the second device can assist the first device in monitoring the performance of the first model.

[0101] In some embodiments, the second device supports AI / ML operations. The first model may be deployed on the second device.

[0102] In some embodiments, the second device may send multiple beams to the first device multiple times to enable the first device to perform measurements and to perform model training, model inference, and / or model monitoring on the first model based on the measurement results.

[0103] In the above embodiments, regardless of whether the first model is deployed on the terminal device side or the network device side, the first model can be one of multiple models deployed on that side. Multiple models can be used to predict the transmission beams of different scenarios or different beam transmitting devices.

[0104] Referring to Figure 3, in step S310, the first device receives the first reference signal. The first reference signal is transmitted through the first beam set.

[0105] The first reference signal is used by the first device to monitor the performance of the first model. As an example, the first reference signal is used by the first device to collect a dataset for performance monitoring of the first model. This dataset can also be called a performance monitoring set.

[0106] In some embodiments, the first reference signal is at least one of dedicated reference signals related to performance monitoring. The dedicated reference signal may also be referred to as a proprietary reference signal, a performance monitoring reference signal, or an auxiliary reference signal. As an example, the first reference signal may be a single dedicated reference signal. As an example, the first reference signal may be multiple dedicated reference signals.

[0107] As an example, a dedicated reference signal can be used to determine the performance parameters of the first model. These performance parameters can also be referred to as key performance indicators (KPIs) for performance monitoring. Optionally, the performance parameters of the first model may include the prediction accuracy and / or the prediction error of the first model.

[0108] The first reference signal corresponds to the first monitoring instance. The first reference signal is used by the first device to monitor the performance of the first model at its corresponding time instance; therefore, the first monitoring instance can also be called the first performance monitoring instance. As an example, the first monitoring instance can be one or more monitoring instances among multiple monitoring instances used to monitor the performance of the first model.

[0109] In some embodiments, the first monitoring instance may be one or more monitoring instances within a monitoring period of the first model. The monitoring period of the first model may be referred to as the first monitoring period. As one embodiment, the first monitoring period includes at least one monitoring instance. Within the first monitoring period, the first device may receive and measure a reference signal, and then generate a monitoring report based on the measurement results.

[0110] As an example, the first monitoring cycle includes a monitoring instance on which the first device performs performance monitoring.

[0111] As an example, the first monitoring period includes multiple monitoring instances, each corresponding to a reference signal at multiple points in time.

[0112] As an example, the first monitoring period is equal to or less than the reporting period of the monitoring report. When the reporting period is longer than the first monitoring period, the first device can perform multiple monitoring operations before sending a monitoring report.

[0113] The first reference signal is transmitted through a first beam set, or alternatively, the first reference signal is associated with the first beam set. The first beam set corresponds to a first monitoring instance. Therefore, the measurement results of the first beam set can serve as the monitoring dataset for the first model. For example, some or all of the beams in the first beam set can be used to transmit the first reference signal.

[0114] The first beam set is determined based on the inference beam set of the first model to ensure the efficiency of data collection and the accuracy of the monitoring model. The inference beam set is, for example, set A as described above. When the first beam set is determined based on the inference beam set, the first device can directly compare the measurement results of the first beam set with the inference results of the first model. As an example, the first reference signal transmitted by the first beam set can span the entire set A or only cover a subset of set A, so that the first device can compare the predictions about set A (the inference target set) with the actual measurements from the beams in set A.

[0115] In some embodiments, the inference beam set of the first model can be determined based on the first model. That is, the beams used for inference mentioned above and throughout the text can be determined based on the first model. As an example, the inference beam set of the first model can be determined based on specific features of the first model. The features of the first model are, for example, features related to model modeling and learning.

[0116] It should be noted that the inference beam can be called the prediction beam, and the inference beam set can also be called the prediction beam set.

[0117] In some embodiments, the first beam set includes all beams in the inference beam set. When monitoring the performance of the first model, the performance monitoring reference signal corresponding to each monitoring instance can be associated with the entire inference beam set. In this scenario, the performance monitoring dataset collected by the first device is equal to the dataset of the inference beam set.

[0118] As an example, the first beam set is a reasoning beam set. When all beams in the reasoning beam set (set A) transmit the first reference signal, the first device can obtain measurement values ​​from the entire set A and define monitoring procedures and metrics accordingly.

[0119] To facilitate understanding, the method of using the first beam set as the inference beam set is illustrated below with reference to Figure 4. In Figure 4, set B is the training beam set of the first model, and set A is the inference beam set of the first model. As shown in Figure 4, on the time axis, the training beam set of the first model is respectively at time instance [t] -3 ,t0,t +3 Transmission on [t]; the inference beamset of the first model is respectively in time instance [t] - 2,t -1 , t +1 , t +2 , t +4 Uploaded; Time instance [t] -2 , t -1 , t +1 , t +2 , t +4 This corresponds to multiple performance monitoring instances.

[0120] Referring to Figure 4, at each time instance corresponding to the inference beam set, the inference beam set is the same as the first beam set of the performance monitoring instance. Optionally, at each time instance [t] -2 , t -1 , t +1 , t +2 , t +4 On the above, the beam set of the first reference signal is the same as the inference beam set.

[0121] In some embodiments, the first beam set includes a subset of beams from the inference beam set. In certain scenarios, measuring all beams in the inference beam set can be highly complex and resource-intensive. For example, for large inference beam sets (e.g., 128 beams or more), measuring the entire inference beam set in each monitoring instance can lead to excessive power consumption for the first device. Therefore, selecting only a subset of the inference beam set for measurement in each performance monitoring instance reduces the measurement burden on the first device.

[0122] As an example, some beams in the inference beam set can form a first beam subset. That is, when the first beam set only includes some beams in the inference beam set, the first beam set can also be called the first beam subset of the inference beam set.

[0123] In the above embodiments, when the first beam set includes a portion of the beams, the portion of the beams may be the Top-K beams in the inference beam set.

[0124] In some embodiments, when the first beam set includes a portion of the beams in the inference beam set, the multiple monitoring instances used for performance monitoring can each correspond to a multiple beam subset. The first beam subset can be any beam subset from the multiple beam subsets that corresponds to the first reference signal. As an example, the multiple monitoring instances can include a first monitoring instance and a second monitoring instance, where the first monitoring instance corresponds to the first beam subset and the second monitoring instance corresponds to the second beam subset.

[0125] As an example, the first beam subset shares at least one beam with the second beam subset. That is, the first beam subset may overlap with some beams in the beam subsets corresponding to other monitoring instances. For instance, both the first and second beam subsets include beam 1 from the inference beam set. Beam 1 could be a beam with good signal quality from the inference beam set.

[0126] As an example, the first beam subset differs from at least one beam in the second beam subset. For instance, the first beam subset differs from all beams in the second beam subset. Or, the first beam subset differs from some beams in the second beam subset.

[0127] As an example, the first beam subset is dynamically selected. That is, the multiple beam subsets corresponding to multiple monitoring instances are dynamically determined. Each dynamically selected monitoring beam subset may have some beam overlap.

[0128] In some embodiments, when the first device selects a first subset of beams from the inference beam set, the first subset of beams can be determined based on first information. The first information may include information related to some or all of the beams in the inference beam set. The first information may include one or more of the following: beam priority in the inference beam set, historical performance data of the beams in the inference beam set, perceived quality of the beams in the inference beam set, coverage area of ​​the beams in the inference beam set, and network load of the cell where the first device is located.

[0129] As an example, the first beam subset can be determined based on the priority of some or all of the beams in the inference beam set. That is, the first device can perform the selection of the first beam subset based on beam priority.

[0130] As an example, beam priorities within an inference beamset can be determined based on the beam's importance in the network, beam quality, and / or the amount of resources required to monitor the beam. When beam priorities are correlated with these factors, it ensures that the first device rationally selects the first beamset. For instance, system resources can be used to monitor beams that have the greatest impact on network performance, reducing the monitoring burden on less important beams.

[0131] In the example above, the importance of a beam in the network can be determined based on the number of users covered by the beam or the user level.

[0132] In the example above, beam quality can include historical performance data. This historical performance data can include quality metrics such as the beam's historical RSRP and signal-to-interference-plus-noise ratio (SINR), and can also include other parameters representing beam quality, such as the ratio between the beam RSRP and a relevant threshold.

[0133] In the example above, the beam priority of the first beam in the inference beam set satisfies one or more of the following conditions: the beam priority of the first beam is positively correlated with the importance of the first beam; the beam priority of the first beam is positively correlated with the quality of the first beam; and the beam priority of the first beam is negatively correlated with the amount of resources used to monitor the first beam.

[0134] As one implementation, the priority P of the i-th beam in the inference beam set... i It can be: P i =U i *Q i / D i Among them, U i The importance of beamforming; Q i It is a quality indicator of the beam; D i It represents the amount of resources required to monitor the i-th beam.

[0135] As an example, the first beam set can be determined based on historical performance data of some or all of the beams in the inference beam set. Exemplarily, historical performance data can be used to construct a statistical model of beam performance, and the first beam set can be selected based on this model. Based on the statistical model, it is also possible to predict which beams' performance will change in the future. Therefore, based on the statistical model derived from historical performance data, the system can more effectively select the beams that need to be monitored.

[0136] In the example above, statistical models can prioritize monitoring beams whose performance is likely to fluctuate significantly in the future, thereby improving the effectiveness of monitoring. A commonly used statistical model is the auto-regressive moving average (ARMA) model. ARMA models can predict future beam states based on historical performance data.

