Method and device for communication and computer readable medium

Abstract

The embodiment of the invention relates to a communication method, equipment and a computer readable medium. According to an embodiment of the present disclosure, a first device receives first information about a line of sight (LOS) state between the first device and a second device. The first device determines whether the LOS state is within a first state range based on the first information. The first device then selects a first model classified for the first state range in response to determining that the LOS state is within the first state range.

Description

Technical Field

[0001] Embodiments of the present disclosure relate generally to the field of communications, and in particular to methods, devices, and computer-readable media for artificial intelligence (AI) or machine learning (ML) model-assisted communications. Background Art

[0002] With the development of communication technology, multi-antenna transmission has been introduced. In multi-antenna transmission, multiple antennas on the transmitter side can be used to provide directivity by carefully adjusting the phase of each antenna element, and possibly the amplitude, that is, to concentrate the total transmit power in a certain direction (beamforming) or in more general cases at a specific location in space. Since higher power reaches the target receiver, this directivity can increase the achievable data rate and communication range. In this case, determining the preferred beam, channel or radio path for data transmission between devices is a prerequisite for making full use of multi-antenna transmission technology. In one solution, channel state information (CSI) measurement is performed between the terminal device and the network device, and the network device can schedule a preferred beam, channel or radio path for the terminal device based on a CSI report from the terminal device. In another solution, a beam management (BM) process is performed between the terminal device and the network device, and the network device can update the preferred beam based on a beam report received from the terminal device.

[0003] In some cases, AI or ML techniques are used to enhance communication performance, especially to enhance communication performance under multiple constraints. The selection of AI or ML models is a key aspect of utilizing AI or ML techniques. Summary of the invention

[0004] In general, example embodiments of the present disclosure relate to methods, devices, and computer-readable media for AI or ML model-assisted communication.

[0005] In a first aspect, a communication method is provided. In the method, a first device receives first information about a line of sight (LOS) state between the first device and a second device. The first device determines whether the LOS state is within a first state range based on the first information. Then, in response to determining that the LOS state is within the first state range, the first device selects a first model classified for the first state range.

[0006] In a second aspect, a communication method is provided. In the method, a second device receives a first CSI report including a plurality of compressed bits. The second device receives first information about a line of sight (LOS) state between the first device and the second device. The second device determines whether the LOS state is within a first state range based on the first information. Then, in response to determining that the LOS state is within the first state range, the second device selects a first model for decompressing the first CSI report, the first model being classified for the first state range.

[0007] In a third aspect, a terminal device is provided, comprising a processor and a memory coupled to the processor, wherein the memory stores instructions, which, when executed by the processor, cause the terminal device to perform the method of the first aspect.

[0008] In a fourth aspect, a network device is provided, comprising a processor and a memory coupled to the processor, wherein the memory stores instructions, which, when executed by the processor, cause the network device to perform the method of the second aspect.

[0009] In a fifth aspect, a computer-readable medium is provided, on which instructions are stored, and when the instructions are executed on at least one processor, the at least one processor performs the method of any one of the first to second aspects.

[0010] It should be understood that the abstract is not intended to identify the key or essential features of the exemplary embodiments of the present disclosure, nor is it intended to be used to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Some example embodiments will now be described with reference to the accompanying drawings, in which:

[0012] Figure 1 illustrates an example environment in which some embodiments of the present disclosure may be implemented;

[0013] Figure 2 illustrates a signaling process for CSI reporting according to some embodiments of the present disclosure;

[0014] FIG. 3A to FIG. 3D illustrates examples of input data and output data according to some embodiments of the present disclosure;

[0015] Figure 3E illustrates an example of an AI or ML model according to some embodiments of the present disclosure;

[0016] FIG. 4A to FIG. 4D illustrates a signaling process of a BM according to some embodiments of the present disclosure;

[0017] FIG. 5A to FIG. 5Billustrates example beam patterns according to some embodiments of the present disclosure;

[0018] FIG. 6A to FIG. 6C illustrates an example historical measurement instance as input to a model according to some embodiments of the present disclosure;

[0019] Figure 7 illustrates a flow chart of an example method implemented at a first device according to some embodiments of the present disclosure;

[0020] Figure 8 A flowchart illustrating an example method implemented at a second device according to some embodiments of the present disclosure; and

[0021] Fig. 9 A simplified block diagram of a device suitable for implementing an example embodiment of the present disclosure is illustrated.

[0022] Throughout the drawings, the same or similar reference numerals denote the same or similar elements. DETAILED DESCRIPTION

[0023] The principle of the present disclosure will now be described with reference to some example embodiments. It should be understood that these embodiments are only for illustrative purposes, and contribute to those skilled in the art to understand and implement the present disclosure, without implying any limitation on the scope of the present disclosure. The present disclosure described herein can be implemented in various ways except the way described below.

[0024] In the following description and claims, unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.

[0025] As used herein, the term "terminal device" refers to any device with wireless or wired communication capabilities. Examples of terminal devices include, but are not limited to, user equipment (UE), personal computers, desktops, mobile phones, cellular phones, smartphones, personal digital assistants (PDAs), portable computers, tablets, wearable devices, Internet of Things (IoT) devices, ultra-reliable and low-latency communication (URLLC) devices, Internet of Everything (IoE) devices, machine type communication (MTC) devices, vehicle-mounted devices for V2X communication (where X refers to pedestrians, vehicles, or infrastructure / network), devices for integrated access and backhaul (IAB), small data transmission (SDT), mobility, multicast and broadcast services (MBS), positioning, dynamic / flexible duplexing in commercial networks, reduction Capabilities (RedCap), space vehicles or air vehicles in non-terrestrial networks (NTN), including satellites and high altitude platforms (HAPs), including unmanned aerial systems (UAS), extended reality (XR) devices (including different types of reality, such as augmented reality (AR), mixed reality (MR) and virtual reality (VR)), unmanned aerial vehicles (UAV) (commonly known as drones, which are aircraft without any human pilots), devices on high-speed trains (HST), or image capture devices (such as digital cameras), sensors, gaming devices, music storage and playback devices, or Internet devices that support wireless or wired Internet access and browsing, etc. "Terminal devices" may also have "multicast / broadcast" features to support public safety and critical missions, V2X applications, transparent IPv4 / IPv6 multicast delivery, IPTV, smart TV, radio services, wireless software delivery, group communication and Internet of Things applications. It may also incorporate one or more user identity modules (SIMs), which are called multi-SIMs. The term "terminal device" may be used interchangeably with UE, mobile station, user station, mobile terminal, user terminal or wireless device.

[0026] As used herein, the term "network equipment" refers to equipment that can provide or host a cell or coverage area, and terminal devices in the cell or coverage area can communicate. Examples of network equipment include, but are not limited to, Node B (NodeB or NB), evolved Node B (eNodeB or eNB), next generation Node B (gNB), transmit receive point (TRP), remote radio unit (RRU), radio head (RH), remote radio head (RRH), IAB node, low power node (such as femto node, micro node), reconfigurable intelligent surface (RIS), network controlled repeater, etc.

[0027] The terminal device or network device may have artificial intelligence (AI) or machine learning capabilities. It usually includes a model for a specific function trained from a large amount of collected data and can be used to predict certain information. The terminal or network device can operate in several frequency ranges, such as FR1 (410MHz-7125 MHz), FR2 (24.25GHz-71GHz), 71GHz-114 GHz, bands above 100GHz, and terahertz (THz). It can further operate on licensed / unlicensed / shared spectrum. In the multi-radio dual connection (MR-DC) application scenario, the terminal device can have more than one connection with the network device. The terminal device or network device can operate in full-duplex, flexible duplex, and cross-division duplex modes.

[0028] The network equipment may have network energy saving, self-organizing network (SON) / minimization of drive test (MDT) functions. The terminal may have a power saving function.

[0029] The embodiments of the present disclosure may be implemented in a test device, for example, a signal generator, a signal analyzer, a spectrum analyzer, a network analyzer, a test terminal device, a test network device, or a channel simulator.

[0030] Embodiments of the present disclosure may be performed according to any generation of communication protocols currently known or to be developed in the future. Examples of communication protocols include, but are not limited to, first generation (1G), second generation (2G), 2.5G, 2.75G, third generation (3G), fourth generation (4G), 4.5G, fifth generation (5G) communication protocols, 5.5G, advanced 5G networks, or sixth generation (6G) networks.

