Beam selection methods, devices, communication systems, storage media, and chip systems

Abstract

This application discloses a beam selection method, apparatus, communication system, storage medium, and chip system, belonging to the field of communication technology. The method includes: a target device acquiring multiple first measurement values, each corresponding one-to-one with multiple first beam pairs, wherein each first beam pair is obtained by combining B transmit beams corresponding to reference signals in a first reference signal set with multiple receive beams in pairs; and obtaining multiple predicted values ​​based on the multiple first measurement values ​​through a target AI model, each predicted value corresponding one-to-one with multiple second beam pairs, wherein each second beam pair is obtained by combining A transmit beams corresponding to reference signals in a second reference signal set with multiple receive beams in pairs, wherein the first reference signal set is a subset of the second reference signal set. In this application, beam selection can be performed relatively quickly and accurately from multiple second beam pairs based on these multiple predicted values, thereby improving measurement timeliness and communication efficiency.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a beam selection method, apparatus, communication system, storage medium, and chip system. Background Technology

[0002] In beam management scenarios, beam selection requires executing P1, P2, and P3 procedures. In the P1 procedure, the network device communicates with the terminal device to determine the wide transmit beam with better signal quality. In the P2 procedure, the network device communicates with the terminal device to determine the narrow transmit beam with better signal quality from the wide transmit beams selected in the P1 procedure. In the P3 procedure, the network device communicates with the terminal device to determine the receive beam with better signal quality. This beam management procedure suffers from significant signaling overhead and measurement latency, resulting in low measurement timeliness throughout the process. Summary of the Invention

[0003] This application provides a beam selection method, apparatus, communication system, storage medium, and chip system, which help improve measurement timeliness. The technical solution is as follows:

[0004] Firstly, a beam selection method is provided. This method can be executed by a target device, or by a component (such as a circuit, chip, or chip system) configured in the target device, or by a logic module or software capable of implementing all or part of the functions of the target device. This application does not limit the scope of the method. The following description uses a target device as an example.

[0005] The target device acquires multiple first measurement values, each corresponding one-to-one with multiple first beam pairs. Each first beam pair is obtained by combining B transmit beams corresponding to reference signals in a first reference signal set with multiple receive beams in pairs. Each first measurement value is obtained by measuring the reference signal corresponding to the transmit beam in the receive beam pair, where B is a positive integer. Based on these first measurement values, a target AI model obtains multiple predicted values, each corresponding one-to-one with multiple second beam pairs. Each second beam pair is obtained by combining A transmit beams corresponding to reference signals in a second reference signal set with multiple receive beams in pairs. The first reference signal set is a subset of the second reference signal set. These predicted values ​​are used for beam selection, where A is an integer greater than or equal to 2.

[0006] In this application, the target device can obtain multiple predicted values ​​corresponding to multiple second beam pairs through a target AI model based on multiple first measurement values ​​corresponding to multiple first beam pairs. Based on these multiple predicted values, beam selection can be performed relatively quickly and accurately from the multiple second beam pairs, thereby improving measurement timeliness and communication efficiency.

[0007] In one possible implementation, the operation of the target device obtaining multiple predicted values ​​through the target AI model based on the multiple first measurements can be as follows: generating a first matrix based on the multiple first measurements, the first matrix including the multiple first measurements; randomizing the element positions of the first matrix to obtain a second matrix; inputting the second matrix into the target AI model to obtain a third matrix output by the target AI model; and derandomizing the element positions of the third matrix to obtain a fourth matrix including the multiple predicted values.

[0008] In this application, after inputting a second matrix into the target AI model, the third matrix output by the target AI model can include multiple predicted values. Since the input second matrix is ​​obtained by randomizing the element positions of the first matrix, the positions of multiple predicted values ​​in the output third matrix have not yet been recovered, and the second beam pairs corresponding to each predicted value cannot yet be confirmed. Therefore, the element positions of the third matrix can be derandomized to obtain a fourth matrix, thereby recovering the positions of multiple predicted values.

[0009] In one possible implementation, after the target device obtains multiple predicted values ​​through the target AI model, it can further determine the target predicted value corresponding to each of the A transmit beams from the multiple predicted values. The target predicted value corresponding to the transmit beam is the largest predicted value among all the predicted values ​​corresponding to the transmit beam. The K transmit beams with the largest target predicted values ​​are determined from the A transmit beams, where K is a positive integer. For any one of the K transmit beams, the L receive beams corresponding to the largest L predicted values ​​are determined from all the predicted values ​​corresponding to the transmit beam, where L is a positive integer. Multiple beam identifier pairs are generated based on the beam identifiers of the K transmit beams and the beam identifiers of the L receive beams corresponding to each of the K transmit beams.

[0010] In this application, firstly, K transmit beams with the largest target predicted values ​​are selected from A transmit beams. Then, L receive beams corresponding to the L largest predicted values ​​are selected from all predicted values ​​of each of the K transmit beams, thereby determining multiple beam identifier pairs. Each of these multiple beam identifier pairs includes the beam identifier of one transmit beam from the K transmit beams and the beam identifier of one receive beam from the L receive beams corresponding to that transmit beam. In this way, not only are the TopK transmit beams determined, but also the TopL receive beams corresponding to each of the TopK transmit beams are determined. Subsequently, during the secondary confirmation in the P2 procedure, the optimal transmit beam and the optimal receive beam can be directly determined from the multiple beam pairs identified by these multiple beam identifier pairs. In this case, there is no need to perform receive beam confirmation in the P3 procedure, thereby reducing measurement latency, saving signaling overhead, reducing power consumption, and thus helping to improve measurement timeliness and communication efficiency.

[0011] In one possible implementation, the target device is a terminal device. After generating multiple beam identifier pairs, the terminal device can send these multiple beam identifier pairs to the network device. The network device uses the corresponding transmit beam to transmit a reference signal through the receiving beam identified by each of the multiple beam identifier pairs to obtain multiple second measurement values, which correspond one-to-one with the multiple beam identifier pairs. The multiple second measurement values ​​are then sent to the network device.

[0012] In this application, after receiving multiple beam identifier pairs sent by a terminal device, the network device can use the multiple transmit beams identified by the multiple beam identifier pairs to send reference signals to the terminal device, so as to obtain multiple second measurement values ​​corresponding one-to-one with the multiple beam identifier pairs through the terminal device. Based on the multiple second measurement values, the network device can determine the optimal transmit beam and the optimal receive beam from the beam pairs identified by the multiple beam identifier pairs. In this case, the optimal transmit beam and the optimal receive beam are determined in the P2 process, eliminating the need for the P3 process. This reduces measurement latency, saves signaling overhead, and reduces power consumption, thereby helping to improve measurement timeliness and communication efficiency.

[0013] In one possible implementation, before acquiring multiple first measurement values, the terminal device may also send inference capability information to the network device, the inference capability information including report content indication information, the report content indication information including information for indicating support for reporting beam pairs; and receive inference configuration information sent by the network device, the inference configuration information including information for indicating reporting beam pairs.

[0014] In this application, when the terminal device supports reporting beam pairs, the network device can instruct the terminal device to report beam pairs. This allows the network device to quickly and accurately select beams based on the beam pairs reported by the terminal device, improving measurement timeliness and communication efficiency.

[0015] In one possible implementation, before the terminal device transmits a reference signal using the corresponding transmit beam of the receiving beam measurement network device identified by each of the plurality of beam identifier pairs, it can also receive scheduling information sent by the network device, which includes the plurality of beam identifier pairs.

[0016] In this application, the terminal device can subsequently measure the signal quality corresponding to the multiple beam pairs identified by the multiple beam identifiers carried in the scheduling information based on the scheduling information.

[0017] In one possible implementation, the target device is a network device. After generating multiple beam identifier pairs, the network device can also use the transmit beams identified by the multiple beam identifier pairs to send reference signals to the terminal device; receive multiple second measurement values ​​sent by the terminal device, which correspond one-to-one with the multiple beam identifier pairs; take the transmit beam identified by the beam identifier pair corresponding to the largest second measurement value among the multiple second measurement values ​​as the optimal transmit beam, and take the receive beam identified by the beam identifier pair corresponding to the largest second measurement value among the multiple second measurement values ​​as the optimal receive beam.

[0018] In this application, after the network device obtains multiple beam identifier pairs, it can use the multiple transmit beams identified by the multiple beam identifier pairs to send reference signals to the terminal device. The terminal device then obtains multiple second measurement values ​​corresponding one-to-one with the multiple beam identifier pairs. Based on these second measurement values, the optimal transmit beam and the optimal receive beam can be determined from the beam pairs identified by the multiple beam identifier pairs. In this case, the optimal transmit beam and the optimal receive beam are determined in the P2 process, eliminating the need for the P3 process. This reduces measurement latency, saves signaling overhead, and lowers power consumption, thereby improving measurement timeliness and communication efficiency.

[0019] In one possible implementation, before the network device uses the multiple beam identifier pairs to send a reference signal to the terminal device via the identified transmit beam, it may also send scheduling information to the terminal device, which includes the multiple beam identifier pairs.

[0020] In this application, the terminal device can subsequently measure the signal quality corresponding to the multiple beam pairs identified by the multiple beam identifiers carried in the scheduling information based on the scheduling information.

[0021] In one possible implementation, the target device acquires multiple first samples, each of which includes multiple third measurements, which correspond one-to-one with the multiple second beam pairs; each of the multiple first samples is randomly row-masked to obtain multiple second samples; and a model is trained based on the multiple second samples to obtain the target AI model.

[0022] In this application, the second sample obtained by randomly row-masking the first sample can simulate the situation where a terminal device receives a first reference signal set in a real communication scenario. In this case, the trained target AI model can predict the signal quality corresponding to multiple second beam pairs based on the measured signal quality corresponding to multiple first beam pairs.

[0023] In one possible implementation, the target device trains a model based on the multiple second samples to obtain the target AI model. This can be achieved by: randomizing the element positions of each of the multiple second samples to obtain multiple third samples; inputting any one of the multiple third samples into the AI ​​model to obtain the first data output by the AI ​​model; derandomizing the element positions of the first data to obtain the second data; and adjusting the parameters in the AI ​​model based on the loss value between the second data and the first sample corresponding to the third sample. The AI ​​model with the adjusted parameters is the target AI model.