[0137] As an implementation, the ARMA model can be represented as: y t =α1y t―1 +α2y t―2 +…+β1e t―1 +β2e t―2 +…;where y t This reflects the current performance of the beam (such as RSRP or SINR); e t This is the error term, representing the difference between the observed and predicted values; α * and β * These are model parameters, derived by fitting historical data.

[0138] As an example, the first beam set can be determined based on the quality of experience (QoE) corresponding to some or all of the beams in the inference beam set. In other words, the first device can select a monitoring dataset for performance monitoring based on user perception.

[0139] In the example above, the first device can prioritize monitoring beams or resources that have a greater impact on user experience. For instance, when users report a decrease in signal quality or application lag in a certain area, the system can immediately select the relevant beams for monitoring. This method directly optimizes the user experience. Therefore, the system can prioritize monitoring beams that cause a decline in user experience.

[0140] As an example, the first beam set can be determined based on the coverage area of ​​some or all of the beams in the inference beam set. The number of users and / or user level within the beam coverage area can be used to determine beam priority, or it can be used directly to determine the first beam set.

[0141] As an example, the first beam set can be determined based on the network load of the cell where the first device is located. The first beam set can be dynamically configured based on the current load. When the network load changes in the first monitoring instance, the first beam set can be adjusted in a timely manner.

[0142] As an example, the first beam set can also be determined based on one or more of the information in the first information mentioned above, such as beam priority, historical performance data, perceived quality, coverage area, and network load of the cell.

[0143] In some embodiments, the subset of beams used to transmit the reference signal for each monitoring instance can be dynamically adjusted based on the current network status and beam prediction requirements to ensure that the collected data is timely and representative.

[0144] In some embodiments, the first beam subset determined based on the first information may be one of multiple beam subsets in the inference beam set. Multiple beam subsets can be determined in the inference beam set based on the first information. Different monitoring levels can be set for each of the multiple beam subsets. When performing performance monitoring, beam subsets with different monitoring levels are selected according to monitoring requirements.

[0145] As an example, the priority of monitoring a beam or resource can be determined based on historical network data, user needs, or service type. For instance, high-priority beams might cover signal coverage in critical areas (such as high-density user areas or VIP user locations), while low-priority beams might be located in areas with lower or stable loads and thus receive lower monitoring levels.

[0146] In some embodiments, to fully cover all beams in the inference beam set, the first beam set can be determined through a polling mechanism related to the inference beam set. This polling mechanism can divide all beams in the inference beam set into multiple beam subsets corresponding to multiple monitoring instances based on a polling method. Therefore, to avoid the first device measuring the entire inference beam set (which may contain a large number of beams) in each monitoring cycle, the inference beam set can be divided into multiple subsets, and all beams can be gradually covered in multiple monitoring cycles using a polling method.

[0147] As an example, the inference beam set is divided into several non-overlapping subsets S1, S2, ..., S... n , where each subset S i It includes a subset of beam resources. For example, if the inference beam set contains 128 beams, it can be divided into 4 subsets, each containing 32 beams.

[0148] In the above embodiments, multiple beam subsets may correspond to the same reference signal or different reference signals. The reference signals corresponding to the multiple beam subsets can be cyclically traversed across the inference beam set based on a polling mechanism.

[0149] In some embodiments, after the inference beam set is divided, a polling period can be defined, during which each beam subset is measured sequentially. Each beam subset can correspond to a monitoring instance, such as a first monitoring instance. Multiple monitoring instances, including the first monitoring instance, constitute a polling period, i.e., the first polling period. Therefore, the polling mechanism can be used to monitor the performance of all beams in the inference beam set within the first polling period. That is, when the first beam set only includes a portion of the beams in the inference beam set, all beams in the inference beam set can be monitored through the polling mechanism and multiple monitoring instances.

[0150] In the above embodiments, different portions of the inference beamset are selected for monitoring in different performance monitoring instances to ensure that the entire inference beamset can be gradually covered over multiple monitoring periods. This method reduces the measurement complexity of each monitoring instance and ensures data integrity through continuous monitoring.

[0151] In the above embodiments, based on the polling mechanism, the performance monitoring of the first model can cover different subsets of the inference beam set within a certain time interval. This time interval can be referred to as the duration of a polling cycle.

[0152] As an example, the first beam set is a subset of the inference beam set, determined according to the polling mechanism of the inference beam set. The first beam subset can be one of a set of non-overlapping beam subsets defined for the inference beam set. Multiple non-overlapping beam subsets are used to implement polling monitoring of the entire inference beam set across multiple time instances.

[0153] As an example, the first polling cycle includes multiple monitoring instances, which include a first monitoring instance. Each of the multiple monitoring instances corresponds to a multiple beam subset that includes the first beam subset. Any two beam subsets in the multiple beam subsets have different beams.

[0154] As one implementation approach, it is assumed that multiple beams in the inference beam set can be divided into n beam subsets based on a polling mechanism. The beam subset selected by each monitoring instance can be: S i ={B i,1 B i,2 ,…,B i,ki}, i = 1, 2, ..., n; where S i B is the i-th beam subset; ki is the number of beams in the i-th beam subset;i,j It is the j-th beam in the i-th beam subset.

[0155] To facilitate understanding, the method for determining the first beam set according to the polling mechanism is illustrated below with reference to Figure 5. Similar to Figure 4, set B in Figure 5 represents the training beam set of the first model, and set A represents the inference beam set of the first model. Likewise, on the time axis, the training beam set of the first model is determined at time instance [t]. -3 ,t0,t +3 Transmission on [t]; the inference beamset of the first model is respectively in time instance [t] -2 , t -1 , t +1 , t +2 , t +4 Uploaded; Time instance [t] -2 , t -1 , t +1 , t +2 , t +4 This corresponds to multiple performance monitoring instances.

[0156] Unlike Figure 4, for each time instance corresponding to the inference beam set, the subset of beams used for performance monitoring is only the portion of the beams filled in with shaded areas. In other words, the inference beam set is different from the beam set of the performance monitoring instance.

[0157] See Figure 4 for a time instance [t] -2 , t -1 , t +1 , t +2 , t +4 This can be used as a polling cycle to cover all 9 beams in the entire inference beam set. Among them, the time instance [t] -2 , t -1 , t +1 , t +2 The corresponding beam subsets each include 2 beams, time instance t +4 The corresponding beam subset includes the remaining 1 beam.

[0158] In some embodiments, the first polling period may be greater than or equal to the first monitoring period. For example, when the first monitoring period includes one monitoring instance, the first polling period is greater than the first monitoring period. Conversely, when the first monitoring period includes multiple monitoring instances, the first polling period may be equal to the first monitoring period.

[0159] As an example, the duration of the first polling period is a positive integer multiple of the duration of the first monitoring period. The first monitoring period may include at least one monitoring instance. Optionally, this positive integer multiple can be determined based on the configuration of the RS resources and the size of the CSI reporting period.

[0160] As an example, the number of monitoring periods included in the first polling cycle is equal to the number of beam subsets divided into the inference beam set. For example, in the example in Figure 4, beam subset S1 is measured in the first monitoring cycle, beam subset S2 is measured in the second monitoring cycle, and so on, until all beam subsets in the inference beam set are covered, at which point a new round of polling begins.

[0161] In some embodiments, the value of the first polling period can be dynamically adjusted. When the first polling period is determined based on one or more pieces of information, the value of the first polling period can be adjusted based on changes in this information. The one or more pieces of information determining the first polling period may include: the moving speed of the first device; the network load of the cell where the first device is located; and the longest polling period of the cell where the first device is located.

[0162] As an example, the first polling period is linearly related to the moving speed of the first device. When the moving speed of the first device increases, the duration of the first polling period can be shortened to ensure that the channel state can be updated in a timely manner.

[0163] As an example, the first polling cycle is linearly related to the network load of the cell where the first device is located. When the network load of the cell where the first device is located increases, the duration of the first polling cycle can be shortened to accommodate more frequent data reporting needs.

[0164] As an example, the first polling period is related to the longest polling period of the cell where the first device is located. The first polling period is less than or equal to the longest polling period. For example, the first polling period can be set to 80% of the longest polling period.

[0165] As an example, the first polling cycle is also related to the monitoring cycle of the first model or the reporting cycle of the monitoring report. The duration of the first polling cycle is an integer multiple of the duration of the first monitoring cycle, or the duration of the first polling cycle is an integer multiple of the duration of the reporting cycle.

[0166] As an example, the first polling period can be dynamically adjusted based on the first device's movement speed, network load, and the longest polling period. For instance, the value of the first polling period could be the longest polling period minus the corresponding duration for movement speed and network load.

[0167] As one implementation, the first polling period T can be: T = T max ―α*v―β*L; where α and β are weighting coefficients, T max The longest polling period is represented by v, where v represents the impact of movement speed on the first polling period, and L represents the impact of network load on the first polling period.

[0168] Optionally, T and T max Both represent the length of the polling period, which can be in seconds. Where T... max This can be pre-configured by the network. For example, the network can determine this value based on cell size and whether the cell covers high-speed mobile terminal devices.

[0169] Optionally, the two weighting coefficients can be used to balance the impact of the first device's movement speed and network load on the first polling cycle.

[0170] Optionally, v can convert the movement speed of the first device into the length of time that affects the first polling cycle.

[0171] Optionally, L can convert the cell's current load into the duration affecting the first polling cycle, which can be calculated using the current network congestion rate.