[0031] In one embodiment, the terminal device may be connected to a first network device and a second network device. One of the first network device and the second network device may be a primary node, and the other may be a secondary node. The first network device and the second network device may use different radio access technologies (RATs). In one embodiment, the first network device may be a first RAT device, and the second network device may be a second RAT device. In one embodiment, the first RAT device is an eNB, and the second RAT device is a gNB. Information related to different RATs may be transmitted to the terminal device from at least one of the first network device and the second network device. In one embodiment, the first information may be transmitted to the terminal device from the first network device, and the second information may be transmitted to the terminal device directly or via the first network device from the second network device. In one embodiment, information related to the configuration of the terminal device configured by the second network device may be transmitted from the second network device via the first network device. Reconfiguration-related information of the terminal device configured by the second network device may be transmitted to the terminal device directly or via the first network device from the second network device.

[0032] As used herein, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. The term "including" and its variations are to be understood as open-ended terms, meaning "including but not limited to". The term "based on" is to be understood as "based at least in part on". The terms "one embodiment" and "embodiment" are to be understood as "at least one embodiment". The term "another embodiment" is to be understood as "at least one other embodiment". The terms "first", "second", etc. may refer to different or identical objects. Other definitions, both explicit and implicit, may be included below.

[0033] In some examples, values, procedures, or devices are referred to as "best," "lowest," "highest," "minimum," "maximum," etc. It will be appreciated that such descriptions are intended to indicate that a selection may be made among a number of functional alternatives employed, and that such selection is not necessarily better, lesser, higher, or more desirable than other selections.

[0034] In the present disclosure, the term "beam" may be represented by "beam pair", "beam pair link", "Tx beam", "Rx beam", "resource", "resource set" or "resource setting", "resource configuration". The term "beam quality" used herein refers to "RSRP", "SINR", "RSSI" or "RSRQ" measured on the corresponding resource or via the corresponding beam. The terms "precoder", "precoding", "precoding matrix", "beam", "spatial relationship information", "spatial domain transmission filter", "spatial domain filter", "spatial parameters", "spatial relationship information", "spatial relationship information", "TPMI", "precoding information", "precoding information and number of layers", "precoding matrix indicator (PMI)", "precoding matrix indicator", "transmission precoding matrix indication", "precoding matrix indication", "TCI state", "transmission configuration indicator", "quasi co-location (QCL)", "quasi co-location", "QCL parameters" and "spatial relationship" may be used interchangeably. The terms "SRI", "SRS resource set index", "UL TCI", "UL spatial domain filter", "UL beam", "joint TCI" may be used interchangeably. The terms "candidate", "prediction", "target", and "potential" may be used interchangeably. The terms "group", "subset", and "set" may be used interchangeably. The terms "relationship", "mapping", and "correspondence" may be used interchangeably. The terms "recovery", "decoding", "reconstruction", and "decompression" may be used interchangeably, and the terms "encoding" and "compression" may be used interchangeably. The terms "compressed CSI", "compressed PMI", and "compressed CSI / PMI" may be used interchangeably.

[0035] In the present disclosure, the term “CSI” or “CSI report” may include at least one of the following:

[0036] ● Channel quality indicator (CQI), precoding matrix indicator (PMI), CSI-RS resource indicator (CRI), SS / PBCH block resource indicator (SSBRI), layer indicator (LI), rank indicator (RI), capability [set] index, panel index, wideband / subband PMI, wideband / subband CQI, wideband / subband amplitude / phase, wideband / subband co-phase, index of discrete Fourier transform (DFT) / discrete cosine transform (DCT) basis in space / frequency / time domain

[0037] or linear combinations;

[0038] ●Signal-to-noise ratio (SNR), reference signal received power (RSRP), signal-to-interference-plus-noise ratio (SINR), received signal strength indicator (RSSI), reference signal received quality (RSRQ), supported modulation schemes, supported coding rates, supported efficiencies;

[0039] A Type I single-panel codebook, a Type I multi-panel codebook, a Type II codebook, a Type II port selection codebook, an enhanced Type II codebook, an enhanced Type II port selection codebook, and a further enhanced Type II port selection codebook;

[0040] ●The channel matrix or the transformation domain of the channel matrix, including the spatial frequency domain, the time frequency domain, the angular delay domain, etc.

[0041] The term "circuit" as used herein may refer to a combination of hardware circuits and / or hardware circuits and software. For example, a circuit may be a combination of analog and / or digital hardware circuits and software / firmware. As another example, a circuit may be any part of a hardware processor and software, including (multiple) digital signal processors, software, and (multiple) memories, which work together to enable a device (such as a terminal device or a network device) to perform various functions. In yet another example, a circuit may be a hardware circuit and / or a processor (such as a microprocessor or a part of a microprocessor) that requires software / firmware to operate, but the software may not exist when not needed. As used herein, the term "circuit" also encompasses the implementation of only a hardware circuit or (multiple) processors or a part of a hardware circuit or (multiple) processors and its (their) accompanying software and / or firmware.

[0042] As mentioned above, in some cases, AI or ML models have been introduced to enhance communication performance, especially under multiple constraints (e.g., multi-antenna processing or multiple-input multiple-output (MIMO) processing). In this case, the selection of AI or ML models should be further considered. For example, for different channel characteristics, AI or ML models trained with different data samples can achieve different communication performance. Generally, the state of line of sight (LOS) between the communicating parties is an important factor affecting the channel characteristics.

[0043] The example embodiments of the present disclosure propose a mechanism for artificial intelligence or machine learning model-assisted communication, in which a first device receives first information about a line of sight (LOS) state between the first device and a second device. The first device determines whether the LOS state is within a first state range based on the first information. Then, in response to determining that the LOS state is within the first state range, the first device selects a first model classified for the first state range.

[0044] In this way, the first device for communication (e.g., a terminal device or a network device) determines an AI or ML model for auxiliary communication based on the LOS state. In this way, the AI ​​or ML model can be trained in a targeted and efficient manner, and the selected AI or ML model is more adaptive. Additionally, the size, type, or time span of the input and output data can depend on the selected AI or ML model.

[0045] Figure 1 An example environment 100 is illustrated in which example embodiments of the present disclosure may be implemented.

[0046] Environment 100 may be part of a communication network including terminal device 110, network device 120, and location server 130. Without limitation, Figure 1 The terminal devices and network devices in the system can perform data transmission in different spatial directions based on the multi-beam capability.

[0047] In some embodiments, the location server 130 may include a location management server (LMF). The location server 130 may provide the terminal device 110 and the network device 120 with first information of the LOS state between the terminal device 110 and the network device 120. Additionally or alternatively, the location server 130 may also provide first information for an AI or ML model in the terminal device 110 and another AI or ML model in the network device 120. Additionally or alternatively, the MI or ML model used for positioning in the location server may also provide first information for the terminal device 110 and the network device 120, or for the AI ​​or ML model in the terminal device 110 and another AI or ML model in the network device 120. Additionally or alternatively, the first information of the LOS state may also be directly exchanged between the terminal device 110 and the network device 120, which will be discussed in the following embodiments.

[0048] It should be understood that the number of terminal devices and network devices shown in environment 100 is for illustrative purposes only and does not imply any limitation on the scope of the present disclosure. In some embodiments, environment 100 may include additional terminal devices to communicate information with additional network devices. In some embodiments, environment 100 may also include an AI or ML server. The AI ​​or ML server may provide AI or ML capabilities for terminal devices 110, network devices 120, and location servers 130. AI or ML capabilities may include: AI or ML model download, AI or ML model upload, AI or ML model update, AI or ML model training, and AI or ML model reasoning. The provided AI or ML model may be provided with first information about the LOS state. Additionally or alternatively, the provided AI or ML model may also be provided with first information about the LOS state.

[0049] Communications in environment 100 may follow any appropriate communication standards or protocols that already exist or will be developed in the future, such as Universal Mobile Telecommunications System (UMTS), Long Term Evolution (LTE), Advanced LTE (LTE-A), Fifth Generation (5G) New Radio (NR), Wireless Fidelity (Wi-Fi), and Worldwide Interoperability for Microwave Access (WiMAX) standards, and employ any appropriate communication technology, including, for example, Multiple Input Multiple Output (MIMO), Orthogonal Frequency Division Multiplexing (OFDM), Time Division Multiplexing (TDM), Frequency Division Multiplexing (FDM), Code Division Multiplexing (CDM), Bluetooth, ZigBee, as well as Machine Type Communication (MTC), Enhanced Mobile Broadband (eMBB), Massive Machine Type Communication (mMTC), Ultra-Reliable Low Latency Communication (URLLC), Carrier Aggregation (CA), Dual Connectivity (DC), and New Radio Unlicensed (NR-U) technology.