[0024] In this application, after randomizing the row mask of the first sample, the element positions are randomized. In this way, the correlation between elements in the first sample can be disrupted by random row and column permutations, thereby increasing the sample richness and enhancing the robustness of the AI ​​model trained on it.

[0025] Secondly, a communication device is provided, comprising a processing module and a communication module. The processing module acquires multiple first measurement values, each corresponding one-to-one with multiple first beam pairs. Each first beam pair is obtained by combining B transmit beams corresponding to reference signals in a first reference signal set with multiple receive beams in pairs. Each first measurement value is obtained by measuring the reference signal corresponding to the transmit beam in the receive beam pair, where B is a positive integer. Based on the multiple first measurement values, multiple predicted values ​​are obtained through a target AI model. These predicted values ​​correspond one-to-one with multiple second beam pairs, each obtained by combining A transmit beams corresponding to reference signals in a second reference signal set with multiple receive beams in pairs. The first reference signal set is a subset of the second reference signal set. The predicted values ​​are used for beam selection, where A is an integer greater than or equal to 2.

[0026] The second aspect is the implementation on the device side, which corresponds to the first aspect. The explanations, supplements, and descriptions of the beneficial effects of the first aspect also apply to the second aspect, and will not be repeated here.

[0027] Thirdly, a communication device is provided, including a processor. The processor is coupled to a memory and can be used to execute instructions or data in the memory to implement the methods in any possible implementation of any of the above aspects. Optionally, the communication device further includes a memory. Optionally, the communication device further includes a communication interface, and the processor is coupled to the communication interface.

[0028] In one implementation, the communication interface can be a transceiver, or an input / output interface.

[0029] In another implementation, the communication device is a chip configured in a terminal device or network device. When the communication device is a chip configured in a terminal device or network device, the communication interface can be an input / output interface.

[0030] Fourthly, a computer program product is provided, comprising: a computer program (also referred to as code or instructions) that, when run, causes a computer to perform the method in any possible implementation of any of the above aspects.

[0031] Fifthly, a computer-readable storage medium is provided that stores a computer program (also referred to as code or instructions) that, when run on a computer, causes the computer to perform the methods in any possible implementation of any of the above aspects.

[0032] Sixthly, embodiments of this application provide a chip system including one or more processors for calling and executing instructions stored in memory, causing the methods in any of the possible implementations of the above aspects to be executed. The chip system may be composed of chips or may include chips and other discrete devices.

[0033] The chip system may include input circuits or interfaces for transmitting information or data, and output circuits or interfaces for receiving information or data.

[0034] In a seventh aspect, a communication system is provided, including the aforementioned terminal device and network device. Optionally, the communication system may further include other devices that communicate with the terminal device and / or network device. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of a communication system provided in an embodiment of this application.

[0036] Figure 2 This is a schematic diagram of beam prediction provided in an embodiment of this application.

[0037] Figure 3 This is a schematic diagram of another beam prediction method provided in an embodiment of this application.

[0038] Figure 4 This is a schematic diagram of a technical architecture provided in an embodiment of this application.

[0039] Figure 5 This is a flowchart of a model training method provided in an embodiment of this application.

[0040] Figure 6 This is a schematic diagram of a model training process provided in an embodiment of this application.

[0041] Figure 7 This is a flowchart of another model training method provided in the embodiments of this application.

[0042] Figure 8 This is a flowchart of a model reasoning method provided in an embodiment of this application.

[0043] Figure 9 This is a schematic diagram of a model reasoning process provided in an embodiment of this application.

[0044] Figure 10 This is a flowchart of a beam selection method provided in an embodiment of this application.

[0045] Figure 11 This is a flowchart of another beam selection method provided in the embodiments of this application.

[0046] Figure 12 This is a schematic block diagram of a communication device provided in an embodiment of this application.

[0047] Figure 13 This is a schematic block diagram of another communication device provided in the embodiments of this application. Detailed Implementation

[0048] In the following description, specific details such as particular system architectures and technologies are set forth for illustrative purposes and not for limiting purposes, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details.

[0049] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0050] It should be understood that "one or more" as used in this application refers to one, two, or more, and "multiple" as used in this application refers to two or more. In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. "And / or" in this document 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, and B existing alone.

[0051] To facilitate a clear description of the technical solutions of this application, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" do not necessarily imply that they are different.

[0052] The terms "one embodiment" or "some embodiments" used in this application mean that one or more embodiments of this application include the specific features, structures, or characteristics described in that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in other embodiments," etc., appearing in different parts of this application do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized.

[0053] The embodiments of this application can be applied to various communication systems. For example, Global System for Mobile Communications (GSM) systems, General Packet Radio Service (GPRS) systems, Wireless Local Area Network (WLAN) systems (such as Wireless Fidelity (Wi-Fi) systems), Long Term Evolution (LTE) systems, LTE Frequency Division Duplex (FDD) systems, LTE Time Division Duplex (TDD) systems, sidelink communication systems, Universal Mobile Telecommunication System (UMTS), Worldwide Interoperability for Microwave Access (WiMAX) communication systems, non-terrestrial network (NTN) communication systems, fourth-generation (4G) mobile communication systems, fifth-generation (5G) mobile communication systems, new radio access technology (NR) systems, and sixth-generation (6G) mobile communication systems. The 5G mobile communication system may include non-standalone (NSA) and / or standalone (SA) networking. It is understood that the embodiments of this application can also be applied to future communication systems, and the embodiments of this application do not limit this application.

[0054] Figure 1 This is a schematic diagram of a communication system 100 provided in an embodiment of this application. The communication system 100 may include network (NW) devices, such as... Figure 1 The network device 110 shown. The communication system 100 may also include terminal devices, such as... Figure 1 The terminal device 120 shown can communicate with the network device via a wireless link. Figure 1 An exemplary network device 110 and a terminal device 120 are shown. Optionally, the communication system 100 may also include multiple network devices and / or multiple terminal devices.

[0055] The network device in this application embodiment can be a network-side device such as an access network device or a core network device.

[0056] Access network equipment is sometimes also called access node. Access network equipment has wireless transceiver capabilities and can communicate with terminal equipment. For example, access network equipment can be a base station, an evolved NodeB (eNodeB), a transmission reception point (TRP), next-generation radio access network (NG-RAN) equipment (such as a next-generation NodeB (gNB)) in a 5G mobile communication system, access network equipment or modules of access network equipment in an open RAN (ORAN) system, satellites in an NTN communication system, base stations in a future mobile communication system, or access points (APs) in a Wi-Fi system. Access network equipment can also be modules or units capable of implementing some of the functions of a base station, such as macro base stations, micro base stations, indoor stations, relay nodes, or donor nodes. Multiple access network devices in the communication system 100 can be of the same type or different types. This application does not limit the specific technology or device form used in the access network equipment.

[0057] Core network equipment possesses functions such as data processing, session management, network interconnection, operation administration and maintenance (OAM), and location management function (LMF). Core network equipment can perform user access authentication, service bearer establishment, and data interaction with external networks. Through OAM, it completes network configuration monitoring, resource scheduling optimization, and fault maintenance tasks. Through LMF, it provides terminal location calculation, trajectory tracking, and spatial data analysis. For example, core network equipment can be network elements such as the mobility management entity (MME), serving gateway (SGW), and packet data network gateway (PGW) in a 4G mobile communication system. Alternatively, core network equipment can be network elements such as the access and mobility management function (AMF), session management function (SMF), user plane function (UPF), and LMF in a 5G mobile communication system. Or, core network equipment can be a functional entity specifically providing operation and maintenance services or location services, such as an independent server closely cooperating with the core network. Alternatively, the core network equipment can be a virtualized network element integrating OAM or LMF capabilities within a Network Functions Virtualization (NFV) architecture. The core network equipment can also be a novel core network entity in future communication systems. Multiple core network equipment in communication system 100 can be deployed in a centralized or distributed architecture, with each core network equipment undertaking the same or different types of network functions. This application does not limit the specific technologies or equipment forms used in the core network equipment.

[0058] In this application embodiment, the apparatus for implementing the functions of a network device can be a network device itself, or an apparatus capable of supporting the network device in implementing those functions, such as a processor, circuit, chip, or chip system. This apparatus can be installed in the network device or connected to and used with the network device. In this application embodiment, taking a network device as an example to illustrate the technical solution provided by this application, we will describe it accordingly.

[0059] The terminal device in this application embodiment can be a wireless terminal device capable of receiving network device scheduling and instructions. The wireless terminal device can be a device providing voice and / or data connectivity to a user, a handheld device with wireless connectivity, or other processing devices connected to a wireless modem. For example, the terminal device can communicate with one or more core networks or the Internet via a radio access network (RAN). The terminal device can also be referred to as a terminal, user equipment (UE), mobile terminal (MT), mobile station (MS), mobile unit (MU), radio unit, remote unit, user agent, mobile client, etc. Terminal devices can be widely used in various scenarios, such as device-to-device (D2D), vehicle-to-everything (V2X) communication, machine-type communication (MTC), Internet of Things (IoT), ultra-reliable low-latency communication (URLLC), virtual reality (VR), augmented reality (AR), industrial control, self-driving, remote medical surgery, smart grid, intelligent transportation, smart homes, smart cities, or satellite communication. Terminal devices can be mobile phones, tablets, computers with wireless transceiver capabilities, wearable devices, vehicles, aircraft (such as drones, helicopters, and airplanes), hot air balloons, ships, robots, robotic arms, or smart home devices. This application does not limit the form of the terminal device.

[0060] In this application embodiment, the device for implementing the functions of the terminal device can be the terminal device itself, or any device capable of supporting the terminal device in implementing the functions, such as a processor, circuit, chip, or chip system. This device can be installed in the terminal device or connected to and used with the terminal device. In this application embodiment, taking the terminal device as an example to illustrate the technical solution provided by this application, we will describe it accordingly.

[0061] To facilitate understanding of the embodiments of this application, the technical terms involved in the embodiments of this application will be briefly explained first. Optionally, the explanation of some terms can also refer to the explanation in the 3rd Generation Partnership Project (3GPP) standard protocol.