[0172] In some embodiments, the number of beams in each beam subset, including the first beam subset, can also be dynamically adjusted in the polling mechanism. As an example, by combining the dynamic adjustment of the first polling period T and the beam subset size S, a joint optimization model can be obtained, thereby ensuring that network performance is maximized while meeting the needs of different scenarios.

[0173] As one implementation approach, assuming that all beams in the inference beam set are gradually covered over multiple cycles, the total delay T for covering all beams is... delay It can be represented as: Where: N is the total number of beams in the inference beamset; S is the dynamically adjusted subset size; T monitor For each monitoring cycle.

[0174] In the above implementation, the network (NW) can dynamically adjust S and T so that T delay To achieve performance optimization, the goal is to minimize the impact while maintaining monitoring accuracy.

[0175] The above sections, with reference to Figures 4 and 5, introduced two methods for determining the first beam subset. Regardless of whether the first beam subset is determined based on a polling mechanism, the number of beams in the first beam subset can be dynamically adjusted to meet actual needs. To improve the monitoring accuracy of the first model, during dynamic adjustment, the number of beams in the first beam subset can be related to one or more of the following: the prediction accuracy of the first model; the network load of the cell where the first device is located; and the minimum number of beams in the first beam subset.

[0176] As an example, the number of beams in the first beam subset is linearly related to the prediction accuracy of the first model. When higher prediction accuracy is required, the number of beams can be increased to obtain more beam data, thereby improving prediction accuracy.

[0177] As an example, the number of beams in the first beam subset is linearly related to the network load of the cell where the first device is located. When the network load increases, the number of beams can be reduced to decrease the amount of measurement data and reduce network pressure.

[0178] As an example, the number of beams in the first beam subset is related to the minimum number of beams in the first beam subset. This minimum number of beams can be configured by the network. The number of beams in the first beam subset is greater than or equal to this minimum number of beams.

[0179] As an example, the number of beams in the first beam subset can be dynamically adjusted based on the prediction accuracy of the first model, the cell network load, and the minimum number of beams. For instance, the number of beams in the first beam subset can be the value obtained by adding the minimum number of beams to the prediction accuracy parameter and subtracting the network load parameter.

[0180] As one implementation, when the first beam subset is the i-th beam subset among multiple beam subsets, where i is an integer greater than or equal to 1, the number of beams S in the first beam subset is... i It can be: Where γ and δ are weighting coefficients, The minimum number of beams is given by E, which represents the impact of prediction accuracy on the number of beams, and L represents the impact of network load on the number of beams.

[0181] Optionally, E can translate the required prediction accuracy into relevant parameters. For example, a target threshold for prediction accuracy, between 0 and 1, might be required.

[0182] Alternatively, L can convert the cell’s current load into relevant parameters that affect the number of beams, which can be calculated using the current network congestion rate.

[0183] Optionally, S i It can also represent the size of the beam subset for each CSI reporting period, i.e., the number of beams measured in each reporting period.

[0184] In some embodiments, the first reference signal may be sent directly by the second device or sent by the second device upon request. As an example, the first device may send a first request to the second device to request the first reference signal for performance monitoring.

[0185] In some embodiments, the dedicated resource set of the first beam set or the first reference signal is determined according to a dedicated CSI report configuration. When configuring the dedicated resource set for monitoring and the report configuration for monitoring in the dedicated CSI report configuration for monitoring, it is necessary to identify the connection between the resource sets RS for monitoring. The first device can also measure the monitored beam according to this connection and report the measurement results.

[0186] As an example, a dedicated resource set used for performance monitoring is the RS resource set configured for the network, which can also be called the monitoring RS resource set.

[0187] As an example, the current beam measurement and reporting framework allows configuring an RS resource set for a first device. This reporting configuration can also include a CSI-RS resource set corresponding to the first beam set (inference beam set or a portion thereof).

[0188] As an example, the first beam set can be associated with the same identity (ID) as the inference beam set or training beam set of the first model to facilitate measurement by the first device. When the associated IDs of the inference beam set and the training beam set are the same, the first beam set is associated with the same ID as both the inference beam set and the training beam set. That is, the first associated identity corresponding to the first beam set is the same as the associated identity of the inference beam set and / or the training beam set of the first model. For example, set A, set B, and the monitoring set are configured with the same associated ID. This will be illustrated exemplarily below with reference to Figures 6 to 8.

[0189] As an implementation approach, within the CSI framework, the association ID can be sent and tagged along with the CSI Reference Signal (CSI-RS), enabling AI / ML models to utilize the same association ID to determine beam selection during the measurement, inference, and monitoring phases. For example, the association ID can be indicated through a CSI-RS resource index. Each CSI-RS resource has its unique ID, which the AI / ML model can use to identify CSI-RS measurements associated with different beam directions.

[0190] As an example, the inference beam set and the training beam set are associated with an ID, while the first beam set is not associated with an ID.

[0191] In some embodiments, when monitoring the inference beam set using a polling method (i.e., a polling mechanism), the resource configuration and reporting configuration for the first beam set need to take into account the polling period in order to cover all beams in the inference beam set.

[0192] As an example, when the first beam set is one of multiple beam subsets determined by a polling mechanism for the inference beam set, the CSI report configuration needs to ensure effective coverage of all beams in the inference beam set while maintaining monitoring resource efficiency. For instance, multiple csi-ReportConfig configurations can be used to complete all configurations within a single polling cycle. Similarly, in configuring RS resources, it's necessary to ensure reasonable resource allocation for each beam subset to support effective measurement.

[0193] As an example, multiple beam subsets of the first beam set correspond one-to-one with multiple resource configurations within the first polling cycle. Each resource configuration can be used to configure one RS resource set. Therefore, multiple beam subsets can each correspond one-to-one with multiple RS resource sets. For example, an independent RS resource set can be configured for each beam subset Si to ensure that the first device can measure the beams in that beam subset within a polling cycle. This will be illustrated exemplarily below with reference to Figure 9.

[0194] As one implementation, each beam subset Si in the inference beam set can be configured with an independent CSI-RS resource or a demodulation reference signal (DMRS) of the physical downlink control channel (PDCCH) for performance monitoring within a polling cycle. Within each beam subset, key beams can also be selected as measurement targets for RS resources to ensure that the reported Top-K beam information fully reflects the overall performance of that beam subset. Optionally, beams with historically good performance or those with high signal strength in the current environment can be prioritized for measurement.

[0195] In some embodiments, the network can dynamically adjust the configuration of RS resources according to real-time needs to improve the flexibility of RS resource configuration. For example, in high mobility scenarios, the network can shorten the polling cycle and increase the RS resource density to improve the timeliness and accuracy of measurements.

[0196] As an example, multiple beam subsets of the first beam set correspond to M resource configurations, where M is the number of reporting cycles within the first polling period. For instance, if the first polling period includes two reporting cycles, resources for multiple beam subsets can be configured simultaneously using a single resource configuration. In other words, resources can be configured for multiple beam subsets at the same time. This will be illustrated later with reference to Figure 10.

[0197] In some embodiments, AI / ML models can predict future locations based on a user's historical movement trajectory, thereby dynamically adjusting the CSI reporting cycle and subset size. For example, when it is predicted that a user will enter a high-mobility area, RS resources are configured intensively, with multiple monitoring subsets configured in one resource configuration, which can shorten the polling cycle in advance.

[0198] In some embodiments, the AI / ML model can dynamically adjust subsequent polling configurations based on real-time CSI report results and the AI / ML algorithm. For example, if significant fluctuations in beam signal strength are detected in the current subset, the subset can be expanded in the next cycle to ensure coverage of more beams. By dynamically adjusting the polling cycle length, CSI report triggering mechanism, and subset size, the network can flexibly optimize the CSI report configuration under different network conditions and user needs, achieving efficient and accurate performance monitoring. This dynamic adjustment strategy ensures full coverage of all beams in the inference beam set while optimizing resource utilization to meet performance requirements in different scenarios.

[0199] Referring again to Figure 3, in step S320, the first device monitors the performance of the first model based on the measurement results of the first beam set.

[0200] The measurement results of the first beam set may include direct measurement parameters of the first beam set, and may also include performance parameters of the first model determined based on the measurement parameters. Direct measurement parameters include, for example, RSRP. Performance parameters, as described above, will not be repeated here.

[0201] In some embodiments, the performance parameters of the first model can be determined by comparing prediction results with actual measurement results. These results can be Top-1 or Top-K beams in the inference beam set, or parameters such as RSRP difference. For example, when the first beam set is an inference beam set or multiple beam subsets including the first beam set cover the inference beam set, comparing the predicted Top-K beams with the measured Top-K beams can determine whether the first model has a problem. In other words, when the RS of performance monitoring covers the entire inference beam set, the inference error can be measured by comparing the Top-1 / Top-K beams predicted by the model with the actual optimal beams based on measurements. Similarly, comparing the prediction results with the measured values ​​of the resource set / resources obtained from performance monitoring can yield the prediction accuracy of the Top-1 or Top-K beams.

[0202] As an example, the accuracy of the first model is determined based on the L1-RSRP difference information between actual L1-RSRP measurements of one or more predicted optimal beams and L1-RSRP measurements from resource sets / resources used for monitoring.

[0203] As an example, the prediction accuracy of Top-1 beams with margin can be evaluated by measuring the RSRP difference between the predicted beam and the optimal beam. When this RSRP difference is less than a set threshold, the prediction is considered successful, and the first model is accurate.