[0050] AI or ML models can be used in the CSI reporting process and the BM process. The AI ​​or ML models used for CSI reporting will refer to Figure 2 Let's discuss.

[0051] Figure 2 The signaling process 200 of CSI reporting according to some embodiments of the present disclosure is illustrated. For the purpose of discussion, reference will be made to Figure 1 Process 200 is described.

[0052] Typically, in the CSI reporting process, the network device 120 sends a downlink (DL) beam reference signal (RS) for the terminal device 110 to measure CSI information. The DL RS may include at least one of a CSI-RS, a synchronization signal and a physical broadcast channel block (SSB), and a demodulation reference signal (DMRS). After measuring the CSI information, the terminal device 110 sends a CSI measurement report to the network device 120. The CSI measurement report includes an indication of a preferred beam for the terminal device 110, for example, an indication of a DL RS quasi-co-located with the preferred beam. Specifically, as mentioned above, the CSI measurement report may include CQI, PMI, CRI, SSB resource indicator (SSBRI), layer indicator (LI), rank indicator (RI), L1-RSRP, L1-SINR, or capability (set) index. The network device 120 may schedule communication resources for the terminal device 110 based on the CSI measurement report.

[0053] In the present disclosure, the terminal device 110 selects an AI or ML model for encoding / compressing a CSI measurement report based on the LOS state between the terminal device 110 and the network device 120. The network device 120 selects another AI or ML model based on the LOS state for decoding / decompressing the CSI measurement report from the terminal device 110. The details are discussed below. In some embodiments, compression or encoding may be used for the entire CSI report or a portion of the CSI report to obtain compressed bits. For example, one or more of the following are compressed or encoded: CQI, PMI, CRI, SSBRI, LI, RI, L1-RSRP, L1-SINR, or capability (set) index. In particular, compression or encoding is applied to the PMI in the CSI report to obtain multiple compressed bits.

[0054] At 201, the terminal device 110 and the network device 120 may exchange capability information. The capability information indicates whether the terminal device 110 or the network device 120 supports the selection of an AI or ML model based on the LOS state. For example, the network device 120 may notify the network device 120 whether the selection of an AI or ML model is supported through an RRC configuration. The terminal device 110 may notify the terminal device 120 whether the selection of an AI or ML model is supported through a UE capability report. Additionally or alternatively, the configuration or UE capability report may also indicate the start / end / request to start / request to end AI / ML model selection based on LOS / NLOS information. In some embodiments, the LOS or NLOS information is provided to an entity that supports the selection of an AI or ML model or is deployed with an AI or ML model.

[0055] At 203, the terminal device 110 receives first information about the LOS state between the terminal device 110 and the network device 120. Additionally or alternatively, the network device 110 also receives the first information. In the present disclosure, the first device may refer to the terminal device 110 or the network device 120, and then the second device may refer to the network device or the terminal device accordingly. For example, the first device that receives the first information about the LOS state between the first device and the second device may refer to:

[0056] • The terminal device 110 receives first information about the LOS status between the terminal device 110 and the network device 120; or

[0057] The network device 120 receives first information about the LOS status between the network device 120 and the terminal device 110 .

[0058] In some embodiments, the first information about the LOS state is received from the location server 130. In some embodiments, the terminal device 110 may receive the first information from the network device 120. Additionally or alternatively, the network device 120 may receive the first information from the terminal device 110. Additionally or alternatively, the terminal device 110 and the network device 120 may receive the first information from the location server 130. Additionally or alternatively, the terminal device 110 and the network device 120 may also provide the first information to the location server 130 for UE positioning. For clarity of discussion, refer to Figure 3A-3E The first information about LOS or non-LOS (NLOS) and the exchange of the first information are further discussed. Using the received first information, the first device (terminal device 110 or network device 120) can select a corresponding AI / ML model. In some embodiments, the AI ​​or ML model server can also provide the first information to the terminal device 110, the network device 120, and the location server 130. Additionally or alternatively, the first information can be provided to the AI ​​or ML model server from any one of the terminal device 110, the network device 120, and the location server 130.

[0059] The LOS state indicates the channel condition between the first device and the second device. For example, a LOS state having a first value (e.g., "1") indicates that the radio path between the first device and the second device is "line-of-sight" within a certain radio frequency range. In addition, a LOS state having a second value (e.g., "0") indicates that the radio path between the first device and the second device is "non-line-of-sight" within a certain radio frequency range. In some embodiments, the LOS state indicating "line of sight" spans a first state range, while another LOS state indicating "non-line of sight" spans a second state range that is at least partially different from the first state range. Additionally or alternatively, the LOS state is expressed by a probability value or possibility, such as a scaling value of 0, 0.1, 0.2, ..., 1. Additionally or alternatively, the LOS state can be indicated by multipath information, such as the number of paths, the number of dominant paths, the ratio of the strongest path to the second strongest path, etc.

[0060] At 205 , the terminal device 110 may perform CSI measurement or channel measurement on the DL RS received from the network device 120 .

[0061] At 207, the terminal device 110 determines whether the LOS state is within a first state range based on the first information. Then, in response to determining that the LOS state is within the first state range, the terminal device 110 selects a first model classified for the first state range to encode / compress the CSI measurement report. In some embodiments, the first information may be a hard value, such as 0 or 1, indicating LOS or NLOS, as discussed above. Additionally or alternatively, the first information may be a soft value, such as a probability or a possible value. Additionally or alternatively, the first information may also be the number of multipaths, or other parameters as discussed above. In some other embodiments, the first information may be other parameters associated with channel conditions.

[0062] At 208, the network device 120 determines whether the LOS state is within the first state range based on the first information. Then, in response to determining that the LOS state is within the first state range, the network device 120 selects another first model classified for the first state range to decode / decompress the CSI measurement report.

[0063] In some embodiments, the first device (terminal device 110 or network device 120) may select a model from at least two models, the at least two models including a first model and a second model classified for a second state range. The second state range is at least partially different from the first state range. In some embodiments, the first state range is specific to a LOS communication environment, and the second state range is specific to a non-LOS (NLOS) communication environment. Additionally or alternatively, if the LOS state is within the second state range, the first device may select a second model classified for the second state range. Additionally or alternatively, at least two models also include a third model classified for a third state range, and the third state range is specific to a predefined default communication environment. In some embodiments, the third state range does not overlap with the first state range and the second state range, or partially overlaps with the first state range or the second state range; if the LOS state is not within the first state range and the second state range, the first device may select a third model classified for the third state range. The third state range may be specific to a predefined default communication environment. In some embodiments, the first device may select the first model by activating the first model or by deactivating the corresponding other AI / ML models. Additionally or alternatively, the first device may change the model state of the AI / ML model (e.g., from "on" to "off" or from "off" to "on"). Additionally or alternatively, if multiple AI / ML models are configured for LOS / NLOS, additional model index information may be provided.

[0064] In some embodiments, the first model may be an AI or ML model for CSI compression or CSI decompression. For example, the first device as the terminal device 110 may select an AI or ML model for CSI compression, and the first device as the network device 120 may select an AI or ML model for CSI decompression. Additionally or alternatively, the first model may be a Bayesian model, a clustering model, an SVM model, or any other statistical model.

[0065] The selection of the first model based on the first information about the LOS state may include: selecting an identification (ID) of the first model to be applied; determining the type and size of input data for the first model; determining historical data as input to the first model; and determining the output of the first model. The output of the first model may include the type of output, the size of the output, and the effective time of the output. In some embodiments, models classified for different LOS state ranges are suitable for different channel conditions, and the input, output, training data, algorithm, model type or other parameters of the model are also different accordingly. In addition, the selection of the model may occur in multiple stages of the life cycle of the model, for example, the selection of the model may occur in the training of the model, the verification of the model, the testing of the model and / or the reasoning of the model.

[0066] In the example, the available AI / ML models are categorized or classified as "used in a LOS environment or a first state range" and "used in a NLOS environment or a second state range". Additionally or alternatively, the available AI / ML models may be classified depending on different values / ranges of LOS / NLOS likelihood or number of paths, etc. Different AI / ML models may be characterized by: different types (CNN, RNN, Transformer, Inception, etc.), number of layers, branches, real-valued or complex-valued parameters, number of compressed bits, input / output size, etc. The differences between AI or ML models may also be reflected in other parameters and hyperparameters used to describe the AI ​​or ML mode (or, for aligning encoders / decoders). Other parameters and hyperparameters include: number or maximum number of layers, number or maximum number of hidden layers, layer type, layer shape (i.e. filter size, number of channels / filters), number and maximum number of neurons in each layer, number and maximum number of neurons and connections between layers, learning rate, loss function, cost function, activation function, mini-batch size, number of training iterations, momentum, number of hidden units, weight decay, activation sparsity, nonlinearity, weight initialization, regularization constant, number of epochs, number of branches in decision trees, number of clusters in clustering algorithms, and other hyperparameters.