[0062] 1. Beam

[0063] A beam is a directional energy radiation pattern formed when electromagnetic waves propagate in space. Its energy is concentrated in the target direction, with energy attenuation in non-target directions. The beam described in the embodiments of this application can also be referred to as a beam direction; the two terms can be used interchangeably.

[0064] 2. Beamforming

[0065] Beamforming is a technique that dynamically constructs and tracks the optimal beam pattern by optimizing the amplitude weighting, phase offset, and time delay compensation parameters of a multi-antenna array. Its core objective is to achieve coherent signal superposition in the target direction and energy destructive cancellation in the interference direction, thereby overcoming path loss and improving spatial multiplexing capabilities.

[0066] Beamforming relies on decision-making information provided by beam management (BM). During beam management, network devices can transmit reference signals using multiple candidate beams. Terminal devices can measure the signal quality of each reference signal (including but not limited to reference signal receiving power (RSRP)) and provide feedback to the network devices. The network devices can then perform beam switching based on this feedback.

[0067] It should be noted that one reference signal corresponds to one beam. That is, when a reference signal is transmitted using a certain beam, then that reference signal corresponds to the beam identifier of that beam. For example, the beam identifier can be a beam number, beam index, or channel state information-reference signal resource indicator (CRI), etc., but this application embodiment does not limit this.

[0068] The signal quality of the reference signal described in the embodiments of this application can also be referred to as the signal quality of the beam corresponding to the reference signal. In other words, the signal quality of a certain beam refers to the signal quality of the reference signal transmitted using that beam.

[0069] 3. Reference signal (RS)

[0070] A reference signal is a known signal used in a communication system for channel estimation or channel sounding. Its core function is to provide channel state information to the receiver, assisting in efficient resource scheduling and data transmission.

[0071] Optionally, the reference signal described in the embodiments of this application may include a channel state information-reference signal (CSI-RS), a synchronization signal / physical broadcast channel block (SSB) reference signal, etc., and the embodiments of this application do not limit this.

[0072] It should be understood that the technical terminology used in the embodiments of this application is for illustrative purposes only and not as a limitation. As technology evolves, technical terminology may also change; however, other technical terms with the same technical meaning should also be applicable to the embodiments of this application.

[0073] The application scenarios involved in the embodiments of this application are described below.

[0074] Beam management involves selecting, maintaining, and optimizing directional beams between network devices and terminal devices to ensure reliable and high-quality communication. Its goal is to establish and maintain suitable beam pairs, i.e., selecting appropriate transmit beams (also known as Tx beams) at the transmitting end and appropriate receive beams (also known as Rx beams) at the receiving end. This application's embodiments relate to AI-based beam management in artificial intelligence (AI) (also known as machine learning (ML)) air interface applications. Through AI technology, more intelligent beamforming and resource management can be achieved.

[0075] The following is a brief explanation of the P1, P2, and P3 procedures in beam management:

[0076] 1. P1: Beam Selection

[0077] The P1 process is the initial beam discovery and alignment process. Its main task is to enable network devices and terminal devices to find a preliminary, usable beam pair among many possible beams for connection. The operation is as follows: the network device transmits a reference signal using a relatively wide set of transmit beams, while the terminal device performs measurements through different receive beams and reports the wide transmit beam with the best signal quality to the network device.

[0078] 2. P2: Transmit Beam Refinement (Tx Beam Refinement)

[0079] After the initial beam connection is established in the P1 procedure, the P2 procedure begins. Its purpose is to further refine the selection of a more precise and higher-performing transmit beam based on the relatively optimal wide transmit beam selected in the P1 procedure. The operation is as follows: the network device transmits a reference signal using a set of narrower, more precise beams on the wide transmit beam selected in the P1 procedure. The terminal device consistently uses a receiving beam for measurement and reports the narrow transmit beam with the best signal quality to the network device.

[0080] 3. P3: Receive beam refinement (Rx Beam Refinement)

[0081] The P3 process focuses on optimizing the receiving beam of the terminal device. When the network device's transmit beam (determined through the P1 and P2 processes) is relatively fixed, the P3 process ensures that the terminal device receives the signal through the optimal receive beam. The process works as follows: the network device consistently transmits a reference signal using a high-quality transmit beam (such as the narrow transmit beam selected in the P2 process), and the terminal device measures using its supported set of receive beams, identifies the most suitable receive beam for the current transmit beam, and reports it to the network device.

[0082] The P2 process can be implemented based on the model inference function in the terminal device. Model inference refers to obtaining the signal quality prediction value of a specific beam set using a trained AI model. The purpose of beam prediction is to use a reference signal set SetB with lower overhead to predict a reference signal set SetA with higher overhead, while relying on lifecycle management (LCM) to set key performance indicators (KPIs) for performance monitoring and model switching.

[0083] For example, AI-based beam prediction mainly includes two major application scenarios: spatial beam prediction and temporal beam prediction. The following describes the AI ​​use cases (BM-Case 1 and BM-Case 2 as described below) for these two scenarios:

[0084] 1. BM-Case 1: Based on the measurement results of Set B, perform spatial downlink beam prediction for Set A.

[0085] In BM-Case 1, the network device sends a set of reference signals, Set B. The terminal device measures the signal quality of Set B and uses this measurement as input or part of the input to an AI model. The AI ​​model then predicts the signal quality of a set of reference signals, Set A, yielding the predicted value for Set A. Based on the AI ​​model's prediction, the terminal device reports the beam identifiers of the Top K reference signals in Set A to the network device. The Top K reference signals in Set A refer to the K predicted reference signals with the best signal quality in Set A.

[0086] Set B is a subset of Set A. For example, Set B includes CSI-RS#[2,6], and Set A includes CSI-RS#[1,2,3,4,5,6]. Figure 2 As shown, assuming the terminal device measures the signal quality of Set B and obtains RSRP#[2,6], it uses this as input to the AI ​​model. The AI ​​model outputs the predicted RSRP values ​​of the six reference signals in Set A, or the probability that each reference signal in Set A has the largest RSRP value in Set A. Clearly, without the AI ​​model, the network device would need to send the entire Set A to the terminal device, increasing its resource overhead.

[0087] 2. BM-Case2: Based on the historical measurement results of Set B, perform time-domain downlink beam prediction for Set A.

[0088] Figure 3 This is a schematic diagram of a BM-Case2 provided in an embodiment of this application. For example... Figure 3 As shown, the terminal device can measure the signal quality of Set B at one or more past transmission occasions, using the historical measurement results of Set B as input or part of the input to an AI model, and then use the AI ​​model to predict the signal quality of Set A at one or more future occasions. Based on the prediction results of the AI ​​model, the terminal device reports the beam identifiers of the TopK reference signals in Set A at one or more future occasions to the network device.

[0089] It can be seen that BM-Case1 is a single-slot beam prediction, while BM-Case2 is a multi-slot beam prediction. Therefore, in some cases, BM-Case1 can be regarded as a special case of BM-Case2 in the beam prediction process.

[0090] It should be noted that AI models can include terminal-side models and network-side models. The above only uses terminal-side models as an example to introduce AI use cases, but it does not mean that AI use cases can only have terminal-side models.

[0091] Furthermore, the above is merely a simple illustrative description of the model inference process using BM-Case1 and BM-Case2 as two AI use cases. The AI ​​use cases described above do not limit the embodiments of this application. In practical applications, other AI use cases can also be used.

[0092] As described in the model reasoning process above, the terminal device does not need to measure all reference signals (i.e., Set A), but only a small portion of the reference signals (i.e., Set B). Using these limited measurement results, the terminal device uses an AI model to predict the TopK beam with the best signal quality in Set A and feeds it back to the network device. Currently, the beams predicted by the AI ​​model are still only a candidate set; the final transmit beam used for data transmission needs further selection. For example, based on the prediction results reported by the terminal device, the network device can send reference signals back to the terminal device. The terminal device then performs actual measurements on these reference signals and reports the results. The network device then selects the beam corresponding to the reference signal with the best signal quality as the optimal transmit beam.

[0093] In relevant application scenarios, after determining the optimal transmit beam through the P2 process, determining the optimal receive beam requires executing the complete P3 process. This leads to increased measurement latency, increased signaling overhead, and increased power consumption, which in turn reduces the measurement timeliness in dynamic scenarios and reduces communication efficiency.

[0094] To address this, embodiments of this application provide a beam selection method that predicts transmit-receive (Tx-Rx) beam pairs in the P2 procedure and determines the optimal transmit and receive beams accordingly. This eliminates the need for the P3 procedure to determine the optimal beam, thereby reducing measurement delay, saving signaling overhead, and lowering power consumption, ultimately improving measurement timeliness and communication efficiency.

[0095] The technical architecture of the beam selection method provided in the embodiments of this application is described below:

[0096] Figure 4 This is a schematic diagram of a technical architecture provided in an embodiment of this application. For example... Figure 4 As shown, the technical architecture can include an AI model training part, an AI model inference part, and an optimal beam selection part.

[0097] 1. AI Model Training

[0098] The AI ​​model training in this application embodiment can be performed on any device, with the aim of training an AI model that takes multiple measurement values ​​corresponding one-to-one with multiple first beam pairs as input and can output multiple predicted values ​​corresponding one-to-one with multiple second beam pairs as input.

[0099] Multiple first beam pairs are obtained by combining B transmit beams corresponding to reference signals in the first reference signal set with multiple receive beams in pairs. Multiple second beam pairs are obtained by combining A transmit beams corresponding to reference signals in the second reference signal set with multiple receive beams in pairs. The first reference signal set is a subset of the second reference signal set. For example, the first reference signal set can be a reference signal set SetB, and the second reference signal set can be a reference signal set SetA.

[0100] 2. AI Model Inference

[0101] The trained AI model can be deployed on terminal devices or network devices. The terminal devices or network devices can input multiple measurement values ​​corresponding to multiple first beam pairs into the AI ​​model, and the AI ​​model can output multiple predicted values ​​corresponding to multiple second beam pairs.

[0102] 3. Optimal beam selection

[0103] Beam selection can be performed based on multiple predicted values ​​that correspond one-to-one with multiple second beam pairs.