[0204] As an example, when the highest measured L1-RSRP of a Top-K predicted beam is within a margin of the highest measured L1-RSRP of a beam in the first beam set, the best beam in the Top-K predicted beams is used. This scenario can occur after the first device measures the Top-K predicted beams and finds the best beam with the highest L1-RSRP. As long as the L1-RSRP of the best beam in the Top-K predicted beams is within a set or calculated margin compared to the L1-RSRP of the best beam in the first beam set, the performance degradation caused by using the predicted Top-K beam or the best beam compared to the actual best beam is considered tolerable, and this is interpreted as a success event.

[0205] As an example, the performance parameters of the first model can be represented by the monitoring margin of the first model. For instance, the monitoring margin of the first model can be determined based on the actual number of monitoring sessions and the quality difference between the inference beam set and the corresponding beam in the first beam set.

[0206] As one implementation method, the monitoring margin can be: Where N is the actual number of monitoring times, ε is the margin threshold, and RSRP(b) i Let be the predicted RSRP of beam b in the inference beam set during the i-th monitoring. The measured RSRP of beam b in the first beam set during the i-th monitoring is given.

[0207] Optionally, ε is a preset margin threshold, which can be used to determine whether the RSRP difference between the predicted beam and the actual beam is acceptable.

[0208] Optionally, 1(·) is an indicator function. It returns 1 if the difference between the predicted result and the actual measurement result is within the preset margin range, and 0 otherwise. Multiple optimal prediction beams can use the same margin value, or different margin sizes can be set for each.

[0209] Optionally, beam b can be the best predicted beam or a plurality of superior beams in the first beam set.

[0210] In some embodiments, the first beam set includes the predicted Top-K beams from the inference beam set. Assume the total number of monitored instances is N. instance The actual number of monitored instances is N. monitorThe performance monitoring accuracy related to Top-K beam prediction is N. monitor / N instance .

[0211] In some embodiments, after performance monitoring, the first device may also send a monitoring report of the first model to the second device. For example, for network device-side performance monitoring at the first device used for beam prediction, the network may need to configure / indicate a set of RS resources as monitoring RS resources, and the first device may measure these monitoring RS resources and send a monitoring report to the network.

[0212] In some embodiments, the monitoring report may include performance parameters of the first model. As described above, the performance parameters of the first model include the prediction accuracy and / or the prediction error of the first model. For example, based on report measurements and inference-related reports corresponding to the monitored RS resources, performance metrics or relevant KPIs (e.g., beam prediction accuracy, RSRP difference, etc.) can be calculated. The calculated performance metrics or relevant KPIs can be used to evaluate the operability of CSI reports related to beam prediction.

[0213] As an example, the CSI reporting framework can be used to configure and monitor RS resource sets and obtain beam measurement reports corresponding to those RS resource sets. These beam measurement reports can include CSI-RS resource indicators (CRI) or parameters such as RSRP and SINR for the measured RS resources.

[0214] As an example, a network can use different CSI reports (monitoring reports) to obtain beam measurements / reports for monitored RS resource sets within the CSI reporting framework. The report content can include the L1-RSRP and RS index of the Top-K beams in the monitored RS group.

[0215] In some embodiments, the first device can send monitoring reports based on different CSI reporting mechanisms to support network-side performance monitoring. As an example, the network can use different time behavior configurations to associate the CSI reporting framework with the monitoring reports. These time behavior configurations include aperiodic (AP), periodic (P), or semi-persistent (SP). Correspondingly, the first device can configure and send periodic, aperiodic, and semi-persistent beam reports.

[0216] As an example, based on the relevant L1-CSI reporting framework, periodic, non-periodic, and semi-persistent signal reporting mechanisms can be used to send monitoring reports to achieve accuracy reporting in model monitoring or Top-K beam measurements.

[0217] In some embodiments, the monitoring report for the first model is configured using a first CSI report configuration. As an example, the first CSI report configuration may be the same as the CSI report configuration corresponding to the inference beamset. As an example, the first CSI report configuration may also be different from the CSI report configuration corresponding to the inference beamset.

[0218] As an example, when the first beam set is the inference beam set, the first CSI report configuration is the same as the CSI report configuration corresponding to the inference beam set. That is, the monitoring RS resource set is equal to the resource set of the inference beam set. In this scenario, model inference and model monitoring use the same RS resource set.

[0219] As an example, when the first beam set is a subset of the inference beam set, the first CSI report configuration is the same as the CSI report configuration corresponding to the inference beam set. That is, the first beam set only includes a portion of the inference beam set. In this scenario, the same CSI report configuration is used for both model monitoring and model inference.

[0220] As an example, when the first beam set is an inference beam set, the configuration of the first CSI report differs from the CSI report configuration corresponding to the inference beam set. In other words, the resource sets of the monitoring RS resource set and the inference beam set are configured / indicated separately.

[0221] As an example, when the first beam set is a subset of the inference beam set, the first CSI report configuration is different from the CSI report configuration corresponding to the inference beam set.

[0222] As an example, regardless of whether the first beam set is a subset of or equal to the inference beam set, the first CSI report configuration and the CSI report configuration of the inference beam set can be configured separately, or the same CSI report configuration can be used.

[0223] As an example, the first CSI report configuration also considers the configuration and indication of the monitoring RS resource set under different scenarios. The monitoring RS resource set corresponds to the first beam set. The setting of the first beam set can be configured explicitly or implicitly. For example, when the first beam set is a subset of the inference beam set, the network can configure / indicate the CSI-RS resource set corresponding to the beam subset of the inference beam set as the monitoring RS resource set, and further configure the timeline and reporting volume in the monitoring RS resource set.

[0224] In some embodiments, the first device may send a monitoring report of the first model based on a performance monitoring reporting cycle. The performance monitoring reporting cycle may be determined based on different CSI reporting mechanisms.

[0225] As an example, the reporting cycle can be the same as the monitoring cycle. Within a polling cycle, each reporting cycle can measure only a subset of beams, significantly reducing the measurement burden on the first device and saving resources. Through multiple reporting cycles, the first device can gradually cover all beams of the inference beam set through polling, ensuring that all beams are measured. As mentioned earlier, the network can dynamically adjust the length of the polling cycle and the size of the subset according to network conditions, ensuring the effectiveness of performance monitoring in different scenarios.

[0226] In some embodiments, in a performance monitoring scenario based on a polling mechanism, the first configuration of the monitoring report can configure the beam subset corresponding to the polling mechanism, and can support phased monitoring reports. Optionally, the first configuration may include one or more of the following: the trigger type of the monitoring report, the beam subset indicator, and the content of the monitoring report.

[0227] As an example, when the first configuration is a CSI report configuration based on the CSI framework, the first configuration can include the CSI report trigger type, beam subset indication, and CSI report content. The three contents are detailed below.

[0228] For CSI report triggering types, periodic, non-periodic, or semi-persistent CSI report triggering mechanisms can be used to ensure that the primary device can periodically report monitoring data. CSI reports for the corresponding beam subset are activated within different polling cycles.

[0229] For beam subset indication, the network can configure multiple CSI reports within each polling cycle to specify the beam subset that needs to be measured. For example, the network can configure beam subset S1 as the measurement target of the CSI report in the first monitoring cycle of each polling cycle, and then configure subset S2 in the next monitoring cycle, as shown in Figure 9.

[0230] The CSI report content can include the L1-RSRP values ​​and RS indices of the Top-K beams within each beam subset. This allows the network to identify the best-performing beam in each subset and calculate the accuracy of its model predictions based on this information.

[0231] The preceding text, with reference to Figures 3 to 5, introduced the method for determining the first beam set used for performance monitoring. As mentioned earlier, the first beam set can be configured with the same association ID as the training beam set and the inference beam set. The following section uses the CSI framework configuration as an example, with reference to Figures 6 to 8, to illustrate the configuration method of the first beam set association ID.

[0232] The beam sets and associated IDs in Figures 6 to 8 are configured through the CSI report configuration (csi-ReportConfig). Throughout the AI / ML-based beam management process, model training, model inference, and model detection (monitoring) can be configured with associated IDs based on the CSI-RS resource index. For example, in the CSI framework, the inference beam set and the training beam set are configured as different CSI resource sets in csi-ResourceConfig, but each is associated with an ID. Specifically, the resource set for the inference beam set is configured through csi-ResourceSetA, the resource set for the training beam set is configured through csi-ResourceSetB, and the resource set for the monitoring set (the first beam set) is configured through csi-ResourcemonitorSet. The following sections, with reference to the accompanying figures, describe three different configuration methods.

[0233] Referring to Figure 6, in csi-ReportConfig, the inference beam set, training beam set, and first beam set corresponding to different CSI resource sets are configured in the same CSI resource configuration (csi-ResourceConfig). The three beam sets share the same association ID, and this association ID is also in this CSI resource configuration.

[0234] Compared to Figure 6, although the inference beam set, training beam set, first beam set, and associated ID in Figure 7 are all in the same CSI resource configuration, the inference beam set, training beam set, and first beam set correspond to different CSI resource configuration indices (csi-ResourceConfigId). In other words, the inference beam set, training beam set, and first beam set are configured separately based on different CSI resource configuration indices.

[0235] Compared to Figure 7, although the inference beam set, training beam set, and first beam set in Figure 8 are also configured separately, the associated IDs corresponding to different beam sets are configured in different CSI resource configuration indexes. In the method shown in Figure 8, since the associated IDs are configured separately in different csi-ReportConfigIds, it can satisfy scenarios where the associated IDs are different. Furthermore, even if the associated IDs are different, no additional csi-ResourceConfig is required, which is more beneficial in terms of reporting configuration overhead.

[0236] In Figures 7 and 8, each csi-ReportConfigId is constructed in a csi-ReportConfig. It should be noted that different csi-ReportConfigIds can also be configured in different csi-ReportConfigs.