[0067] In addition, within different LOS state ranges, the number of compressed bits output by the first model selected by the first device as the terminal device 110 may be different. For example, for LOS conditions, the channel characteristics may be simpler than NLOS conditions, so that a smaller effective load in the CSI report is sufficient to indicate the LOS channel. In this way, a smaller number of compressed bits can be used to reduce overhead. In one example, fewer or fewer bits are required in the LOS scenario. Additionally or alternatively, a CSI report format with less effective load or overhead can be used to send compressed bits to a first device or a second device as a network device 120. In the present disclosure, the entire CSI report or a portion of the CSI report may be compressed or encoded to obtain compressed bits. For example, one or more of the following are compressed or encoded: CQI, PMI, CRI, SSBRI, LI, RI, L1-RSRP, L1-SINR or capability (set) index. In particular, compression or encoding is applied to the PMI in the CSI report to obtain multiple compressed bits.

[0068] Additionally or alternatively, the report includes information about whether the compressed bits are obtained by the LOS AI model or the NLOS AI model. In some embodiments, the first device may send a CSI report to the second device. If the first device selects a model classified for a LOS communication environment, the CSI report (which may also be referred to as a first CSI report) may include a first number of compressed bits. Additionally or alternatively, if the first device selects a model classified for a NLOS communication environment, the CSI report (which may also be referred to as a second CSI report) may include a second number of compressed bits. As described above, the first number of compressed bits may be less than the second number of compressed bits. In addition, the first CSI report may include a first indication, and the first indication indicates that the first number of compressed bits is determined by a model classified for a LOS communication environment. Additionally or alternatively, the second CSI report may include a second indication, and the second indication indicates that the second number of compressed bits is determined by a model classified for a NLOS communication environment.

[0069] Further, in one solution, the compression bits are calculated by the following formula:

[0070] Number of compression bits = 2*Nt*Ns*α*Q (1)

[0071] Where N t is the number of ports, N s is the number of subbands, α is the compression ratio, and Q is the quantization bits.

[0072] In some embodiments, the compression ratios of models classified for different state ranges are different. In one example, a higher ratio can be used in a LOS scenario (e.g., LOS: 1 / 16; NLOS: 1 / 4). In another example, in a LOS situation, a higher ratio is used to compress PMIs with fewer ranks, and the rank can be fixed to 1 or 2.

[0073] Additionally or alternatively, the first input type of the first model classified for LOS communication is different from the second input type of the second model classified for NLOS communication. For example, the first input may include an angular delay domain channel matrix because multipath components can be removed without reducing compression performance. Correspondingly, the second input type may include a spatial frequency domain channel matrix.

[0074] Additionally or alternatively, the first input type may also include an eigenvector of the (original) channel matrix, channel information (e.g., matrix, eigenvector, precoder matrix indication, PMI) and PMI. The second input type may also include the (original) channel matrix, the archiving of the received reference signal (e.g., PDP, amplitude, phase, etc.).

[0075] Additionally or alternatively, a first input size for a first model classified for LOS communication is different from a second input size for a second model classified for NLOS communication. In one example, the first input size is less than or smaller than the second input size. Specifically, the first input size or the second input size can be any of the following: the number of subbands, the number of rows or columns of a channel matrix, the bandwidth of a subband, the number of beams, the number of discrete Fourier transform (DFT) vectors, the number of eigenvalues ​​or eigenvectors, the rank value of a channel.

[0076] Additionally or alternatively, the historical channel information used as input for the first model may also be different from the historical channel information used as input for the second model. Typically, channel characteristics in NLOS situations may change rapidly. For the second model classified for NLOS communication environments, less historical channel information may be used. For the first model classified for LOS communication environments, more historical channel information may be used. In one example, the input to the first model may include M1 latest measurement instances, such as M1 channel matrices. The input to the second model may include M2 ​​latest measurement instances, such as M2 channel matrices, and M1 is greater than M2. Additionally or alternatively, M1 is equal to or less than M2, but the first interval between adjacent measurement instances of the first model is greater than the second interval between adjacent measurement instances of the second model, such that M1*first interval is greater than M2*second interval. The input domain parameters may be further described in FIG. 6A to FIG. 6C In general, LOS channel conditions require less information in the time domain, frequency domain, beam / angle domain, and delay domain for inference.

[0077] Additionally, data processing for different models may also be different. Data processing includes at least cleaning raw data samples, normalizing / truncating input / output sizes. Data samples used to train models (e.g., first models and second models) are marked with LOS or NLOS information when the data samples are collected. For a first model classified for a LOS communication environment, the first model is trained based on data samples marked with LOS. Additionally or alternatively, the reasoning input for the first device needs to remove data samples marked with "NLOS". For a second model classified for a NLOS communication environment, the second model is trained based on data samples marked with NLOS. Additionally or alternatively, the reasoning input for the first device needs to remove data samples marked with "LOS". Additionally or alternatively, for an AI / ML model trained for a specific LOS possibility (value or range), the input for reasoning of the AL / ML model only includes data samples within a specific range.

[0078] Further, for the input or output of the model of CSI measurement report decompression at the network device 120 (or the first device or the second device as the network device 120 as mentioned above), the input size of the model used as a decoder is based on the reported compressed bits, i.e., the output of the encoder (e.g., the first number of compressed bits or the second number of compressed bits as mentioned above). In addition, the output size of the decoder is also based on the input type / size of the encoder.

[0079] Additionally, some other parameters of the first model and the second model may also be different. In some embodiments, the quantization parameters are different. The quantization parameters may include a quantization type (e.g., uniform quantization or vector quantization), a quantization step size (e.g., larger in a LOS communication environment), and a quantization bit (e.g., smaller in a LOS communication environment). In some embodiments, time-related parameters are different, such as the periodicity of reporting CSI measurements (e.g., larger in a LOS communication environment). In some embodiments, calculation-related parameters are different, and the calculation-related parameters include complexity and time (e.g., smaller in a LOS communication environment). In some embodiments, the loss function or convergence condition is different. For example, the first criterion for the first loss function of the first model is looser than the second criterion for the second loss function of the second model. In another example, the first constraint for the first model is looser than the second constraint for the second model.

[0080] Additionally or alternatively, if no LOS / NLOS information is provided, or a special state of LOS / NLOS is provided, a default AI / ML model may be employed. In some embodiments, default values ​​or predefined values ​​of the above parameters are applied. Additionally or alternatively, the default configuration may be a configuration for non-AI based CSI feedback.

[0081] In some embodiments, the terminal device 110 and the network device 120 are provided with an association between a model or at least one of the parameters associated with the model and a LOS state range. The first device can select the first model by determining at least one of the first ID, the number of compressed bits, the compression ratio, the input data size, the pattern of the beam set for beam measurement, the number of predicted beams, or the effective load size of the beam report for the first model based on the first information and the association. For example, the first device (terminal device 110 or network device 120) can select the first model or the second model based on the association, and in some embodiments, the provided association is predefined. Additionally or alternatively, the provided association is configured by the first device as the network device 120. Additionally or alternatively, the provided association is received by the terminal device 110 in response to a request for association sent by the terminal device 110 to the network device 120. In some embodiments, the association can be expressed as Table 1 below.

[0082] Table 1

[0083]

[0084] There may be more parameters for selecting the first model or the second model, which may be defined in the association. Without any limitation, Table 1 lists only a portion of these parameters. In addition, another model listed in the above table is a general AI or ML model that can be trained based on data samples labeled with LOS and data samples labeled with NLOS.

[0085] Additionally or alternatively, the first device selects the first model by determining at least a portion of the above parameters based on the first information and the scaling value. In one example, the scaling value can be used based on the LOS likelihood. In some embodiments, the scaling function is N=f(LOS likelihood)*N_basic, if the LOS likelihood is 0.5, N_basic=N1 is the number of compressed bits configured for LOS=1, f(x) is a function of likelihood=1 / x, then N=2*N1. The first device can determine at least one of the following parameters based on the scaling value: the number of compressed bits, the compression ratio, the input data size, the number of beams used for beam measurement, the predicted number of beams, or the payload size of the beam report.