[0104] Optionally, the AI ​​model described in the embodiments of this application may be a long short-term memory network (LSTM) model, a self-attention model (such as a transformer model), a convolutional neural network (CNN) model, a graph neural network (GNN) model, etc., and the embodiments of this application do not limit it.

[0105] Optionally, the signal quality of the beam can be determined based on one or more of the following indicators: layer 1 reference signal received power (L1-RSRP), RSRP, reference signal receiving quality (RSRQ), received signal strength indicator (RSSI), and signal to interference plus noise ratio (SINR).

[0106] It should be noted that, in some implementations, the technical architecture in the embodiments of this application can be directly configured within the LCM framework without adding new signaling and overhead.

[0107] The beam selection method provided in this application embodiment will be described in detail below with reference to the corresponding flowcharts. It is understood that the illustrative flowcharts provided in this application embodiment mainly use different devices (such as network devices and terminal devices) as examples of the execution subjects of the interaction to illustrate the beam selection method, but this application embodiment does not limit the execution subject of the interaction. For example, the device (such as a network device or terminal device) in the illustrative flowchart can also be a chip, chip system, or processor that supports the device in implementing the beam selection method, or it can be a logic module or software that can implement all or part of the functions of the device.

[0108] As a general statement, the message or signaling interactions involved in the interaction process of this application embodiment can be standard messages or signaling or newly introduced messages or signaling. This application embodiment does not limit this.

[0109] Understandable, the following text Figures 5 to 11 The network device described in the implementation method can be the one mentioned above. Figure 1 Any of the network devices described in the embodiments can also be devices within a network device (such as processors, chips, or chip systems). (The following...) Figures 5 to 11 The terminal device described in the implementation method can be as described above. Figure 1 Any of the terminal devices described in the embodiments can also be devices within the terminal device (such as processors, chips, or chip systems).

[0110] The beam selection method provided in this application includes an AI model training part, an AI model inference part, and an optimal beam selection part.

[0111] The following is combined with Figure 5 The AI ​​model training part of the embodiments of this application will be explained in detail.

[0112] Figure 5 This is a flowchart illustrating a model training method provided in an embodiment of this application. This model training method can be applied to a first device, which may be a network device, a terminal device, or other devices; this embodiment does not limit the application to this. Figure 5 As shown, the model training method may include the following steps:

[0113] Step 501: The first device acquires multiple first samples.

[0114] Each of the plurality of first samples may include multiple third measurements. It is assumed that each first sample includes A × C third measurements, where A and C are integers greater than or equal to 2. Each of the A × C third measurements corresponds one-to-one with one of the A × C second beam pairs. Any one of the A × C second beam pairs includes a transmit beam and a receive beam. The A × C second beam pairs are obtained by combining A transmit beams supported by the network device with C receive beams supported by the terminal device. For example, the A transmit beams may be the beams corresponding to the reference signals in the second reference signal set, i.e., the beams used to transmit the second reference signal set.

[0115] Each of the A×C third measurements is obtained by measuring the reference signal corresponding to the transmit beam of the receiving beam pair in the second beam pair. That is, for any one of the A×C third measurements, the third measurement is the signal quality measured by using the reference signal transmitted by the transmit beam of the corresponding second beam pair through the receiving beam pair.

[0116] For example, the first sample can be a matrix as shown in formula (1). .matrix In the matrix, row index i is the beam identifier of the transmitted beam. The column index j in the matrix is ​​the beam identifier of the receiving beam. Any element in the matrix represents the third measurement value corresponding to a second beam pair. The element in the i-th row and j-th column is the signal quality measured by the terminal device when "the network device transmits a signal using the i-th transmit beam out of A transmit beams, and the terminal device receives a signal through the j-th receive beam out of C receive beams".

[0117] (1)

[0118] in, It refers to a real number matrix with A rows and C columns. That is, both A and C are natural numbers. , is a matrix The element in the i-th row and j-th column. , .

[0119] It should be noted that the matrix The subscripts (i.e., i, j) of the elements are the beam identifiers of the transmit and receive beams in the second beam pair corresponding to that element. Formula (1) is illustrated by taking the beam identifiers of the A transmit beams as 1, 2, ..., A-1, A and the beam identifiers of the B receive beams as 1, 2, ..., C-1, C as an example. The subscripts of each element in Formula (1) do not limit the embodiments of this application. In actual applications, the subscripts of each element can be the actual beam identifiers of the transmit and receive beams in the second beam pair corresponding to that element.

[0120] Step 502: The first device performs a random row mask on each of the multiple first samples to obtain multiple second samples.

[0121] Random row masking refers to the operation of masking certain rows of data in a first sample by randomly generating masking rules, while preserving the remaining rows of data.

[0122] For example, for the first sample (i.e., the matrix) Perform random row masking to obtain the second sample (i.e., matrix). The operation of ) can be represented by the following formula (2).

[0123] (2)

[0124] in, Function representation from matrix The operation of randomly masking several rows while retaining the remaining rows.

[0125] Assuming that the A transmitted beams are used to transmit the second reference signal set, then the matrix The unmasked rows in the matrix are then applied to the beam that transmits the first set of reference signals; that is, the number of these rows equals the number of reference signals in the first set of reference signals. Thus, the matrix... It can simulate the situation where a terminal device receives a first set of reference signals in a real communication scenario.

[0126] Step 503: The first device trains the model based on the multiple second samples to obtain the target AI model.

[0127] The target AI model can predict the signal quality of A×C second beam pairs based on the signal quality of multiple first beam pairs. These multiple first beam pairs are obtained by combining B transmit beams supported by the network device with C receive beams supported by the terminal device; that is, there are B×C first beam pairs, where B is a positive integer. For example, the B transmit beams are the beams corresponding to the reference signals in the first reference signal set, i.e., the beams used to transmit the first reference signal set.

[0128] In some implementations, step 503 can be performed in two possible ways:

[0129] In the first possible approach, for any one of the multiple second samples, the first device inputs the second sample into the AI ​​model to obtain the third data output by the AI ​​model; the parameters in the AI ​​model are then adjusted based on the loss value between the third data and the corresponding first sample. The AI ​​model with adjusted parameters can be called the target AI model.

[0130] Each of the multiple second samples corresponds one-to-one with a single first sample. By performing random row masking on each first sample sequentially, a corresponding second sample can be obtained. In this case, the first sample corresponding to a second sample is the label (ground truth) of that second sample.

[0131] This AI model can be called a reconstruction model, which is used to perform matrix reconstruction. That is, this AI model can recover the masked elements in the second sample.

[0132] The third data is the predicted value output by the AI ​​model, and the first sample corresponding to the second sample is the label. Therefore, the parameters in the AI ​​model can be adjusted according to the loss value between the third data and the first sample corresponding to the second sample.

[0133] It should be noted that the first device can sequentially input the multiple second samples into the AI ​​model to train the AI ​​model. The first device can stop training when a first preset condition is met, and the AI ​​model obtained at this time can be called the target AI model.

[0134] The first preset condition can be set in advance. For example, the first preset condition can be that the parameters in the AI ​​model have been adjusted based on each of the plurality of second samples, or the first preset condition can be that the number of iterations has reached a first preset number of iterations, or the first preset condition can be that the loss value is less than a first preset loss value. Of course, the first preset condition can also be other conditions, and this application embodiment does not limit them. Among them, one parameter adjustment of the AI ​​model based on one second sample can be called one iteration.

[0135] The second possible approach is as follows: A first device randomizes the element positions of each of the multiple second samples to obtain multiple third samples. For any one of these third samples, it inputs it into an AI model to obtain the first data output by the AI ​​model. The first data is then derandomized to obtain second data. The parameters in the AI ​​model are adjusted based on the loss value between the second data and the corresponding first sample. The AI ​​model with adjusted parameters can be called the target AI model.

[0136] Element position randomization refers to the operation of randomly adjusting the positions of elements in a matrix. Its purpose is to change the original state of the matrix through random transformation, generating a matrix with uncertainty. In this case, after randomizing the element positions of the second sample to obtain the third sample, each element in the third sample comes from the second sample, but the positions of all elements have been randomly rearranged.

[0137] For example, for the second sample (i.e., the matrix) Randomize the element positions to obtain the third sample (i.e., the matrix). The operation of ) can be represented by the following formula (3).

[0138] (3)

[0139] in, The function is used to perform random permutation operations.

[0140] For example, The algorithm flow for the function can be as follows: Denote the original matrix... Total number of elements Record the original position The linear index (i.e., the one-dimensional index) is ,satisfy Remember the new location The linear index is ,satisfy After the mask, define a randomly permuted operator. , it is A random bijection that satisfies Therefore, for each original position First calculate its linear index. Then use the random bijective operator Get a new linear index Finally, through the formula Linear index Map to new location This algorithm can randomize the positions of matrix elements, satisfying the following conditions: .

[0141] Of course, this is not the only way; the randomization of matrix element positions can also be achieved in other ways, and this application does not limit this approach.

[0142] Each of the multiple third samples corresponds one-to-one with a single first sample. By sequentially applying random row masks and randomizing element positions to a first sample, a corresponding third sample can be obtained. In this case, the first sample corresponding to a third sample is the label of that third sample.

[0143] This AI model, which can be called a reconstruction model, is used for matrix reconstruction; that is, it can recover the masked elements in the third sample. This AI module can learn beam correlation, thus enabling the training of a highly stable AI model using sample data.

[0144] For example, the third sample (i.e., the matrix) After inputting the AI ​​model, the processing of the AI ​​model can be represented by the following formula (4).

[0145] (4)

[0146] in, This is the first data output by the AI ​​model. For this AI model, This is the set of trainable parameters in the AI ​​model.

[0147] Derandomization of element positions is the inverse operation of randomization of element positions, used to restore the positions of elements in a matrix. In this case, after derandomizing the first data to obtain the second data, each element in the second data comes from the first data, but the positions of all elements have been restored.

[0148] For example, for the first data (i.e., the matrix) Randomize the element positions to obtain the second data (i.e., the matrix). The operation of ) can be represented by the following formula (5).