[0237] As mentioned earlier, when monitoring performance using multiple beam subsets generated based on a polling mechanism, the RS resource set can correspond one-to-one with multiple beam subsets, or it can be flexibly configured according to the actual situation. The following uses the CSI framework in Figure 6 as an example, combined with Figures 9 and 10, to illustrate the two configuration methods.

[0238] Referring to Figure 9, during the first polling cycle, the inference beam set is divided into n beam subsets, namely sets S1, S2, ..., S... n The CSI resource configuration for the first polling cycle includes n csi-ReportConfigs. Each of the n csi-ReportConfigs corresponds one-to-one with one of the n beam subsets.

[0239] Compared to Figure 9, the inference beam set in Figure 10 is also divided into n beam subsets, but the first polling cycle is determined by two reporting cycles. In this scenario, two csi-ReportConfigs are needed to configure resources for each of the multiple monitoring subsets. Referring to Figure 10, the two csi-ReportConfigs correspond to the beam subset sets {S1,…,S…}, respectively. k}, {S k+1 ,S k+2 ,…,S n}

[0240] Optionally, the two CSI-ReportConfigs in Figure 10 can be two beam subsets {S1,…,S1}, respectively. k}, {S k+1 ,S k+2 ,…,S n Set up a corresponding RS to improve the accuracy of performance monitoring.

[0241] The determination method and related configuration of the first beam set used for performance monitoring have been described above with reference to Figures 3 to 10. The first beam set is used to transmit a dedicated RS for performance monitoring; the related configuration includes the configuration of RS resources and monitoring reports. To more clearly illustrate the application of the embodiments of this application, the following example uses a UE as the first device and an eNB as the second device, with reference to Figure 11, to illustrate an example of the first device requesting a dedicated reference signal from the second device and sending a monitoring report.

[0242] In step S1110, the UE sends UE capability indication information to the base station. For example, the UE sends its own AI / ML capability indication to the base station.

[0243] In step S1120, the UE sends a request (first request) to the base station to request a dedicated RS for performance monitoring in order to perform performance testing on the model.

[0244] In step S1130, the base station sends a dedicated RS for transmission based on the UE's request to support performance detection.

[0245] In step S1140, the UE calculates the detection KPIs or determines the event triggering conditions based on the received performance RS.

[0246] In step S1150, the UE sends the KPIs detection results or trigger event information (information about monitoring KPIs or event occurrence) to the base station based on the performance detection and evaluation results.

[0247] In step S1160, the base station receives KPI detection result information from the UE to evaluate model performance (AI / ML performance). Optionally, the base station can perform performance evaluation and auxiliary monitoring of the model on the UE side, or it can perform performance evaluation of its own model. Conversely, in model monitoring, the UE can perform performance evaluation of its own model or auxiliary monitoring of the model on the base station side.

[0248] In step S1170, based on the evaluation results, the base station will notify the UE-side AI / ML and LCM (life cycle management, LCM) operation information. This information covers aspects such as the model's status, configuration, monitoring, and updates. This information ensures that the UE can maintain optimal performance and consistency during model usage.

[0249] In step S1180, the UE performs LCM operation at the UE based on the LCM information from the base station, which includes activating, generalizing, switching, falling back, or updating the AM / ML model on the UE side.

[0250] In step S1190, the UE will notify the base station of relevant information about the LCM operation it performed, such as the results after the LCM was implemented, based on the updated results.

[0251] As shown in Figure 11, the embodiments of this application can be applied to the monitoring of model performance by terminal devices and network devices. The transmission of LCM information facilitates timely understanding of the relevant model status by terminal devices and network devices, thereby improving the accuracy of the model.

[0252] The method embodiments of this application have been described in detail above with reference to Figures 1 to 11. The apparatus embodiments of this application are described in detail below with reference to Figures 12 to 14. It should be understood that the descriptions of the apparatus embodiments correspond to the descriptions of the method embodiments; therefore, any parts not described in detail can be referred to the foregoing method embodiments.

[0253] Figure 12 is a schematic block diagram of a wireless communication apparatus according to an embodiment of this application. The apparatus 1200 can be any of the first devices described above. The first device can be a terminal device. The apparatus 1200 shown in Figure 12 includes a transceiver unit 1210 and a processing unit 1220.

[0254] The transceiver unit 1210 can be used to receive a first reference signal, which is transmitted through a first beam set.

[0255] The processing unit 1220 can be used to monitor the performance of a first model based on the measurement results of a first beam set; wherein, the first reference signal corresponds to a first monitoring instance, and the first beam set is determined based on the inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes some beams in the inference beam set.

[0256] Optionally, the first beam set is a first subset of the inference beam set, and the first beam subset is determined according to first information, which includes one or more of the following: beam priority in the inference beam set; historical performance data of the beams in the inference beam set; perceived quality of the beams in the inference beam set; coverage area of ​​the beams in the inference beam set; and network load of the cell where the first device is located.

[0257] Optionally, the first beam subset is determined based on beam priority, and the beam priority of the first beam in the inference beam set satisfies one or more of the following conditions: the beam priority of the first beam is positively correlated with the importance of the first beam; the beam priority of the first beam is positively correlated with the quality of the first beam; the beam priority of the first beam is negatively correlated with the amount of resources used to monitor the first beam.

[0258] Optionally, performance monitoring includes multiple monitoring instances, including a first monitoring instance and a second monitoring instance, the second monitoring instance corresponding to a second beam subset, and the first beam subset having at least one beam in the second beam subset.

[0259] Optionally, the first beam set is a first subset of the inference beam set. The first beam subset is determined according to a polling mechanism related to the inference beam set. The polling mechanism is used to monitor the performance of all beams in the inference beam set during the first polling period.

[0260] Optionally, the first polling cycle includes multiple monitoring instances, the multiple monitoring instances include a first monitoring instance, the multiple monitoring instances correspond to multiple beam subsets including the first beam subset, and the beams in any two beam subsets of the multiple beam subsets are different.

[0261] Optionally, the duration of the first polling period is a positive integer multiple of the duration of the first monitoring period, and the first monitoring period includes at least one monitoring instance.

[0262] Optionally, the first polling period is related to one or more of the following: the moving speed of the first device; the network load of the cell where the first device is located; and the longest polling period of the cell where the first device is located.

[0263] Optionally, the first polling period is: T = T max ―α*v―β*L; where α and β are weighting coefficients, T max The longest polling period is represented by v, where v represents the impact of movement speed on the first polling period, and L represents the impact of network load on the first polling period.

[0264] Optionally, the number of beams in the first beam subset is related to one or more of the following: the prediction accuracy of the first model; the network load of the cell where the first device is located; and the minimum number of beams in the first beam subset.

[0265] Optionally, when the first beam subset is the i-th beam subset among multiple beam subsets, where i is an integer greater than or equal to 1, the number of beams in the first beam subset is: Where γ and δ are weighting coefficients, The minimum number of beams is given by E, which represents the impact of prediction accuracy on the number of beams, and L represents the impact of network load on the number of beams.

[0266] Optionally, the first beam set corresponds to a first association identifier, and the first association identifier is the same as the association identifier of the inference beam set and / or the training beam set of the first model.

[0267] Optionally, the first beam set is one of multiple beam subsets determined by the polling mechanism for the inference beam set; the multiple beam subsets correspond one-to-one with multiple resource configurations in the first polling cycle, or the multiple beam subsets correspond to M resource configurations, where M is the number of reporting cycles in the first polling cycle.

[0268] Optionally, the transceiver unit 1210 is further configured to send a first request to the second device, the first request being used to request a first reference signal; wherein the first reference signal is at least one of a plurality of dedicated reference signals, the dedicated reference signals being used to monitor the performance of the first model.

[0269] Optionally, the transceiver unit 1210 is also used to send a monitoring report of the first model based on a performance monitoring reporting cycle; wherein the monitoring report includes the performance parameters of the first model, and the performance parameters of the first model include the prediction accuracy of the first model and / or the prediction error of the first model.

[0270] Optionally, the monitoring report of the first model is configured through the first CSI report configuration; the first CSI report configuration is the same as the CSI report configuration corresponding to the inference beam set, or the first CSI report configuration is different from the CSI report configuration corresponding to the inference beam set.

[0271] Optionally, the performance parameters of the first model are represented by the monitoring margin of the first model. The monitoring margin is determined based on the actual number of monitoring operations and the quality difference between the inference beam set and the corresponding beam in the first beam set.

[0272] Optionally, the monitoring margin is: Where N is the actual number of monitoring times, ε is the margin threshold, and RSRP(b) i Let be the predicted RSRP of beam b in the inference beam set during the i-th monitoring. The measured RSRP of beam b in the first beam set during the i-th monitoring is given.

[0273] Optionally, the first model is an artificial intelligence or machine learning model.

[0274] Figure 13 is a schematic block diagram of another device for wireless communication according to an embodiment of this application. The device 1300 can be any of the second devices described above. The second device is a network device or a terminal device. The device 1300 shown in Figure 13 includes a transceiver unit 1310.

[0275] The transceiver unit 1310 can be used to transmit a first reference signal, which is transmitted through a first beam set; wherein, the first reference signal corresponds to a first monitoring instance, the measurement results of the first beam set are used to monitor the performance of a first model, and the first beam set is determined based on the inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes a portion of the beams in the inference beam set.