[0086] At 209, the terminal device 110 compresses or encodes the CSI report with the aid of the above embodiments. For example, the terminal device 110 compresses or encodes the channel matrix or the precoding matrix. At 211, the terminal device 110 sends the compressed / encoded CSI measurement report to the network device 120, for example, the terminal device 110 sends the CSI measurement report in a compressed bit manner via uplink control information (UCI) or a physical uplink shared channel (PUSCH). At 212, the network device 120 decompresses or decodes the CSI measurement report as described above. For example, the network device 120 recovers the channel matrix / precoding matrix from the compressed bits, and then the network device 120 can schedule communication resources for the terminal device 110 and perform data transmission accordingly.

[0087] As described above, based on the first information of the LOS state, the model can be selectively used for CSI reporting. In this way, the CSI prediction accuracy can be enhanced accordingly, and the payload of the CSI measurement report can be significantly reduced.

[0088] As mentioned above, the first information may be obtained or provided in a variety of ways. FIG. 3A to FIG. 3D This was discussed.

[0089] FIG. 3A to FIG. 3D Examples of input data and output data according to some embodiments of the present disclosure are illustrated.

[0090] In some embodiments, LOS / NLOS information may be exchanged between the terminal device 110 and the network device 120. Additionally or alternatively, LOS / NLOS information may also be exchanged between different AI / ML models. Additionally or alternatively, LOS / NLOS information is associated with each data sample collected for the AI / ML model for model training / validation / testing / inference.

[0091] In some embodiments, LOS / NLOS information can be explicitly or implicitly exchanged between the network device and the terminal device, for example, it can be expressed as a hard value (e.g., 0 and 1) or a soft value (e.g., 1, probability or likelihood-scaled value of 0, 0.1, 0.2, ..., 1). LOS / NLOS information can also be expressed as multipath information, such as the number of paths, the number of dominant paths, the ratio of the strongest path to the second strongest path, etc.

[0092] Additionally, the terminal device 110 may provide LOS / NLOS information to the network device 120 via: radio resource control (RRC) signaling, medium access control (MAC) control unit (CE), UCI, CSI report, beam report, etc. In some embodiments, LOS / NLOS information may be provided according to at least one of the following: transmission path, beam, antenna, path, resource set, CSI-reference signal (RS) resource indicator (CRI) transmit-receive point (TRP), cell, bandwidth part (BWP), and spectrum band. In addition, LOS / NLOS information may be provided according to DL reference signal time difference (RSTD), DL reference signal received power (RSRP), and UE Rx-Tx time difference. In addition, LOS / NLOS information may be provided for each AI / ML model, including reporting LOS / NLOS information and each (or a group of) data samples collected for model training / validation / testing / inference. LOS / NLOS information may also be provided by the network device 120 in the same manner.

[0093] Furthermore, the network device 120 may send the LOS / NLOS information to the terminal device 110 via RRC signaling, MAC CE, or DCI. In addition, the LOS / NLOS information may be configured for each AI / ML model, including the configured LOS / NLOS information and each (or a group of) data samples collected for model training / validation / testing / inference. Additionally or alternatively, when the LOS / NLOS condition changes or changes by more than a threshold, the first device (terminal device 110 or network device 120) may send / request resources to send update information.

[0094] like Figure 3A As shown in , the AI / ML model for CSI or BM (which will be discussed below) can also provide LOS / NLOS information as part of the output of the AI / ML model. Figure 3B As shown in , LOS / NLOS information can be used as part of the AI / ML model input for CSI / BM.

[0095] In some other embodiments, such as FIG. 3D to FIG. 3EAs shown in , the location server 130 may provide LOS / NLOS information to the source. The location server 130 may provide LOS / NLOS information to the terminal device 110 or the network device 120. The location server 130 may be provided with LOS / NLOS information from the terminal device 110 or the network device 120. The AI / ML model server may provide LOS / NLOS information to the terminal device 110 or to the network device 120. The AI / ML model server may be provided with LOS / NLOS information from the terminal device 110 or the network device 120. The AI / ML model for positioning may provide LOS / NLOS information to the AI / ML model for CSI / BM. The AI / ML model for CSI / BM may provide LOS / NLOS information to the AI / ML model for positioning. The provided LOS / NLOS information may be based on path, beam, panel, resource, resource set, CRI, TRP, cell, BWP, band, DL RSTD, DL RSRP and UE Rx-Tx time difference, AI / ML model, etc.

[0096] In some further embodiments, the LOS / NLOS indicator may also be another environmental indicator (e.g., indicating a propagation scenario, such as indoor / outdoor, urban / rural, micro / macro, dry air, leaves, fog, rain, cloud, temperature, etc.), a location indicator (cell edge, cell center), a mobility indicator (UE-NW relative moving speed / moving direction), a UE category / type, a BS category / type. This may be further discussed as follows.

[0097] As mentioned above, the first information about the LOS state (or LOS / NLOS information) can be exchanged between any two of the terminal device, the network device, and the location server, and the AI ​​or ML model integrated therein. In this way, the first device and the network device can obtain the LOS / NLOS information for further use (e.g., model selection, model training, model reasoning). In addition, if the channel conditions change, the first device and the network device can update the LOS / NLOS information.

[0098] Figure 3E An example of an AI or ML model according to some embodiments of the present disclosure is illustrated.

[0099] like Figure 3E As shown in , the AI ​​or ML model used in the present disclosure can also be expressed as follows:

[0100]

[0101]

[0102] As described above, in addition to CSI reporting, a first model may be selected for a beam management (BM) process, the first model used for BM may be different from the first model used for CSI reporting, and the so-called "first" model is only used to refer to a model selected by the first device based on the LOS state for different purposes without any limitation. Figure 6C The choice of the first model for the BM process is discussed.

[0103] FIG. 4A to FIG. 4D The signaling processes 400A, 400B, 400C and 400D of the BM according to some embodiments of the present disclosure are illustrated.

[0104] Compared with the CSI report, which may require the selection of the first model on both the terminal device and the network device, the first model may be selected by only one side during the BM process. For example, the first model may be located at the terminal device 110 side or the network device 120 side.

[0105] like Figure 4A As shown in , the first model is selected by the network device 120 for predicting / inferring at least one beam in the beam set for beam prediction. In other words, the first model is located at Figure 4A The network device side in.

[0106] In the signaling process 400A, the terminal device 110 and the network device 120 exchange their capabilities or configurations for selecting an AI or ML model at 401. In some embodiments, the terminal device 110 may exchange capabilities or receive configurations from the network device 120 in the same manner as a CSI report.

[0107] At 403, the terminal device 110 and the network device 120 receive or provide first information about the LOS state. In some embodiments, the terminal device 110 and the network device 120 may receive or provide the first information in the same manner as discussed in the above embodiments. At 405, the network device 120 sends a beam RS set to the first device 110 for measuring beam quality. The beam RS may be sent periodically, semi-persistently, or aperiodically.

[0108] At 407, the terminal device 110 determines a beam set (also referred to as beam set B) for measurement. Beam set B may be determined differently for different LOS states. For a first model classified for a first state range or a LOS communication environment, the terminal device 110 may determine a first beam set for measurement. For a second model classified for a second state range or a NLOS communication environment, the terminal device 110 may determine a second beam set for measurement. The first model and the second model may be selected by a first device as a network device 120 based on the first information in the same manner as in the above-described embodiment. In some embodiments, the network device 120 may select the first model by activating the first model or by deactivating the corresponding other AI / ML models. Additionally or alternatively, the first device may change the model state of the AI / ML model (e.g., from "on" to "off" or from "off" to "on"). Additionally or alternatively, if multiple AI / ML models are configured for LOS / NLOS, additional model index information may be provided. In some embodiments, the second number of beams in the second beam set is more than the first number of beams in the first beam set. In one example, the number of beams measured and reported in a LOS communication environment is relatively small. Conversely, the number of beams measured and reported in a NLOS communication environment is relatively large.

[0109] Additionally or alternatively, a first beam pattern of a first beam set for a first model is different from a second beam pattern of a second beam set for a second model. In the present disclosure, a beam pattern refers to a distribution of beams in a beam grid. For example, the beams in the first beam set are distributed as a plurality of adjacent beams, and wherein the beams in the second beam set are randomly or uniformly distributed in the spatial domain. For clarity of discussion, reference is made to FIG. 5A to FIG. 5B Discuss the differences in beam patterns.