[0149] (5)

[0150] in, The function is used to perform inverse permutation operations.

[0151] For example, The algorithm flow of the function can be as follows: For each position First calculate its linear index. Then use the inverse random bijective operator To obtain the linear index Finally, through the formula Linear index Mapped to position This algorithm can achieve the derandomization of matrix element positions, i.e., satisfy: .

[0152] The second data is the predicted value output by the AI ​​model, and the first sample corresponding to the third sample is the label. Therefore, the parameters in the AI ​​model can be adjusted according to the loss value between the second data and the first sample corresponding to the third sample.

[0153] For example, the first device can determine the loss value between the second data and the first sample corresponding to the third sample by using a preset loss function.

[0154] The preset loss function can be set in advance. For example, the preset loss function can be shown in the following formula (6):

[0155] (6)

[0156] in, The loss value. For tags, This is the second data point.

[0157] For example, the training objective could be: .

[0158] It should be noted that the first device can sequentially input the multiple third samples into the AI ​​model to train it. The first device can stop training when a second preset condition is met, and the AI ​​model obtained at this point can be called the target AI model.

[0159] The second preset condition can be set in advance. For example, the second preset condition can be that the parameters in the AI ​​model were adjusted based on each of the plurality of third samples, or the second preset condition can be that the number of iterations reached a second preset number of iterations, or the second preset condition can be that the loss value is less than a second preset loss value. Of course, the second preset condition can also be other conditions, and this application embodiment does not limit them. Among them, one parameter adjustment of the AI ​​model based on one third sample can be called one iteration.

[0160] For example, Figure 6 This is a schematic diagram illustrating a model training process provided in an embodiment of this application. For example... Figure 6 As shown, the original matrix M can be randomly row-masked and its element positions randomized before being input into the AI ​​model. After obtaining the output data of the AI ​​model, the element positions of the output data are de-randomized, and a loss value is calculated with the original matrix M. Based on this loss value, the gradient is calculated using the backpropagation algorithm, and the parameters in the AI ​​model are adjusted using the stochastic gradient descent (SGD) algorithm to train the AI ​​model. It is worth noting that in this embodiment, after randomly row-masking the original matrix M, element positions are also randomized. In this way, random row and column permutations can disrupt the element correlation of the matrix, thereby increasing the sample richness and enhancing the robustness of the AI ​​model trained accordingly.

[0161] It should be noted that in some embodiments, multiple first samples can be pre-set, and the third measurement value among these multiple first samples can be obtained based on historical data. In other embodiments, the signal quality of a second reference signal set can be measured, and multiple first samples can be generated accordingly for model training. The following will combine... Figure 7 This needs to be explained.

[0162] Figure 7 This is a flowchart illustrating a model training method provided in an embodiment of this application. For example... Figure 7 As shown, the model training method may include the following steps:

[0163] Step 701: The terminal device sends data collection capability information to the network device.

[0164] This data collection capability information is used to report the data collection capabilities supported by the terminal device to the network device. For example, this data collection capability information may include one or more of the following: AI model support capabilities: maximum number of inputs, maximum number of outputs, etc.; computing resources: available memory, processor computing power (which determines whether local training is possible), etc.; data collection constraints: maximum continuous measurement duration, upper limit of sampling frequency, etc.

[0165] Step 702: After receiving the data collection capability information sent by the terminal device, the network device sends data collection configuration information to the terminal device.

[0166] This data collection configuration information is used to configure data collection. For example, the data collection configuration information may include one or more of the following: reference signal type, such as CSI-RS or SSB; reference signal information: specifying the beam identifier to be measured, such as CSI-RS #[1,2,3,4,5,6]; sampling period: measurement time interval, such as 20 milliseconds; training mode: network-centralized training or terminal local training.

[0167] Step 703: The network device uses A transmit beams to send A reference signals from the second reference signal set to the terminal device.

[0168] Step 704: The terminal device measures the signal quality of each of the A reference signals through each of the C receiving beams, generates a first sample based on the A×C measured values, and trains the model based on the first sample to obtain the target AI model.

[0169] It should be noted that the network device can send A reference signals to the terminal device multiple times. Each time, the terminal device measures the signal quality of each of the A reference signals and obtains A×C measurement values, which can then be used to generate a training sample.

[0170] Optionally, the terminal device may perform model training based on a first sample after each first sample is generated; or, the terminal device may perform model training based on all the generated first samples after all the required first samples are generated.

[0171] It should be noted that the above text Figure 7 This implementation uses a terminal device for model training as an example. In practical applications, a network device can also perform model training. In the scenario where a network device performs model training, after the terminal device measures A×C values ​​in step 704, it can send these values ​​to the network device, which can then generate a first sample for model training. Similarly, the network device can generate a first sample each time it receives A×C measurement values ​​from the terminal device and perform model training based on that first sample; or, the network device can generate all the required first samples based on the measurement values ​​sent by the terminal device, and then use all the generated first samples for model training.

[0172] In this embodiment, a target AI model can be trained. The target AI model can predict the signal quality of A×C second beam pairs based on the measured signal quality of B×C first beam pairs. Optionally, the target AI model can predict the signal quality of A×C second beam pairs at a certain moment based on the measured signal quality of B×C first beam pairs at a certain moment; or, the target AI model can predict the signal quality of A×C second beam pairs at one or more future moments based on the measured signal quality of B×C first beam pairs at one or more moments.

[0173] It should be noted that after the target AI model is trained, it can be deployed on terminal devices or network devices to assist in beam selection.

[0174] The following is combined with Figure 8 The AI ​​model inference part of the embodiments of this application will be explained in detail.

[0175] Figure 8 This is a flowchart illustrating a model inference method provided in an embodiment of this application. This model inference method can be applied to a second device (also called a target device), which can be a network device or a terminal device. The first device and the second device can be the same device or different devices. Figure 8 As shown, the inference method of this model may include the following steps:

[0176] Step 801: The second device acquires B×C first measurement values.

[0177] The B×C first measurements correspond one-to-one with the B×C first beam pairs. The B×C first beam pairs are obtained by combining the B transmit beams and C receive beams corresponding to the reference signals in the first reference signal set.

[0178] Each of the B×C first measurements is obtained by measuring the reference signal corresponding to the transmit beam of the receiving beam pair in the first beam pair. That is, for any one of the B×C first measurements, the first measurement is the signal quality measured by using the reference signal transmitted by the transmit beam of the corresponding first beam pair through the receiving beam pair.

[0179] For example, when the second device is a terminal device, step 801 can be performed as follows: the terminal device measures each reference signal in the first reference signal set transmitted by the network device through each of the C receiving beams, and obtains the measurement result of each reference signal in the first reference signal set. The first reference signal set includes B reference signals. The measurement result of a reference signal includes C measurement information of that reference signal, and each of the C measurement information includes the beam identifier of the corresponding transmit beam, the beam identifier of the corresponding receive beam, and a first measurement value.

[0180] For example, if the second device is a network device, step 801 can be performed as follows: the network device receives the measurement results of each reference signal in the first set of reference signals sent by the terminal device.

[0181] The network device can send B reference signals from a first set of reference signals to the terminal device. For any reference signal in the first set of reference signals, the terminal device can measure the reference signal through each of C receiving beams to obtain the measurement result. Then, the terminal device can send the measurement result of each reference signal in the first set of reference signals to the network device. The measurement result of a reference signal includes C measurement information of the reference signal, each of the C measurement information including the beam identifier of the corresponding transmit beam, the beam identifier of the corresponding receive beam, and a first measurement value.

[0182] Step 802: The second device obtains A×C predicted values ​​through the target AI model based on the B×C first measured values.

[0183] The A×C predicted values ​​correspond one-to-one with the A×C second beam pairs. The A×C second beam pairs are obtained by combining the A transmit beams and C receive beams corresponding to the reference signals in the second reference signal set.

[0184] In some implementations, if the target AI model is trained in the first possible way in step 503 above, then step 802 can be implemented in the following way one; if the target AI model is trained in the second possible way in step 503 above, then step 802 can be implemented in the following way two.

[0185] Method 1: The second device generates a first matrix based on the B×C first measurement values; the first matrix is ​​input into the target AI model to obtain the fifth matrix output by the target AI model.

[0186] The first matrix is ​​a real number matrix with A rows and C columns. The row index i of the first matrix represents the beam identifier of each of the A transmit beams corresponding to the reference signals in the second reference signal set, and the column index j of the first matrix represents the beam identifier of each of the C receive beams. The first matrix includes the B×C first measurements. The subscript (i, j) of each of the B×C first measurements in the first matrix represents the beam identifier of the transmit and receive beams in its corresponding first beam pair.

[0187] The target AI model can obtain the prediction results for the second reference signal set based on the measurement results of the first reference signal set. Therefore, after inputting the first matrix into the target AI model, the fifth matrix output by the target AI model can include the prediction results for each of the A reference signals in the second reference signal set. The prediction result for a reference signal includes C prediction information, which includes the beam identifier of the transmitted beam corresponding to the reference signal, the beam identifier of the corresponding received beam, and the predicted value.

[0188] The fifth matrix is ​​a real matrix with A rows and C columns. The row index i of the fifth matrix represents the beam identifier of each of the A transmit beams corresponding to the reference signals in the second reference signal set, and the column index j represents the beam identifier of each of the C receive beams. The fifth matrix includes A×C prediction values. The index (i, j) of each of these A×C prediction values ​​in the fifth matrix corresponds to the beam identifier of the transmit and receive beams in its corresponding second beam pair.

[0189] Method 2: The second device generates a first matrix based on the B×C first measurement values; the element positions of the first matrix are randomized to obtain a second matrix; the second matrix is ​​input into the target AI model to obtain a third matrix output by the target AI model; the element positions of the third matrix are derandomized to obtain a fourth matrix.

[0190] The first matrix has already been explained in Method 1 above, and will not be repeated here.

[0191] After randomizing the element positions of the first matrix to obtain the second matrix, each element in the second matrix comes from the first matrix, but the positions of all elements are randomly rearranged.