[0276] Optionally, the first beam set is a first subset of the inference beam set, and the first beam subset is determined according to first information, which includes one or more of the following: beam priority in the inference beam set; historical performance data of the beams in the inference beam set; perceived quality of the beams in the inference beam set; coverage area of ​​the beams in the inference beam set; and network load of the cell corresponding to the second device.

[0277] Optionally, the first beam subset is determined based on beam priority, and the beam priority of the first beam in the inference beam set satisfies one or more of the following conditions: the beam priority of the first beam is positively correlated with the importance of the first beam; the beam priority of the first beam is positively correlated with the quality of the first beam; the beam priority of the first beam is negatively correlated with the amount of resources used to monitor the first beam.

[0278] Optionally, performance monitoring includes multiple monitoring instances, including a first monitoring instance and a second monitoring instance, the second monitoring instance corresponding to a second beam subset, and the first beam subset having at least one beam in the second beam subset.

[0279] Optionally, the first beam set is a first subset of the inference beam set. The first beam subset is determined according to a polling mechanism related to the inference beam set. The polling mechanism is used to monitor the performance of all beams in the inference beam set during the first polling period.

[0280] Optionally, the first polling cycle includes multiple monitoring instances, the multiple monitoring instances include a first monitoring instance, the multiple monitoring instances correspond to multiple beam subsets including the first beam subset, and the beams in any two beam subsets of the multiple beam subsets are different.

[0281] Optionally, the duration of the first polling period is a positive integer multiple of the duration of the first monitoring period, and the first monitoring period includes at least one monitoring instance.

[0282] Optionally, the first polling period is related to one or more of the following: the moving speed of the first device receiving the first reference signal; the network load of the cell corresponding to the second device; and the longest polling period of the cell corresponding to the second device.

[0283] Optionally, the first polling period is: T = T max ―α*v―β*L; where α and β are weighting coefficients, T max The longest polling period is represented by v, where v represents the impact of movement speed on the first polling period, and L represents the impact of network load on the first polling period.

[0284] Optionally, the number of beams in the first beam subset is related to one or more of the following: the prediction accuracy of the first model; the network load of the cell corresponding to the second device; and the minimum number of beams in the first beam subset.

[0285] Optionally, when the first beam subset is the i-th beam subset among multiple beam subsets, where i is an integer greater than or equal to 1, the number of beams in the first beam subset is: Where γ and δ are weighting coefficients, The minimum number of beams is given by E, which represents the impact of prediction accuracy on the number of beams, and L represents the impact of network load on the number of beams.

[0286] Optionally, the first beam set corresponds to a first association identifier, and the first association identifier is the same as the association identifier of the inference beam set and / or the training beam set of the first model.

[0287] Optionally, the first beam set is one of multiple beam subsets determined by the polling mechanism for the inference beam set; the multiple beam subsets correspond one-to-one with multiple resource configurations in the first polling cycle, or the multiple beam subsets correspond to M resource configurations, where M is the number of reporting cycles in the first polling cycle.

[0288] Optionally, the transceiver unit 1310 is further configured to receive a first request sent by the first device, the first request being used to request a first reference signal; wherein the first reference signal is at least one of a plurality of dedicated reference signals, the dedicated reference signals being used to monitor the performance of the first model.

[0289] Optionally, the transceiver unit 1310 is further configured to receive a monitoring report of the first model based on a performance monitoring reporting cycle; wherein the monitoring report includes performance parameters of the first model, and the performance parameters of the first model include the prediction accuracy of the first model and / or the prediction error of the first model.

[0290] Optionally, the monitoring report of the first model is configured through the first CSI report configuration; the first CSI report configuration is the same as the CSI report configuration corresponding to the inference beam set, or the first CSI report configuration is different from the CSI report configuration corresponding to the inference beam set.

[0291] Optionally, the performance parameters of the first model are represented by the monitoring margin of the first model. The monitoring margin is determined based on the actual number of monitoring operations and the quality difference between the inference beam set and the corresponding beam in the first beam set.

[0292] Optionally, the monitoring margin is: Where N is the actual number of monitoring times, ε is the margin threshold, and RSRP(b) i Let be the predicted RSRP of beam b in the inference beam set during the i-th monitoring. The measured RSRP of beam b in the first beam set during the i-th monitoring is given.

[0293] Optionally, the first model is an artificial intelligence or machine learning model.

[0294] Figure 14 is a schematic diagram of the structure of a communication device according to an embodiment of this application. The dashed lines in Figure 14 indicate that the unit or module is optional. This device 1400 can be used to implement the methods described in the above method embodiments. Device 1400 can be a chip, a terminal device, or a network device.

[0295] Apparatus 1400 may include one or more processors 1410. The processor 1410 may support apparatus 1400 in implementing the methods described in the preceding method embodiments. The processor 1410 may be a general-purpose processor or a special-purpose processor. For example, the processor may be a central processing unit (CPU). Alternatively, the processor may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0296] The apparatus 1400 may further include one or more memories 1420. The memories 1420 store a program that can be executed by the processor 1410, causing the processor 1410 to perform the methods described in the preceding method embodiments. The memories 1420 may be independent of the processor 1410 or integrated within the processor 1410.

[0297] The device 1400 may also include a transceiver 1430. The processor 1410 can communicate with other devices or chips via the transceiver 1430. For example, the processor 1410 can send and receive data with other devices or chips via the transceiver 1430.

[0298] This application also provides a computer-readable storage medium for storing a program. This computer-readable storage medium can be applied to a terminal device or network device provided in this application embodiment, and the program causes a computer to execute the methods performed by the terminal device or network device in the various embodiments of this application.

[0299] The computer-readable storage medium can be any available medium that a computer can read, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., digital video discs, DVDs), or semiconductor media (e.g., solid-state disks, SSDs), etc.

[0300] This application also provides a computer program product. The computer program product includes a program. This computer program product can be applied to a terminal device or network device provided in the embodiments of this application, and the program causes a computer to execute the methods performed by the terminal device or network device in the various embodiments of this application.

[0301] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0302] This application also provides a computer program. This computer program can be applied to a terminal device or network device provided in this application, and the computer program causes the computer to execute the methods performed by the terminal or network device in various embodiments of this application.

[0303] In this application, the terms "system" and "network" are used interchangeably. Furthermore, the terminology used in this application is only for explaining specific embodiments of the application and is not intended to limit the application. The terms "first," "second," "third," and "fourth," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. In addition, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. In embodiments of this application, determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0304] In the embodiments of this application, the term "instruction" can be a direct instruction, an indirect instruction, or an indication of a relationship. For example, A instructing B can mean that A directly instructs B, such as B being able to obtain information through A; it can also mean that A indirectly instructs B, such as A instructing C, so B can obtain information through C; or it can mean that there is a relationship between A and B.

[0305] In the embodiments of this application, the term "correspondence" may indicate a direct or indirect correspondence between two things, or an association between two things, or a relationship such as instruction and being instructed, configuration and being configured.

[0306] In the embodiments of this application, "predefined" or "preconfigured" can be implemented by pre-storing corresponding codes, tables, or other means that can be used to indicate relevant information in the device (e.g., including terminal devices and network devices). This application does not limit the specific implementation method. For example, predefined can refer to what is defined in the protocol.

[0307] In the embodiments of this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0308] In the embodiments of this application, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0309] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0310] The units described as separate components may or may not be physically separate. 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 the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0311] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0312] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for wireless communication, characterized in that, include: The first device receives a first reference signal, which is transmitted through a first beam set; The first device monitors the performance of the first model based on the measurement results of the first beam set; Wherein, the first reference signal corresponds to the first monitoring instance, and the first beam set is determined according to the inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes some beams in the inference beam set.

2. The method according to claim 1, characterized in that, The first beam set is a first subset of the inference beam set, and the first beam subset is determined based on first information, which includes one or more of the following: Beam priority in the inference beam set; Historical performance data of the beams in the inference beam set; The sensing quality of the beams in the inference beam set; The coverage area of ​​the beams in the inference beam set; The network load of the cell where the first device is located.

3. The method according to claim 2, characterized in that, The first beam subset is determined based on the beam priority, and the beam priority of the first beam in the inference beam set satisfies one or more of the following conditions: The beam priority of the first beam is positively correlated with the importance of the first beam; The beam priority of the first beam is positively correlated with the quality of the first beam; The beam priority of the first beam is negatively correlated with the amount of resources monitoring the first beam.

4. The method according to claim 2 or 3, characterized in that, The performance monitoring includes multiple monitoring instances, including the first monitoring instance and the second monitoring instance. The second monitoring instance corresponds to a second beam subset, and the first beam subset is the same as at least one beam in the second beam subset.

5. The method according to claim 1, characterized in that, The first beam set is a first subset of the inference beam set. The first subset of beams is determined according to a polling mechanism related to the inference beam set. The polling mechanism is used to monitor the performance of all beams in the inference beam set during a first polling period.

6. The method according to claim 5, characterized in that, The first polling period includes multiple monitoring instances, the multiple monitoring instances include the first monitoring instance, the multiple monitoring instances respectively correspond to multiple beam subsets including the first beam subset, and the beams in any two beam subsets of the multiple beam subsets are different.

7. The method according to claim 5 or 6, characterized in that, The duration of the first polling period is a positive integer multiple of the duration of the first monitoring period, and the first monitoring period includes at least one monitoring instance.

8. The method according to claim 5 or 6, characterized in that, The first polling cycle is related to one or more of the following: The moving speed of the first device; The network load of the cell where the first device is located; The longest polling period of the cell where the first device is located.

9. The method according to claim 8, characterized in that, The first polling period is: T = Tmax - α*v - β*L; Where α and β are weighting coefficients, T max The longest polling period is defined as v, where v represents the effect of the movement speed on the first polling period, and L represents the effect of the network load on the first polling period.