[0110] FIG. 5A to FIG. 5B Example beam patterns 500A and 500B are illustrated according to some embodiments of the present disclosure.

[0111] Beam pattern 500A is an example of a first beam pattern. Figure 5A As shown in , the beams are distributed in a beam grid in a relatively focused manner. Pattern 500B is an example of a second beam pattern. Figure 5B As shown in , the beams are distributed in the beam grid in a random manner or a uniform manner (eg, equidistant beams). Additionally or alternatively, the first set of beams may have a fixed pattern and the second set of beams may have a random pattern.

[0112] Return to see Figure 4AAdditionally or alternatively, the first beam set may be a subset of the third beam set, and the second beam set is different from the fourth beam set. The third beam set and the fourth beam set are beam sets for beam prediction. The third beam set is associated with a first model classified for a LOS communication environment; the fourth beam set is associated with a second model classified for a NLOS communication environment. Furthermore, the first model may predict a preferred beam for the terminal device 110 from the third beam set. The second model may predict a preferred beam for the terminal device 110 from the fourth beam set.

[0113] Additionally or alternatively, a beam in the first beam set is narrower than another beam in the second beam set. In turn, the second beam set may have a relatively wider beam. Additionally or alternatively, the beam selection criteria / requirements associated with the first beam set or the second beam set are different from each other. For example, the RSRP threshold for the beam in the first beam set may be higher than the RSRP threshold for the beam in the second beam set.

[0114] At 409, the terminal device 110 sends a BM report to the network device 120. In some embodiments, if the LOS state is determined to be a LOS communication environment, the terminal device 110 may send a first BM report associated with the first beam set to the network device 120. The terminal device 110 may determine the first BM report by determining the following report information: the first beam ID for the beam in the first beam set and the first RSRP for the beam as determined above, or the second beam ID for the receive beam in the first beam set.

[0115] In some embodiments, if the LOS state is determined to be a NLOS communication environment, the terminal device 110 may send a second BM report associated with the second beam set to the network device 120. The information indicated by the second BM report may be different from the information indicated by the first BM report. For example, the second report information in the second BM report may include at least one of the following: a third beam ID for a beam in the second beam set, a second RSRP and a beam orientation of the beam, a third beam ID and a signal to interference plus noise ratio (SINR) associated with the beam, and a fourth beam ID for a transmit beam in the second beam set.

[0116] Additionally or alternatively, the time domain parameters for the first beam set and the second beam set are also different. For clarity of discussion, reference may be made to FIG. 6A to FIG. 6C The time domain parameters of the first beam set and the second beam set are discussed.

[0117] FIG. 6A to FIG. 6C Illustrated are example historical measurement instances 600A, 600B, and 600C as inputs to a model according to some embodiments of the present disclosure.

[0118] In some embodiments, the input of the second model may include M2 ​​latest beam measurement results, such as Fig. 6A The input of the first model may include M1 beam measurement results, such as Fig. 6A In addition or alternatively, the first interval between adjacent beam measurement results of the first model (eg Figure 6C ) is greater than the second interval between adjacent beam measurements of the second model (as shown in 650 in FIG. Figure 6C 630 in FIG. 1 ).

[0119] Reference Figure 4A , at 409, the first model or the second model selected at the network device 120 may predict or infer a beam from the third beam set based on the BM report (e.g., based on the first BM report), or predict or infer a beam from the fourth beam set (e.g., based on the second BM report). In some embodiments, the first model may output a first number of predictions for a first number of future time instances, and each beam in the first number of predicted beams corresponds to a corresponding future time instance in the first number of future time instances. Furthermore, the second model may output a second number of predictions for a second number of future time instances, and each of the second number of predictions corresponds to a corresponding future time instance in the second number of future time instances. In addition, one of the predictions may be at least one predicted beam. In some embodiments, the first number of predictions is more than the second number of predictions. For example, in a LOS communication environment, there may be F1 number predictions for F1 future time instances, and in a NLOS communication environment, there may be F2 number predictions for F2 future time instances. F1 is more than F2.

[0120] Additionally or alternatively, a third time interval between adjacent time instances in the first number of future time instances is greater than a fourth time interval between adjacent time instances in the second number of future time instances. Additionally or alternatively, a first lifetime of the first number of predictions is greater than a second lifetime of the second number of predictions.

[0121] In some other embodiments, if the AI / ML model is located at the network device 120, the BM report content and format depends on the input type / size. Alternatively, if the AI / ML model is located at the terminal device 110, the BM report content and format depends on the output type / size. Additionally or alternatively, the BM report includes information about whether the beam report is obtained by / with the AI ​​model for LOS or NLOS.

[0122] In some embodiments, data processing for different models may also be different. Data processing includes at least cleaning raw data samples and normalizing / truncating input / output sizes. Data samples (e.g., first and second models) used to train the models are marked with LOS or NLOS information when the data samples are collected. For a first model classified for a LOS communication environment, the first model is trained based on data samples marked with LOS. Additionally or alternatively, the reasoning input for the first device needs to remove data samples marked with "NLOS". For a second model classified for a LOS communication environment, the second model is trained based on data samples marked with NLOS. Additionally or alternatively, the reasoning input for the first device needs to remove data samples marked with "LOS". Additionally or alternatively, for an AI / ML model trained for a specific LOS possibility (value or range), the reasoning input for the AL / ML model only includes data samples within a specific range.

[0123] In some embodiments, some other parameters (eg, quantization type, quantization step size, quantization bits, time-related parameters, computation-related parameters, and loss function or convergence condition) are also different from those discussed with reference to the CSI report.

[0124] Additionally or alternatively, if no LOS / NLOS information is provided, or a special state of LOS / NLOS is provided, a default AI / ML model may be employed. In some embodiments, default values ​​or predefined values ​​of the above parameters are applied. Additionally or alternatively, the default configuration may be a configuration based on non-AI CSI feedback.

[0125] In some embodiments, similarly, an association between at least one of the model selected for the BM process or the parameters associated with the model and the LOS state range is provided. In some embodiments, the provided association is predefined. Additionally or alternatively, the provided association is configured by the first device as the network device 120. Additionally or alternatively, the provided association is received by the terminal device 110 in response to a request for association sent by the terminal device 110 to the network device 120. In some embodiments, the association can be expressed as the following Table 2.

[0126] Table 2

[0127]

[0128]

[0129] There may be more parameters for selecting the first model or the second model, which may be defined in the association. Without any limitation, Table 2 lists only a portion of these parameters. In addition, another model listed in the above table is a general AI or ML model, which may be trained based on data samples labeled with LOS and data samples labeled with NLOS.

[0130] Additionally or alternatively, the first device selects the first model by determining at least a portion of the above parameters based on the first information and the scaling value in the same manner as the CSI report.

[0131] After obtaining the prediction, the network device 120 may update the beam for the terminal device 110. For example, the updating of the applied beam may depend on the network device 120, or may depend on triggering / requesting additional beam searches, etc., and is not restricted in this case.

[0132] In addition to the BM RS transmitted by the network device 120 , the BM RS may also be transmitted by the terminal device 110 .

[0133] like Figure 4B As shown in FIG. 4, at 413, the terminal device 110 may Figure 4A The beam set for beam measurement is determined in the same manner. At 415, the terminal device 110 sends an uplink (UL) BMRS on the beam set determined for beam measurement. At 409, based on the LOS state, the first model or the second model at the network device 120 can predict or infer the beam based on the measured UL BM RS. The network device 120 can then update the beam applied to the terminal device 110, as discussed above.

[0134] In addition to the network device 120 selecting the first model for the BM process and performing inference or prediction, the selection of the first or second model may also be performed at the terminal device 110 .

[0135] FIG. 4C to FIG. 4D Illustrated are signaling processes 400C and 400D of a BM in which the first model is at a terminal device according to some embodiments of the present disclosure.

[0136] In the signaling process 400C, steps 401, 403, 405 and 407 may be the same as discussed above. At 421, based on the LOS state, the first model or the second model at the terminal device 110 may infer or predict the beam to be applied. Then, at 423, the terminal device 110 may send a BM report associated with the predicted beam. In some embodiments, if the first information indicates a LOS communication environment, the terminal device 110 may send a third BM report to the network device 120, the third BM report indicating the third number of beams predicted by the first model in the third beam set and the report information associated with the third number of beams. The reported information may be information similar to the first report information discussed above. Additionally or alternatively, if the first information indicates a NLOS communication environment, the terminal device 110 may send a fourth BM report to the network device 120, the fourth BM report indicating the fourth number of beams predicted by the first model in the fourth beam set and the report information associated with the fourth number of beams. The reported information may be information similar to the second report information discussed above.