[0192] The target AI model can obtain the prediction results for the second reference signal set based on the measurement results of the first reference signal set. Therefore, after inputting the second matrix into the target AI model, the third matrix output by the target AI model can include A×C predicted values. However, since the input second matrix is ​​obtained by randomizing the element positions of the first matrix, the positions of the A×C predicted values ​​in the output third matrix have not yet been recovered, and the second beam pairs corresponding to each predicted value cannot yet be confirmed. Therefore, the element positions of the third matrix can be derandomized to obtain a fourth matrix to recover the positions of the A×C predicted values.

[0193] The fourth matrix is ​​a real matrix with A rows and C columns. The row index i of the fourth matrix represents the beam identifier of each of the A transmit beams corresponding to the reference signals in the second reference signal set, and the column index j represents the beam identifier of each of the C receive beams. The fourth matrix includes A×C prediction values. The index (i, j) of each of these A×C prediction values ​​in the fourth matrix corresponds to the beam identifier of the transmit and receive beams in its corresponding second beam pair.

[0194] It should be noted that, in this embodiment, the A×C predicted values ​​can be used for beam selection. Since the A×C predicted values ​​correspond one-to-one with the A×C second beam pairs, beam selection can be performed relatively quickly and accurately from the A×C second beam pairs based on the A×C predicted values, thereby helping to improve measurement timeliness and communication efficiency.

[0195] For example, Figure 9 This is a schematic diagram of a model reasoning process provided in an embodiment of this application. For example... Figure 9 As shown, the first matrix can be randomized in terms of element position and then input into the target AI model. After obtaining the output data of the target AI model, the output data can be derandomized in terms of element position to obtain A×C predicted values ​​that correspond one-to-one with A×C second beam pairs.

[0196] Step 803: The second device determines the target prediction value corresponding to each of the A transmit beams from the A×C predicted values. The target prediction value corresponding to a transmit beam is the largest predicted value among all the predicted values ​​corresponding to that transmit beam. The device then determines the K transmit beams with the largest target prediction values ​​from the A transmit beams, where K is a positive integer. For any one of the K transmit beams, the device determines the L receive beams corresponding to the L largest predicted values ​​from all the predicted values ​​corresponding to that transmit beam, where L is a positive integer. Based on the beam identifiers of the K transmit beams and the beam identifiers of the L receive beams corresponding to each of the K transmit beams, the device generates K×L beam identifier pairs.

[0197] In this embodiment, firstly, K transmit beams with the largest target predicted values ​​are selected from A transmit beams. Then, L receive beams corresponding to the L largest predicted values ​​are selected from all predicted values ​​of each of the K transmit beams, thereby determining K×L beam identifier pairs. Each beam identifier pair in the K×L beam identifier pairs includes the beam identifier of one transmit beam from the K transmit beams and the beam identifier of one receive beam from the L receive beams corresponding to that transmit beam. In this way, not only are the TopK transmit beams determined, but also the TopL receive beams corresponding to each of the TopK transmit beams are determined. Subsequently, during the secondary confirmation in the P2 process, the optimal transmit beam and the optimal receive beam can be directly determined from the K×L beam pairs identified by the K×L beam identifier pairs. In this case, there is no need to perform receive beam confirmation in the P3 process, thereby reducing measurement latency, saving signaling overhead, reducing power consumption ratio, and thus helping to improve measurement timeliness and communication efficiency.

[0198] The following is combined with Figure 10 and Figure 11 The beam selection process is explained. Figure 10 This explanation will take the deployment of the target AI model on a terminal device as an example. Figure 11 The following example illustrates the deployment of a target AI model on a network device.

[0199] Figure 10 This is a flowchart illustrating a beam selection method provided in an embodiment of this application. For example... Figure 10 As shown, this beam selection method may include the following steps:

[0200] Step 1001: The terminal device sends inference capability information to the network device.

[0201] This inference capability information is used to report the AI ​​inference capabilities supported by the terminal device to the network device. For example, this inference capability information may include one or more of the following: the number of reference signals in a first reference signal set, the number of reference signals in a second reference signal set, the number of signals that can be monitored in a single instance, etc. The inference capability information may also include report content indication information, which includes information indicating support for reporting beam pairs.

[0202] For example, this inference capability information can be UE capability information. Report content indication information (i.e., supportedReportContents as described below) can be added to the UE capability information, with a value of 4 indicating that the terminal device supports beampup prediction.

[0203] For example, the relevant content in this UE capability information is as follows:

[0204] UEcapabilityInformation :: = SEQUENCE {

[0205] nonCriticalExtension SEQUENCE {

[0206] AI-BeamPredictionCapability {

[0207] ...

[0208] supportedReportContents BIT STRING(SIZE(4)) / / Predicted content: 0=RSRP, 1=RSRQ, 2=Beam identifier, 3=Confidence level, 4=Beam pair

[0209] ...

[0210] }

[0211] }

[0212] }

[0213] Step 1002: After receiving the inference capability information sent by the terminal device, the network device sends inference configuration information to the terminal device.

[0214] This inference configuration information is used for inference configuration. The inference configuration information may include reference signal configuration information. For example, the inference configuration information may include one or more of the following: reference signal type; signal measurement period; first reference signal information: specifying the beam identifier to be measured; second reference signal information: specifying the beam identifier to be predicted. Optionally, the inference configuration information may also include a preset report type, which is used to indicate the reporting of beam pairs.

[0215] For example, network devices can send this inference configuration information to terminal devices via radio resource control (RRC) messages. Since the terminal devices will subsequently need to report beam pairs, the network device can add a preset report type (i.e., beamPairPredictionSupported as described below) to the RRC message.

[0216] For example, the relevant content in the RRC message is as follows:

[0217] CSI-ReportConfig :: = SEQUENCE {

[0218] reportConfigId = 1,

[0219] ...

[0220] reportQuantity = CRI-RSRP-TopK, / / Report SetB measurements and predicted TopK beams

[0221] groupBasedBeamReporting = TRUE, / / Enable multi-beam reporting

[0222] beamPairPredictionSupported = TRUE, / / Reported predicted beam pairs

[0223] ...

[0224] numberOfTopKBeams = K, / / Report TopK predicted beams

[0225] ...

[0226] }

[0227] Step 1003: The network device sends each reference signal in the first set of reference signals to the terminal device.

[0228] The network device can transmit B reference signals from the first reference signal set using B transmit beams respectively.

[0229] Step 1004: The terminal device measures each reference signal in the first reference signal set through each of the C receiving beams to obtain B×C first measurement values.

[0230] The operation of step 1004 is similar to that of step 801 above, and will not be described again in this embodiment.

[0231] Step 1005: The terminal device obtains A×C predicted values ​​through the target AI model based on the B×C first measurement values; determines the target predicted value corresponding to each of the A transmit beams from the A×C predicted values, wherein the target predicted value corresponding to a transmit beam is the largest predicted value among all predicted values ​​corresponding to that transmit beam; determines the K transmit beams with the largest target predicted values ​​from the A transmit beams; for any one of the K transmit beams, determines the L receive beams corresponding to the largest L predicted values ​​from all predicted values ​​corresponding to that transmit beam; generates K×L beam identifier pairs based on the beam identifiers of the K transmit beams and the beam identifiers of the L receive beams corresponding to each of the K transmit beams.

[0232] The operation of step 1005 is similar to the operation of steps 802 to 803 above, and will not be described again in this embodiment.

[0233] Step 1006: The terminal device sends the K×L beam identifier pairs to the network device.

[0234] In some implementations, the terminal device can send the K×L beam identifier pairs to the network device in uplink control information (UCI). For example, the UCI field design can be as shown in Table 1 below. For example, the UCI format can be Physical Uplink Control Channel (PUCCH) format 3.

[0235] Table 1

[0236]

[0237] This application embodiment is only used as an example to illustrate the field design of UCI, and Table 1 above does not constitute a limitation on the embodiment of this application.

[0238] Optionally, after receiving the K×L beam identifier pairs sent by the terminal device, the network device can directly execute step 1007. Alternatively, the network device can first send scheduling information to the terminal device and then execute step 1007. The scheduling information includes the K×L beam identifier pairs, used to instruct the terminal device to measure the signal quality corresponding to the K×L beam pairs identified by the K×L beam identifier pairs.

[0239] In some implementations, this scheduling information can be downlink control information (DCI). For example, the DCI field design can be as shown in Table 2 below, where the TCI-State ID field can include the K×L beam identifier pairs. For example, the DCI format can be format 1_1.

[0240] Table 2

[0241]

[0242] The embodiments of this application are merely illustrative examples of the field design of DCI as shown in Table 2 above, and Table 2 above does not constitute a limitation on the embodiments of this application.

[0243] Step 1007: After receiving the K×L beam identifier pairs sent by the terminal device, the network device uses the transmit beams identified by the K×L beam identifier pairs to send reference signals to the terminal device.

[0244] The network device can send K reference signals to the terminal device one by one on the K transmit beams identified by the K×L beam identifiers.

[0245] Step 1008: The terminal device obtains K×L second measurement values ​​by using the reference signal transmitted by the corresponding transmit beam of the receiving beam measurement network device identified by each of the K×L beam identifier pairs. The K×L second measurement values ​​correspond one-to-one with the K×L beam identifier pairs.

[0246] For any one of the K reference signals sent by the network device, the terminal device measures the reference signal through each of the L receive beams corresponding to the transmit beam of the reference signal, and obtains L measurement information of the reference signal. The L measurement information includes the beam identifier of the transmit beam corresponding to the reference signal, the beam identifier of the corresponding receive beam, and the second measurement value.

[0247] Step 1009: The terminal device sends the K×L second measurement values ​​to the network device.

[0248] Step 1010: After receiving the K×L second measurement values, the network device uses the beam identifier corresponding to the largest second measurement value among the K×L second measurement values ​​as the optimal transmit beam and the beam identifier corresponding to the largest second measurement value among the K×L second measurement values ​​as the optimal receive beam.

[0249] In this embodiment, after receiving K×L beam identifier pairs from the terminal device, the network device can use the K transmit beams identified by the K×L beam identifier pairs to send reference signals to the terminal device. This allows the terminal device to obtain K×L second measurement values ​​corresponding one-to-one with each of the K×L beam identifier pairs. Based on these K×L second measurement values, the optimal transmit beam and optimal receive beam can be determined from the beam pairs identified by the K×L beam identifier pairs. In this case, the optimal transmit beam and optimal receive beam are determined in the P2 process, eliminating the need for the P3 process. This reduces measurement latency, saves signaling overhead, and lowers power consumption, thereby improving measurement timeliness and communication efficiency.