10. The method according to any one of claims 2-9, characterized in that, The number of beams in the first beam subset is related to one or more of the following: The prediction accuracy of the first model; The network load of the cell where the first device is located; The minimum number of beams in the first beam subset.

11. The method according to claim 10, characterized in that, When the first beam subset is the i-th beam subset among multiple beam subsets, where i is an integer greater than or equal to 1, the number of beams in the first beam subset is: Where γ and δ are weighting coefficients, Let E represent the minimum number of beams, E represent the impact of the prediction accuracy on the number of beams, and L represent the impact of the network load on the number of beams.

12. The method according to any one of claims 1-11, characterized in that, The first beam set corresponds to the first association identifier, and the first association identifier is the same as the association identifier of the inference beam set and / or the training beam set of the first model.

13. The method according to claim 12, characterized in that, The first beam set is one of multiple beam subsets determined by the polling mechanism of the inference beam set; the multiple beam subsets correspond one-to-one with multiple resource configurations in the first polling cycle, or the multiple beam subsets correspond to M resource configurations, where M is the number of reporting cycles in the first polling cycle.

14. The method according to any one of claims 1-13, characterized in that, The method further includes: The first device sends a first request to the second device, the first request being used to request the first reference signal; The first reference signal is at least one of a plurality of dedicated reference signals, which are used to monitor the performance of the first model.

15. The method according to any one of claims 1-14, characterized in that, The method further includes: The first device sends a monitoring report of the first model based on the reporting cycle of the performance monitoring; The monitoring report includes the performance parameters of the first model, which include the prediction accuracy and / or the prediction error of the first model.

16. The method according to claim 15, characterized in that, The monitoring report of the first model is configured through the first channel state information (CSI) report configuration; the first CSI report configuration is the same as the CSI report configuration corresponding to the inference beam set, or the first CSI report configuration is different from the CSI report configuration corresponding to the inference beam set.

17. The method according to any one of claims 1-16, characterized in that, The performance parameters of the first model are represented by the monitoring margin of the first model. The monitoring margin is determined based on the actual number of monitoring operations and the quality difference between the inference beam set and the corresponding beam in the first beam set.

18. The method according to claim 17, characterized in that, The monitoring margin is: Where N is the actual number of monitoring times, ε is the margin threshold, and RSRP(b) i Let be the predicted reference signal received power (RSRP) of beam b in the inference beam set during the i-th monitoring. The measured RSRP of beam b in the first beam set during the i-th monitoring is given.

19. The method according to any one of claims 1-18, characterized in that, The first model is an artificial intelligence or machine learning model.

20. A method for wireless communication, characterized in that, include: The second device transmits a first reference signal, which is transmitted through a first beam set; Wherein, the first reference signal corresponds to the first monitoring instance, the measurement results of the first beam set are used to monitor the performance of the first model, the first beam set is determined according to the inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes some beams in the inference beam set.

21. The method according to claim 20, characterized in that, The first beam set is a first subset of the inference beam set, and the first beam subset is determined based on first information, which includes one or more of the following: Beam priority in the inference beam set; Historical performance data of the beams in the inference beam set; The sensing quality of the beams in the inference beam set; The coverage area of ​​the beams in the inference beam set; The network load of the cell corresponding to the second device.

22. The method according to claim 21, characterized in that, The first beam subset is determined based on the beam priority, and the beam priority of the first beam in the inference beam set satisfies one or more of the following conditions: The beam priority of the first beam is positively correlated with the importance of the first beam; The beam priority of the first beam is positively correlated with the quality of the first beam; The beam priority of the first beam is negatively correlated with the amount of resources monitoring the first beam.

23. The method according to claim 21 or 22, characterized in that, The performance monitoring includes multiple monitoring instances, including the first monitoring instance and the second monitoring instance. The second monitoring instance corresponds to a second beam subset, and the first beam subset is the same as at least one beam in the second beam subset.

24. The method according to claim 20, characterized in that, The first beam set is a first subset of the inference beam set. The first subset of beams is determined according to a polling mechanism related to the inference beam set. The polling mechanism is used to monitor the performance of all beams in the inference beam set during a first polling period.

25. The method according to claim 24, characterized in that, The first polling period includes multiple monitoring instances, the multiple monitoring instances include the first monitoring instance, the multiple monitoring instances respectively correspond to multiple beam subsets including the first beam subset, and the beams in any two beam subsets of the multiple beam subsets are different.

26. The method according to claim 24 or 25, characterized in that, The duration of the first polling period is a positive integer multiple of the duration of the first monitoring period, and the first monitoring period includes at least one monitoring instance.

27. The method according to claim 24 or 25, characterized in that, The first polling cycle is related to one or more of the following: The moving speed of the first device receiving the first reference signal; The network load of the cell corresponding to the second device; The longest polling period for the cell corresponding to the second device.

28. The method according to claim 27, characterized in that, The first polling period is: T = T max ―α*v―β*L; Where α and β are weighting coefficients, T max The longest polling period is defined as v, where v represents the effect of the movement speed on the first polling period, and L represents the effect of the network load on the first polling period.

29. The method according to any one of claims 21-28, characterized in that, The number of beams in the first beam subset is related to one or more of the following: The prediction accuracy of the first model; The network load of the cell corresponding to the second device; The minimum number of beams in the first beam subset.

30. The method according to claim 29, characterized in that, When the first beam subset is the i-th beam subset among multiple beam subsets, where i is an integer greater than or equal to 1, the number of beams in the first beam subset is: Where γ and δ are weighting coefficients, Let E represent the minimum number of beams, E represent the impact of the prediction accuracy on the number of beams, and L represent the impact of the network load on the number of beams.

31. The method according to any one of claims 20-30, characterized in that, The first beam set corresponds to the first association identifier, and the first association identifier is the same as the association identifier of the inference beam set and / or the training beam set of the first model.

32. The method according to claim 31, characterized in that, The first beam set is one of multiple beam subsets determined by the polling mechanism of the inference beam set; the multiple beam subsets correspond one-to-one with multiple resource configurations in the first polling cycle, or the multiple beam subsets correspond to M resource configurations, where M is the number of reporting cycles in the first polling cycle.

33. The method according to any one of claims 20-32, characterized in that, The method further includes: The second device receives a first request sent by the first device, the first request being used to request the first reference signal; The first reference signal is at least one of a plurality of dedicated reference signals, which are used to monitor the performance of the first model.

34. The method according to any one of claims 20-33, characterized in that, The method further includes: The second device receives the monitoring report from the first model based on the performance monitoring reporting cycle; The monitoring report includes the performance parameters of the first model, which include the prediction accuracy and / or the prediction error of the first model.

35. The method according to claim 34, characterized in that, The monitoring report of the first model is configured through the first channel state information (CSI) report configuration; the first CSI report configuration is the same as the CSI report configuration corresponding to the inference beam set, or the first CSI report configuration is different from the CSI report configuration corresponding to the inference beam set.

36. The method according to any one of claims 20-35, characterized in that, The performance parameters of the first model are represented by the monitoring margin of the first model. The monitoring margin is determined based on the actual number of monitoring operations and the quality difference between the inference beam set and the corresponding beam in the first beam set.

37. The method according to claim 36, characterized in that, The monitoring margin is: Where N is the actual number of monitoring times, ε is the margin threshold, and RSRP(b) i Let be the predicted reference signal received power (RSRP) of beam b in the inference beam set during the i-th monitoring. The measured RSRP of beam b in the first beam set during the i-th monitoring is given.

38. The method according to any one of claims 20-37, characterized in that, The first model is an artificial intelligence or machine learning model.

39. A device for wireless communication, characterized in that, The device is a first device, the device comprising: A transceiver unit is used to receive a first reference signal, which is transmitted through a first beam set; The processing unit is used to monitor the performance of the first model based on the measurement results of the first beam set. Wherein, the first reference signal corresponds to the first monitoring instance, and the first beam set is determined according to the inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes some beams in the inference beam set.

40. The apparatus according to claim 39, characterized in that, The first beam set is a first subset of the inference beam set, and the first beam subset is determined based on first information, which includes one or more of the following: Beam priority in the inference beam set; Historical performance data of the beams in the inference beam set; The sensing quality of the beams in the inference beam set; The coverage area of ​​the beams in the inference beam set; The network load of the cell where the first device is located.

41. The apparatus according to claim 40, characterized in that, The first beam subset is determined based on the beam priority, and the beam priority of the first beam in the inference beam set satisfies one or more of the following conditions: The beam priority of the first beam is positively correlated with the importance of the first beam; The beam priority of the first beam is positively correlated with the quality of the first beam; The beam priority of the first beam is negatively correlated with the amount of resources monitoring the first beam.

42. The apparatus according to claim 40 or 41, characterized in that, The performance monitoring includes multiple monitoring instances, including the first monitoring instance and the second monitoring instance. The second monitoring instance corresponds to a second beam subset, and the first beam subset is the same as at least one beam in the second beam subset.

43. The apparatus according to claim 39, characterized in that, The first beam set is a first subset of the inference beam set. The first subset of beams is determined according to a polling mechanism related to the inference beam set. The polling mechanism is used to monitor the performance of all beams in the inference beam set during a first polling period.

44. The apparatus according to claim 43, characterized in that, The first polling period includes multiple monitoring instances, the multiple monitoring instances include the first monitoring instance, the multiple monitoring instances respectively correspond to multiple beam subsets including the first beam subset, and the beams in any two beam subsets of the multiple beam subsets are different.