[0137] Using the third or fourth BM report, the network device 120 may update the beam applied to the terminal device 110 in the same manner as discussed above.

[0138] In the signaling process 400D, steps 401, 403, 405, 407 and 423 may be the same as discussed above. In this process, similar to process 400B, at 427, the BM RS is sent from the terminal device 110 to the network device 120. Further, at 429, the network device 120 performs measurement on the UL BM RS on the beam set B from the terminal device 110 and sends the measurement result to the terminal device 110. At 431, based on the LOS state information, the terminal device 110 obtains a prediction. Then, using the beam report, the network device 120 updates the beam applied to the terminal device 110 in the same manner as discussed above.

[0139] In this way, the BM prediction accuracy can be enhanced accordingly and the payload of the BM report can be significantly reduced.

[0140] Additionally, as discussed above, the LOS / NLOS indicator may also be another environmental indicator. For example, the impact of the LOS / NLOS indicator may be similar to that of other indicators (environment, mobility, UE category, etc.), and the method in the 0 / 1 case may be effective by replacing the LOS / NLOS indicator with other indicators, such as:

[0141] ●LOS communication environment → UE moves at low speed; NLOS communication environment → UE moves at high speed; LOS possible value / range → moving speed value / range;

[0142] ●LOS communication environment → outdoor / rural environment; NLOS communication environment → indoor / dense urban environment;

[0143] ● LOS communication environment → power saving mode; NLOS communication environment → normal mode; and

[0144] ●LOS communication environment → UE with reduced capabilities; NLOS communication environment → normal UE.

[0145] Additionally, LOS / NLOS information can be applied to improve CSI feedback / BM / positioning / mobility / handover / load balancing / scheduling / resource allocation / energy saving without the need for AI / ML models, for example, determining the CSI / beam report content and format based on LOS / NLOS information.

[0146] In addition, LOS / NLOS information can also be used to select data samples for AI or ML models. For example, LOS / NLOS information can be used to determine the size of data samples collected for model training / verification / testing / inference, for example, the number of data samples (measurements / measurement results / reports / resources / resource sets) required to train the corresponding (multiple) AI / ML models, for example, for model training, if it is LOS, L1 data samples are required, if it is NLOS, L2 data samples are required, L2>L1. Additionally, L2=f(LOS possibility)*L1, f(LOS possibility)=1 / LOS possibility. Additionally, LOS / NLOS information can be used to determine the input / output size / type of data samples collected in the model training / verification / testing / inference stages, respectively. In addition, the method can be applied to AI / ML models for other purposes, such as for positioning, mobility, switching, load balancing, scheduling, resource allocation, and energy saving.

[0147] In this way, data collection (eg, measurement, calculation, reporting) overhead may be reduced.

[0148] Figure 7 A flow chart of a communication method 700 implemented at a first device according to some embodiments of the present disclosure is illustrated. The method 700 may be implemented at a first device according to some embodiments of the present disclosure. Figure 1 For the purpose of discussion, reference will be made to Figure 1 Method 700 is described. It should be understood that method 700 may include additional acts not shown and / or may omit some of the acts shown, and that the scope of the present disclosure is not limited in this regard.

[0149] At 710 , a first device receives first information regarding a line of sight (LOS) status between the first device and a second device.

[0150] At 720 , the first device determines whether the LOS state is within a first state range based on the first information.

[0151] At 730 , in response to determining that the LOS state is within a first state range, the first device selects a first model classified for the first state range.

[0152] In some embodiments, selecting the first model includes selecting the first model from at least two models including the first model and a second model classified for a second state range, the second state range being at least partially different from the first state range.

[0153] In some embodiments, one of the at least two models includes at least one of: a first artificial intelligence (AI) or machine learning (ML) model for channel state information (CSI) compression; a second AI or ML model for beam management (BM).

[0154] In some embodiments, the first model and the second model differ in at least one of the following: the type of AI or ML model; the number of layers of the AI ​​or ML model; the branches of the AI ​​or ML model; the real-valued parameters of the AI ​​or ML model; the complex-valued parameters of the AI ​​or ML model; the number of compressed bits of the AI ​​or ML model; the input data size of the AI ​​or ML model; the output data size of the AI ​​or ML model; the input type of the AI ​​or ML model; or the output type of the AI ​​or ML model.

[0155] In some embodiments, the first state range is specific to a LOS communication environment, and wherein the second state range is specific to a non-LOS (NLOS) communication environment, wherein the at least two models further include a third model classified for a third state range, the third state range being specific to a predefined default communication environment.

[0156] In some embodiments, where a first model and a second model are used for CSI compression, a first number of compressed bits output by the first model is less than a second number of compressed bits output by the second model.

[0157] In some embodiments, the method further comprises sending a CSI report to the second device, the CSI report comprising at least one of the first number of compressed bits or the second number of compressed bits.

[0158] In some embodiments, at least one of the following exists: a first CSI report including a first number of compressed bits also includes a first indication indicating that the first number of compressed bits is determined by a model classified for a LOS communication environment; or a second CSI report including a second number of compressed bits also includes a second indication indicating that the second number of compressed bits is determined by a model classified for a NLOS communication environment.

[0159] In some embodiments, a first size of a first input for a first model is smaller than a second size of a second input for a second model.

[0160] In some embodiments, one of the first size or the second size includes at least one of: the number of subbands; the number of rows or columns of the channel matrix; the bandwidth of the subbands; the number of beams; the number of discrete Fourier transform (DFT) vectors; the number of eigenvalues ​​or eigenvectors; or the rank value of the channel.

[0161] In some embodiments, a first set of historical channel information used as input for the first model spans a longer period of time than a second set of historical channel information used as input for the second model.

[0162] In some embodiments, a first group of historical channel information includes a first number of latest measurement instances, and a second group of historical channel information includes a second number of latest measurement instances, wherein at least one of the following exists: the first number of latest measurement instances is greater than the second number of latest measurement instances; or a first time interval between adjacent measurement instances in the first group is greater than a second time interval between adjacent measurement instances in the second group.

[0163] In some embodiments, the first model is trained based on first data samples labeled with a LOS communication environment, wherein the second model is trained based on second data samples labeled with a NLOS communication environment.

[0164] In some embodiments, at least one of the following exists: a first quantization step size for the first model is greater than a second quantization step size for the second model; or a second quantization bit number for the second model is greater than a first quantization bit number for the first model.

[0165] In some embodiments, at least one of the following exists: a first criterion for a first loss function for a first model is looser than a second criterion for a second loss function for a second model; or a first constraint for the first model is looser than a second constraint for the second model.

[0166] In some embodiments, at least one of: an association between at least one of the model or parameters associated with the model and the LOS state range is predefined; the association is received from the second device; or the association is received in response to a request for association sent by the first device.

[0167] In some embodiments, selecting the first model includes determining at least one of the following based on the first information and the association: a first ID for the first model; a number of compression bits; a compression ratio; an input data size; a pattern of a beam set for beam measurement; a number of predicted beams; or a payload size of a beam report.

[0168] In some embodiments, selecting the first model includes determining at least one of the following based on the first information and the scaling value: the number of compression bits; the compression ratio; the input data size; the pattern of the beam set used for beam measurement; the number of predicted beams; or the payload size of the beam report.

[0169] In some embodiments, the method further includes: determining a first beam set for the first model and a second beam set for the second model when the first model and the second model are used for BM, the first beam set and the second beam set being used for beam measurement.

[0170] In some embodiments, the number of second beams in the second beam set is greater than the number of first beams in the first beam set.

[0171] In some embodiments, the beams in the first beam set are distributed into a plurality of adjacent beams, wherein the beams in the second beam set are randomly or uniformly distributed in the spatial domain.

[0172] In some embodiments, the first beam set is a subset of a third beam set, wherein the second beam set is different from a fourth beam set, the third beam set is one beam set used for beam prediction, and the fourth beam set is another beam set used for beam prediction.

[0173] In some embodiments, a first reference signal received power (RSRP) threshold for a first beam set is greater than a second RSRP for a second beam set.