[0250] Figure 11 This is a flowchart illustrating a beam selection method provided in an embodiment of this application. For example... Figure 11 As shown, this beam selection method may include the following steps:

[0251] Step 1101: The network device sends each reference signal in the first set of reference signals to the terminal device.

[0252] The network device can transmit B reference signals from the first reference signal set using B transmit beams respectively.

[0253] Step 1102: The terminal device measures each reference signal in the first reference signal set through each of the C receiving beams to obtain B×C first measurement values.

[0254] Step 1103: The terminal device sends the B×C first measurement values ​​to the network device.

[0255] Step 1104: After receiving the B×C first measurement values ​​sent by the terminal device, the network device obtains A×C predicted values ​​through the target AI model; from the A×C predicted values, it determines the target predicted value corresponding to each of the A transmit beams, where the target predicted value corresponding to a transmit beam is the largest predicted value among all predicted values ​​corresponding to that transmit beam; from the A transmit beams, it determines the K transmit beams with the largest target predicted values; for any one of the K transmit beams, it determines the L receive beams corresponding to the L largest predicted values ​​among all predicted values ​​corresponding to that transmit beam; based on the beam identifiers of the K transmit beams and the beam identifiers of the L receive beams corresponding to each of the K transmit beams, it generates K×L beam identifier pairs.

[0256] Step 1105: The network device sends the K×L beam identifier pairs to the terminal device.

[0257] For example, a network device can send scheduling information to a terminal device, which includes the K×L beam identifier pairs, to instruct the terminal device to measure the K×L beam pairs identified by the K×L beam identifier pairs.

[0258] Step 1106: The network device uses the K×L beam identifiers to send a reference signal to the terminal device for the identified transmit beam.

[0259] Step 1107: The terminal device obtains K×L second measurement values ​​by using the reference signal transmitted by the corresponding transmit beam of the receiving beam measurement network device identified by each of the K×L beam identifier pairs. The K×L second measurement values ​​correspond one-to-one with the K×L beam identifier pairs.

[0260] Step 1108: The terminal device sends the K×L second measurement values ​​to the network device.

[0261] Step 1109: After receiving the K×L second measurement values, the network device uses the beam identifier corresponding to the largest second measurement value among the K×L second measurement values ​​as the optimal transmit beam and the beam identifier corresponding to the largest second measurement value among the K×L second measurement values ​​as the optimal receive beam.

[0262] In this embodiment, after the network device obtains K×L beam identifier pairs, it can use the K transmit beams identified by the K×L beam identifier pairs to send reference signals to the terminal device. The terminal device then obtains K×L second measurement values ​​corresponding one-to-one with each of the K×L beam identifier pairs. Based on these K×L second measurement values, the optimal transmit beam and optimal receive beam can be determined from the beam pairs identified by the K×L beam identifier pairs. In this case, the optimal transmit beam and optimal receive beam are determined in the P2 process, eliminating the need for the P3 process. This reduces measurement latency, saves signaling overhead, and lowers power consumption, thereby improving measurement timeliness and communication efficiency.

[0263] It should be understood that Figures 1 to 11 The flowcharts or scene diagrams shown are for illustrative purposes only and are not intended to limit the embodiments of this application to the examples illustrated. In fact, those skilled in the art can interpret the embodiments based on... Figures 1 to 11 The examples in the document can be transformed into equivalent ways to obtain more implementations.

[0264] The above text combined Figures 1 to 11 The beam selection method provided in the embodiments of this application is described in detail below. Figures 12 to 13The apparatus embodiments of this application are described in detail below. It should be understood that the communication apparatus of this application embodiments can execute the various beam selection methods of the foregoing embodiments of this application, that is, the specific working processes of the various products below can be referred to the corresponding processes in the foregoing method embodiments.

[0265] In the embodiments described above, the network device may execute some or all of the steps in each embodiment; the terminal device may execute some or all of the steps in each embodiment. These steps or operations are merely examples, and other operations or variations thereof may also be performed in the embodiments of this application. Furthermore, the steps may be executed in different orders as presented in the embodiments, and it is not necessary to execute all the operations in the embodiments of this application. The sequence number of each step 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.

[0266] Figure 12 This is a schematic block diagram of a communication device provided in an embodiment of this application. Figure 12 As shown, the communication device 1200 may include a communication module 1220. The communication module 1220 can implement corresponding communication functions, which can be internal communication functions of the communication device 1200 or communication functions between the communication device 1200 and other devices. Optionally, the communication module 1220 may also be referred to as a communication interface or transceiver module. Optionally, the communication device 1200 also includes a processing module 1210. The processing module 1210 can implement corresponding processing functions.

[0267] Optionally, the communication device 1200 further includes a storage module, which can be used to store instructions and / or data; the processing module 1210 can read the instructions and / or data in the storage module so that the communication device 1200 can implement the aforementioned method embodiments.

[0268] For example, the processing module 1210 is used to acquire multiple first measurement values, which correspond one-to-one with multiple first beam pairs. The multiple first beam pairs are obtained by combining B transmit beams corresponding to reference signals in a first reference signal set with multiple receive beams in pairs. Each of the multiple first measurement values ​​is obtained by measuring the reference signal corresponding to the transmit beam in the receive beam pair. B is a positive integer. Based on the multiple first measurement values, multiple predicted values ​​are obtained through the target AI model. The multiple predicted values ​​correspond one-to-one with multiple second beam pairs. The multiple second beam pairs are obtained by combining A transmit beams corresponding to reference signals in a second reference signal set with multiple receive beams in pairs. The first reference signal set is a subset of the second reference signal set. The multiple predicted values ​​are used for beam selection. A is an integer greater than or equal to 2.

[0269] For example, the processing module 1210 is used to generate a first matrix based on the plurality of first measurements, the first matrix including the plurality of first measurements; randomize the element positions of the first matrix to obtain a second matrix; input the second matrix into the target AI model to obtain a third matrix output by the target AI model; and derandomize the element positions of the third matrix to obtain a fourth matrix including the plurality of predicted values.

[0270] For example, the processing module 1210 is used to generate a first matrix based on the plurality of first measurements, the first matrix including the plurality of first measurements; input the first matrix into the target AI model to obtain a fifth matrix output by the target AI model, the fifth matrix including the plurality of predicted values.

[0271] For example, the processing module 1210 is used to acquire a plurality of first samples, each of which includes a plurality of third measurements, which correspond one-to-one with the plurality of second beam pairs; to perform random row masking on each of the plurality of first samples to obtain a plurality of second samples; and to perform model training based on the plurality of second samples to obtain a target AI model.

[0272] For example, the processing module 1210 is used to randomize the element positions of each of the plurality of second samples to obtain a plurality of third samples; for any one of the plurality of third samples, the third sample is input into the AI ​​model to obtain the first data output by the AI ​​model; the element positions of the first data are derandomized to obtain the second data; the parameters in the AI ​​model are adjusted according to the loss value between the second data and the first sample corresponding to the third sample, and the AI ​​model after parameter adjustment is the target AI model.

[0273] For example, the processing module 1210 is used to input any one of the multiple second samples into the AI ​​model to obtain the third data output by the AI ​​model; and adjust the parameters in the AI ​​model according to the loss value between the third data and the first sample corresponding to the second sample. The AI ​​model after parameter adjustment is the target AI model.

[0274] For example, the processing module 1210 is used to determine the target prediction value corresponding to each of the A transmit beams from the plurality of prediction values, wherein the target prediction value corresponding to the transmit beam is the largest prediction value among all prediction values ​​corresponding to the transmit beam; determine the K transmit beams with the largest target prediction values ​​from the A transmit beams, where K is a positive integer; for any one of the K transmit beams, determine the L receive beams corresponding to the largest L prediction values ​​from all prediction values ​​corresponding to that transmit beam, where L is a positive integer; and generate a plurality of beam identifier pairs based on the beam identifiers of the K transmit beams and the beam identifiers of the L receive beams corresponding to each of the K transmit beams.

[0275] In one possible design, the communication device 1200 may correspond to the terminal device in the above method embodiments, or to a component (such as a circuit, chip, or chip system) configured in the terminal device. The communication device 1200 may be used to perform the steps or processes performed by the terminal device in any of the above method embodiments.

[0276] For example, the communication module 1220 is used to send the plurality of beam identifier pairs to the network device.

[0277] The processing module 1210 is used to obtain multiple second measurement values ​​by using the reference signal transmitted by the corresponding transmit beam of the receiving beam measurement network device identified by each of the multiple beam identifier pairs. The multiple second measurement values ​​correspond one-to-one with the multiple beam identifier pairs.

[0278] The communication module 1220 is used to send the plurality of second measurement values ​​to the network device.

[0279] For example, the communication module 1220 is used to send inference capability information to the network device, the inference capability information including report content indication information, the report content indication information including information for indicating support for reporting beam pairs; and to receive inference configuration information sent by the network device, the inference configuration information including information for indicating reporting beam pairs.

[0280] For example, the communication module 1220 is used to receive scheduling information sent by a network device, the scheduling information including the plurality of beam identifier pairs.

[0281] In one possible design, the communication device 1200 may correspond to the network device in the above method embodiments, or to a component (such as a circuit, chip, or chip system) configured in the network device. The communication device 1200 may be used to perform the steps or processes performed by the network device in any of the above method embodiments.

[0282] For example, the communication module 1220 is used to send a reference signal to a terminal device using the plurality of beam identifiers for the identified transmit beam; and to receive a plurality of second measurement values ​​sent by the terminal device, the plurality of second measurement values ​​corresponding one-to-one with the plurality of beam identifiers.

[0283] The processing module 1210 is used to take the beam identifier corresponding to the largest second measurement value among the plurality of second measurement values ​​as the optimal transmit beam and the beam identifier corresponding to the largest second measurement value among the plurality of second measurement values ​​as the optimal receive beam.