45. The apparatus according to claim 43 or 44, characterized in that, The duration of the first polling period is a positive integer multiple of the duration of the first monitoring period, and the first monitoring period includes at least one monitoring instance.

46. ​​The apparatus according to claim 43 or 44, characterized in that, The first polling cycle is related to one or more of the following: The moving speed of the first device; The network load of the cell where the first device is located; The longest polling period of the cell where the first device is located.

47. The apparatus according to claim 46, characterized in that, The first polling period is: T = T max ―α*v―β*L; Where α and β are weighting coefficients, T max The longest polling period is defined as v, where v represents the effect of the movement speed on the first polling period, and L represents the effect of the network load on the first polling period.

48. The apparatus according to any one of claims 40-47, characterized in that, The number of beams in the first beam subset is related to one or more of the following: The prediction accuracy of the first model; The network load of the cell where the first device is located; The minimum number of beams in the first beam subset.

49. The apparatus according to claim 48, characterized in that, When the first beam subset is the i-th beam subset among multiple beam subsets, where i is an integer greater than or equal to 1, the number of beams in the first beam subset is: Where γ and δ are weighting coefficients, Let E represent the minimum number of beams, E represent the impact of the prediction accuracy on the number of beams, and L represent the impact of the network load on the number of beams.

50. The apparatus according to any one of claims 39-49, characterized in that, The first beam set corresponds to the first association identifier, and the first association identifier is the same as the association identifier of the inference beam set and / or the training beam set of the first model.

51. The apparatus according to claim 50, characterized in that, The first beam set is one of multiple beam subsets determined by the polling mechanism of the inference beam set; the multiple beam subsets correspond one-to-one with multiple resource configurations in the first polling cycle, or the multiple beam subsets correspond to M resource configurations, where M is the number of reporting cycles in the first polling cycle.

52. The apparatus according to any one of claims 39-51, characterized in that, The transceiver unit is further configured to send a first request to the second device, the first request being used to request the first reference signal; wherein the first reference signal is at least one of a plurality of dedicated reference signals, the dedicated reference signals being used to monitor the performance of the first model.

53. The apparatus according to any one of claims 39-52, characterized in that, The transceiver unit is further configured to send a monitoring report of the first model based on the reporting cycle of the performance monitoring; wherein the monitoring report includes the performance parameters of the first model, and the performance parameters of the first model include the prediction accuracy of the first model and / or the prediction error of the first model.

54. The apparatus according to claim 53, characterized in that, The monitoring report of the first model is configured through the first channel state information (CSI) report configuration; the first CSI report configuration is the same as the CSI report configuration corresponding to the inference beam set, or the first CSI report configuration is different from the CSI report configuration corresponding to the inference beam set.

55. The apparatus according to any one of claims 39-54, characterized in that, The performance parameters of the first model are represented by the monitoring margin of the first model. The monitoring margin is determined based on the actual number of monitoring operations and the quality difference between the inference beam set and the corresponding beam in the first beam set.

56. The apparatus according to claim 55, characterized in that, The monitoring margin is: Where N is the actual number of monitoring times, ε is the margin threshold, and RSRP(b) i Let be the predicted reference signal received power (RSRP) of beam b in the inference beam set during the i-th monitoring. The measured RSRP of beam b in the first beam set during the i-th monitoring is given.

57. The apparatus according to any one of claims 39-56, characterized in that, The first model is an artificial intelligence or machine learning model.

58. A device for wireless communication, characterized in that, The device is a second device, and the device includes: A transceiver unit is used to transmit a first reference signal, which is transmitted through a first beam set; Wherein, the first reference signal corresponds to the first monitoring instance, the measurement results of the first beam set are used to monitor the performance of the first model, the first beam set is determined according to the inference beam set of the first model; the first beam set includes all beams in the inference beam set, or the first beam set includes some beams in the inference beam set.

59. The apparatus according to claim 58, characterized in that, The first beam set is a first subset of the inference beam set, and the first beam subset is determined based on first information, which includes one or more of the following: Beam priority in the inference beam set; Historical performance data of the beams in the inference beam set; The sensing quality of the beams in the inference beam set; The coverage area of ​​the beams in the inference beam set; The network load of the cell corresponding to the second device.

60. The apparatus according to claim 59, characterized in that, The first beam subset is determined based on the beam priority, and the beam priority of the first beam in the inference beam set satisfies one or more of the following conditions: The beam priority of the first beam is positively correlated with the importance of the first beam; The beam priority of the first beam is positively correlated with the quality of the first beam; The beam priority of the first beam is negatively correlated with the amount of resources monitoring the first beam.

61. The apparatus according to claim 59 or 60, characterized in that, The performance monitoring includes multiple monitoring instances, including the first monitoring instance and the second monitoring instance. The second monitoring instance corresponds to a second beam subset, and the first beam subset is the same as at least one beam in the second beam subset.

62. The apparatus according to claim 58, characterized in that, The first beam set is a first subset of the inference beam set. The first subset of beams is determined according to a polling mechanism related to the inference beam set. The polling mechanism is used to monitor the performance of all beams in the inference beam set during a first polling period.

63. The apparatus according to claim 62, characterized in that, The first polling period includes multiple monitoring instances, the multiple monitoring instances include the first monitoring instance, the multiple monitoring instances respectively correspond to multiple beam subsets including the first beam subset, and the beams in any two beam subsets of the multiple beam subsets are different.

64. The apparatus according to claim 62 or 63, characterized in that, The duration of the first polling period is a positive integer multiple of the duration of the first monitoring period, and the first monitoring period includes at least one monitoring instance.

65. The apparatus according to claim 62 or 63, characterized in that, The first polling cycle is related to one or more of the following: The moving speed of the first device receiving the first reference signal; The network load of the cell corresponding to the second device; The longest polling period for the cell corresponding to the second device.

66. The apparatus according to claim 65, characterized in that, The first polling period is: T = T max ―α*v―β*L; Where α and β are weighting coefficients, T max The longest polling period is defined as v, where v represents the effect of the movement speed on the first polling period, and L represents the effect of the network load on the first polling period.

67. The apparatus according to any one of claims 59-66, characterized in that, The number of beams in the first beam subset is related to one or more of the following: The prediction accuracy of the first model; The network load of the cell corresponding to the second device; The minimum number of beams in the first beam subset.

68. The apparatus according to claim 67, characterized in that, When the first beam subset is the i-th beam subset among multiple beam subsets, where i is an integer greater than or equal to 1, the number of beams in the first beam subset is: Where γ and δ are weighting coefficients, Let E represent the minimum number of beams, E represent the impact of the prediction accuracy on the number of beams, and L represent the impact of the network load on the number of beams.

69. The apparatus according to any one of claims 58-68, characterized in that, The first beam set corresponds to the first association identifier, and the first association identifier is the same as the association identifier of the inference beam set and / or the training beam set of the first model.

70. The apparatus according to claim 69, characterized in that, The first beam set is one of multiple beam subsets determined by the polling mechanism of the inference beam set; the multiple beam subsets correspond one-to-one with multiple resource configurations in the first polling cycle, or the multiple beam subsets correspond to M resource configurations, where M is the number of reporting cycles in the first polling cycle.

71. The apparatus according to any one of claims 58-70, characterized in that, The transceiver unit is further configured to receive a first request sent by the first device, the first request being used to request the first reference signal; wherein the first reference signal is at least one of a plurality of dedicated reference signals, the dedicated reference signal being used to monitor the performance of the first model.

72. The apparatus according to any one of claims 58-71, characterized in that, The transceiver unit is further configured to receive a monitoring report of the first model based on the reporting period of the performance monitoring; wherein the monitoring report includes the performance parameters of the first model, and the performance parameters of the first model include the prediction accuracy of the first model and / or the prediction error of the first model.

73. The apparatus according to claim 72, characterized in that, The monitoring report of the first model is configured through the first channel state information (CSI) report configuration; the first CSI report configuration is the same as the CSI report configuration corresponding to the inference beam set, or the first CSI report configuration is different from the CSI report configuration corresponding to the inference beam set.

74. The apparatus according to any one of claims 58-73, characterized in that, The performance parameters of the first model are represented by the monitoring margin of the first model. The monitoring margin is determined based on the actual number of monitoring operations and the quality difference between the inference beam set and the corresponding beam in the first beam set.

75. The apparatus according to claim 74, characterized in that, The monitoring margin is: Where N is the actual number of monitoring times, ε is the margin threshold, and RSRP(b) i Let be the predicted reference signal received power (RSRP) of beam b in the inference beam set during the i-th monitoring. The measured RSRP of beam b in the first beam set during the i-th monitoring is given.

76. The apparatus according to any one of claims 58-75, characterized in that, The first model is an artificial intelligence or machine learning model.

77. A communication device, characterized in that, It includes a memory and a processor, the memory being used to store a program, and the processor being used to invoke the program in the memory to perform the method as described in any one of claims 1-38.

78. An apparatus, characterized in that, Includes a processor for calling a program from memory to perform the method as described in any one of claims 1-38.

79. A chip, characterized in that, Includes a processor for calling a program from memory, causing a device on which the chip is mounted to perform the method as described in any one of claims 1-38.

80. A computer-readable storage medium, characterized in that, It contains a program that causes a computer to perform the method as described in any one of claims 1-38.

81. A computer program product, characterized in that, Includes a program that causes a computer to perform the method as described in any one of claims 1-38.

82. A computer program, characterized in that, The computer program causes the computer to perform the method as described in any one of claims 1-38.

Images