[0284] For example, the communication module 1220 is used to send scheduling information to the terminal device, the scheduling information including the plurality of beam identifier pairs.

[0285] The above are merely examples; for detailed steps or procedures, please refer to the descriptions in the foregoing embodiments.

[0286] Figure 13 This is a schematic block diagram of a communication device 1300 provided in an embodiment of this application. The communication device 1300 may be a network device, a terminal device, or a circuit, chip, chip system, or processor, etc., used to implement the above methods. The communication device 1300 can be used to implement the methods described in the above method embodiments; for details, please refer to the descriptions in the above method embodiments.

[0287] like Figure 13 As shown, the communication device 1300 may include one or more processors 1310, which may also be referred to as processing units or processing modules, and can implement certain control functions. The processor 1310 may be a general-purpose processor or a dedicated processor, such as a baseband processor or a central processing unit. The baseband processor can be used to process communication protocols and communication data, while the central processing unit can be used to control the communication device 1300 (e.g., a base station, baseband chip, user equipment, user chip), execute software programs, and process data from the software programs.

[0288] In an alternative design, the processor 1310 may also store instructions and / or data that can be executed by the processor 1310 to cause the communication device 1300 to perform the methods described in the above method embodiments.

[0289] In another alternative design, the communication device 1300 may include a communication interface 1320 for implementing receiving and transmitting functions. For example, the communication interface 1320 may be a transceiver circuit, interface, interface circuit, or transceiver. The transceiver circuit, interface, interface circuit, or transceiver for implementing receiving and transmitting functions may be separate or integrated. The aforementioned transceiver circuit, interface, interface circuit, or transceiver may be used for reading and writing code / data, or it may be used for transmitting or relaying signals.

[0290] Optionally, the communication device 1300 may include one or more memories 1330, which may store instructions that can be executed on the processor 1310, causing the communication device 1300 to perform the methods described in the above method embodiments. Optionally, the memories 1330 may also store data. Optionally, the processor 1310 may also store instructions and / or data. The processor 1310 and the memories 1330 may be provided separately or integrated together.

[0291] It should be understood that, in one possible design, the steps in the method embodiments provided in this application can be implemented by integrated logic circuits in the processor's hardware or by instructions in software form. The steps of the methods disclosed in the embodiments of this application can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor. The software modules can reside in mature storage media in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, detailed descriptions are not provided here.

[0292] In one implementation, the communication device 1300 may correspond to the network device in the above method embodiments and may be used to execute the various steps and / or processes executed by the network device in the above method embodiments. The processor 1310 may be used to execute instructions stored in the memory 1330, and when the processor 1310 executes the instructions stored in the memory, the processor 1310 is used to execute the various steps and / or processes of the above method embodiments corresponding to the network device.

[0293] In another implementation, the communication device 1300 may correspond to the terminal device in the above method embodiments, and may be used to execute the various steps and / or processes executed by the terminal device in the above method embodiments. The processor 1310 may be used to execute the instructions stored in the memory 1330, and when the processor 1310 executes the instructions stored in the memory, the processor 1310 is used to execute the various steps and / or processes of the above method embodiments corresponding to the terminal device.

[0294] It should be understood that the aforementioned processing device can be one or more chips. For example, the processing device can be a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), a system-on-chip (SoC), a central processor unit (CPU), a network processor (NP), a digital signal processor (DSP), a microcontroller unit (MCU), a programmable logic device (PLD), or other integrated chips.

[0295] It is understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM). It should be noted that the memory used in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0296] According to the method provided in the embodiments of this application, this application also provides a chip system, which includes one or more processors for calling and executing instructions stored in memory, thereby causing the method described in the embodiments of this application to be executed. The chip system may be composed of chips or may include chips and other discrete devices.

[0297] The chip system may include input circuits or interfaces for transmitting information or data, and output circuits or interfaces for receiving information or data.

[0298] According to the method provided in the embodiments of this application, this application also provides a communication system, which includes the aforementioned network device and terminal device.

[0299] According to the method provided in the embodiments of this application, this application also provides a computer program product, which includes: computer program code, which, when run on a computer, causes the computer to execute the various steps or processes executed by the network device or terminal device in any of the foregoing method embodiments.

[0300] According to the method provided in the embodiments of this application, this application also provides a computer-readable storage medium storing program code, which, when run on a computer, causes the computer to execute the various steps or processes executed by the network device or terminal device in any of the foregoing method embodiments.

[0301] The computer-readable storage medium may be the aforementioned volatile memory or non-volatile memory, or it may include both volatile memory and non-volatile memory.

[0302] In the embodiments of this application, the terms and English abbreviations are exemplary examples given for ease of description and should not be construed as limiting the application in any way. This application does not preclude the possibility of defining other terms that can achieve the same or similar functions in existing or future agreements.

[0303] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in 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 these computer 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.

[0304] 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.

[0305] It should be understood that in the various embodiments of this application, the sequence number of each process 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.

[0306] In summary, the above descriptions are merely optional embodiments of the technical solutions of this application and are not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A beam selection method, characterized in that, Applied to a target device, the method includes: Multiple first measurement values ​​are acquired, and the multiple first measurement values ​​correspond one-to-one with multiple first beam pairs. The multiple first beam pairs are obtained by combining B transmit beams corresponding to the reference signals in the first reference signal set with multiple receive beams in pairs. Each of the multiple first measurement values ​​is obtained by measuring the reference signal corresponding to the transmit beam in the receive beam pair of the first beam pair. B is a positive integer. A first matrix is ​​generated based on the plurality of first measurement values, the first matrix including the plurality of first measurement values; Randomize the element positions of the first matrix to obtain the second matrix; The second matrix is ​​input into the target AI model to obtain the third matrix output by the target AI model; The third matrix is ​​derandomized to obtain a fourth matrix. The fourth matrix includes multiple predicted values, which correspond one-to-one with multiple second beam pairs. The multiple second beam pairs are obtained by combining A transmit beams corresponding to the reference signals in the second reference signal set with the multiple receive beams in pairs. The first reference signal set is a subset of the second reference signal set. The multiple predicted values ​​are used for beam selection, and A is an integer greater than or equal to 2.

2. The method as described in claim 1, characterized in that, After obtaining the fourth matrix, the process also includes: From the plurality of predicted values, determine the target predicted value corresponding to each of the A transmission beams, wherein the target predicted value corresponding to the transmission beam is the largest predicted value among all the predicted values ​​corresponding to the transmission beam; From the A transmitted beams, determine the K transmitted beams with the largest target prediction values, where K is a positive integer; For any one of the K transmit beams, determine the L receive beams corresponding to the largest L predicted values ​​from all predicted values ​​corresponding to the transmit beam, where L is a positive integer; Multiple beam identifier pairs are generated based on the beam identifiers of the K transmit beams and the beam identifiers of the L receive beams corresponding to each of the K transmit beams.

3. The method as described in claim 2, characterized in that, The target device is a terminal device, and after generating multiple beam identifier pairs, the process further includes: Send the plurality of beam identifier pairs to the network device; By measuring the reference signal transmitted by the network device using the corresponding transmit beam through each of the multiple beam identifier pairs, a plurality of second measurement values ​​are obtained, and the plurality of second measurement values ​​correspond one-to-one with the plurality of beam identifier pairs; The plurality of second measurement values ​​are sent to the network device.

4. The method as described in claim 3, characterized in that, Before acquiring the multiple first measurement values, the method further includes: Send inference capability information to the network device, the inference capability information including report content indication information, the report content indication information including information for indicating support for report beam pairs; The network device receives inference configuration information, which includes information for instructing the reporting of beam pairs.

5. The method as described in claim 3, characterized in that, Before measuring the reference signal transmitted by the network device using the corresponding transmit beam through the received beam identified by each of the plurality of beam identifier pairs, the method further includes: The system receives scheduling information sent by the network device, the scheduling information including the plurality of beam identifier pairs.

6. The method as described in claim 2, characterized in that, The target device is a network device, and after generating multiple beam identifier pairs, the process further includes: The plurality of beam identifiers are used to send reference signals to the terminal device for the identified transmit beams; Receive multiple second measurement values ​​sent by the terminal device, wherein the multiple second measurement values ​​correspond one-to-one with the multiple beam identifier pairs; The transmit beam identified by the beam identifier corresponding to the largest second measurement value among the plurality of second measurement values ​​is taken as the optimal transmit beam, and the receive beam identified by the beam identifier corresponding to the largest second measurement value among the plurality of second measurement values ​​is taken as the optimal receive beam.

7. The method as described in claim 6, characterized in that, Before sending a reference signal to the terminal device using the plurality of beam identifiers for the identified transmit beam, the method further includes: Send scheduling information to the terminal device, the scheduling information including the plurality of beam identifier pairs.

8. The method according to any one of claims 1 to 7, characterized in that, The method further includes: Multiple first samples are acquired, each of the multiple first samples including multiple third measurement values, and the multiple third measurement values ​​correspond one-to-one with the multiple second beam pairs; Each of the plurality of first samples is randomly row-masked to obtain a plurality of second samples; The target AI model is obtained by training the model based on the multiple second samples.

9. The method as described in claim 8, characterized in that, The step of training the model based on the plurality of second samples to obtain the target AI model includes: Randomize the element positions of each of the plurality of second samples to obtain a plurality of third samples; For any one of the plurality of third samples, input the third sample into the AI ​​model to obtain the first data output by the AI ​​model; The element positions of the first data are derandomized to obtain the second data. The parameters in the AI ​​model are adjusted based on the loss value between the second data and the first sample corresponding to the third sample. The AI ​​model after parameter adjustment is the target AI model.

10. A communication device, characterized in that, The communication device includes at least one processor coupled to a memory storing a program or instructions, the processor executing the program or instructions to cause the communication device to perform the method as described in any one of claims 1 to 9.

11. A communication system, characterized in that, The communication system includes the communication device as described in claim 10.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program or instructions that, when executed, cause a computer to perform the method as described in any one of claims 1 to 9.

13. A chip system, characterized in that, The chip system includes one or more processors, which are configured to retrieve and execute instructions stored in a memory, such that the method as described in any one of claims 1 to 9 is performed.

Images