[0174] In some embodiments, a first model outputs a first number of predictions for a first number of future time instances, and each of the first number of predictions corresponds to a corresponding future time instance in the first number of future time instances, wherein a second model outputs a second number of predictions for a second number of future time instances, and each of the second number of predictions corresponds to a corresponding future time instance in the second number of future time instances, a prediction in the predictions includes at least one predicted beam, and wherein at least one of the following: the first number of predictions is greater than the second number of predictions; the third time interval between adjacent time instances in the first number of future time instances is greater than the fourth time interval between adjacent time instances in the second number of future time instances; or a first lifetime for the first number of predictions is greater than a second lifetime for the second number of predictions.

[0175] In some embodiments, the method further includes determining first reporting information for the first beam set, the first reporting information including the following: a first beam ID and a first RSRP for a beam in the first beam set; or a second beam ID for a receive beam in the first beam set.

[0176] In some embodiments, the method also includes determining second reporting information for a second beam set, the second reporting information including at least one of: a third beam ID for a beam in the second beam set, a second RSRP, and a beam orientation of the beam; a third beam ID and a signal to interference plus noise ratio (SINR) associated with the beam; or a fourth beam ID for a transmit beam in the second beam set.

[0177] In some embodiments, the first device is a terminal device and the second device is a network device, and wherein the method further comprises at least one of: sending a first BM report to the second device, the first BM report comprising first report information; or sending a second BM report to the second device, the second BM report comprising second report information.

[0178] In some embodiments, the first device is a terminal device and the second device is a network device, and wherein the method further includes at least one of the following: sending a third BM report to the second device, the third BM report indicating a third number of beams predicted by the first model in a third beam set and report information associated with the third number of beams, and sending a fourth BM report to the second device, the fourth BM report indicating a fourth number of beams predicted by the second model in a fourth beam set and report information associated with the fourth number of beams, the fourth number being greater than the third number.

[0179] In some embodiments, the first information is received from at least one of: a second device; a location server; or an AI or ML model.

[0180] In some embodiments, the method further includes sending the first information to at least one of: a second device; a location server; or an AI or ML model.

[0181] In some embodiments, sending the first information includes sending the first information according to at least one of a transmission path, a beam, an antenna path, a resource set, a CSI-reference signal (RS) resource indicator (CSI), a transmit-receive point (TRP), a cell, a bandwidth part (BWP), and a spectrum band.

[0182] In some embodiments, the first information further indicates at least one of: a location of the first device; a speed of the first device; or a trajectory of the first device.

[0183] In some embodiments, the first information is used as input data for the first model.

[0184] In some embodiments, the first device includes at least one of a terminal device or a network device.

[0185] Figure 8 A flow chart of a communication method 800 implemented at a second device according to some embodiments of the present disclosure is illustrated. The method 800 may be implemented at Figure 1 For the purpose of discussion, reference will be made to the network device 120 shown in FIG. Figure 1 Method 800 is described. It should be understood that method 800 may include additional acts not shown and / or may omit some of the acts shown, and that the scope of the present disclosure is not limited in this regard.

[0186] At 810, a second device receives a first channel state information (CSI) report including a plurality of compressed bits.

[0187] At 820 , the second device receives first information regarding a line of sight (LOS) status between the first device and the second device.

[0188] At 830 , the second device determines whether the LOS state is within a first state range based on the first information.

[0189] At 840 , in response to determining that the LOS state is within the first state range, the second device selects a first model for decompressing the first CSI report, the first model being classified for the first state range.

[0190] In some embodiments, one of the at least two models includes at least one of the following: an artificial intelligence (AI) or machine learning (ML) model for channel state information (CSI) decompression.

[0191] In some embodiments, the first information is received by receiving a first CSI report.

[0192] In some embodiments, selecting the first model includes selecting the first model from at least two models including the first model and a second model classified for a second state range, the second state range being at least partially different from the first state range.

[0193] In some embodiments, the first state range is specific to a LOS communication environment, and wherein the second state range is specific to a non-LOS (NLOS) communication environment, and wherein the at least two models further include a third model classified for a third state range, the third state range being specific to a predefined default communication environment.

[0194] In some embodiments, a first number of compressed bits output by the first model is less than a second number of compressed bits output by the second model.

[0195] In some embodiments, the first CSI report including the first number of compressed bits further includes a first indication that the first number of compressed bits is determined by a model classified for a LOS communication environment.

[0196] In some embodiments, the second CSI report including the second number of compressed bits further includes a second indication that the second number of compressed bits is determined by a model classified for a NLOS communication environment.

[0197] In some embodiments, the first information is received from at least one of: a first device; a location server; or an AI or ML model.

[0198] In some embodiments, the method further includes sending the first information to at least one of: a first device; a location server; or an AI or ML model.

[0199] In some embodiments, the first information further indicates at least one of: a location of the first device; a speed of the first device; or a trajectory of the first device.

[0200] In some embodiments, the second device includes a network device or a terminal device.

[0201] Fig. 9 is a simplified block diagram of a device 900 suitable for implementing some embodiments of the present disclosure. The device 900 can be viewed as Figure 1 The network devices 130 and 140 shown in FIG. Figure 1 1 is another exemplary embodiment of the terminal devices 110 and 120 shown in . Therefore, the device 900 can be implemented at the above-mentioned network device or terminal device or be implemented as at least a part of the above-mentioned network device or terminal device.

[0202] As shown, the device 900 includes a processor 910, a memory 920 coupled to the processor 910, an appropriate transmitter (TX) and a receiver (RX) 940 coupled to the processor 910, and a communication interface coupled to the TX / RX 940. The memory 920 stores at least a portion of the program 930. The TX / RX 940 is used for bidirectional communication. The TX / RX 940 has at least one antenna to facilitate communication, but in fact the access node mentioned in this application may have several antennas. The communication interface may represent any interface necessary for communication with other network elements, such as an X2 interface for bidirectional communication between gNBs or eNBs, an S1 interface for communication between a mobility management entity (MME) / serving gateway (S-GW) and a gNB or eNB, a Un interface for communication between a gNB or eNB and a relay node (RN), or a Uu interface for communication between a gNB or eNB and a terminal device.

[0203] Assume that the program 930 includes program instructions that, when executed by the associated processor 910, enable the device 900 to operate in accordance with embodiments of the present disclosure, as described herein with reference to Figure 2-Figure 8 The embodiments of the present invention may be implemented by computer software, hardware, or a combination of software and hardware executable by a processor 1910 of the device 900. The processor 910 may be configured to implement various embodiments of the present disclosure. In addition, the combination of the processor 910 and the memory 920 may form a processing component 950 suitable for implementing various embodiments of the present disclosure.

[0204] The memory 920 may be of any type suitable for the local technology network and may be implemented using any suitable data storage technology, such as, as non-limiting examples, non-transitory computer-readable storage media, semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory, and removable memory. Although only one memory 920 is shown in the device 900, there may be several physically different memory modules in the device 900. The processor 1210 may be of any type suitable for the local technology network and, as non-limiting examples, may include one or more of a general-purpose computer, a special-purpose computer, a microprocessor, a digital signal processor (DSP), and a processor based on a multi-core processor architecture. The device 900 may have multiple processors, such as a dedicated integrated circuit chip synchronized with the clock of a synchronous main processor.

[0205] In some embodiments, the terminal device includes circuitry configured to perform method 600 .

[0206] In some embodiments, a network device includes circuitry configured to perform methods 700 and / or 800 .

[0207] The components included in the apparatus and / or equipment of the present disclosure can be implemented in various ways, including software, hardware, firmware, or any combination thereof. In one embodiment, software and / or firmware (e.g., machine executable instructions stored on a storage medium) can be used to implement one or more units. In addition to or in place of machine executable instructions, some or all of the units in the apparatus and / or equipment can be implemented at least in part by one or more hardware logic components. For example, but not limited to, exemplary types of hardware logic components that can be used include field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), and the like.

[0208] Generally, various embodiments of the present disclosure can be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. Some aspects can be implemented in hardware, while other aspects can be implemented in firmware or software, which can be executed by a controller, a microprocessor, or other computing device. Although various aspects of the embodiments of the present disclosure are illustrated and described as block diagrams, flow charts, or using some other graphical representations, it should be understood that, as non-limiting examples, the blocks, devices, systems, technical terminal devices, or methods described herein can be implemented in hardware, software, firmware, dedicated circuits or logic, general hardware or controllers or other computing devices, or some combination thereof.

[0209] The present disclosure also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer executable instructions, such as those included in a program module, which are executed in a device on a target real or virtual processor to perform a process or method as described above with reference to any one of Figures 3 to 11. Typically, a program module includes routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of a program module can be combined or split between program modules as desired. The machine executable instructions of a program module can be executed in a local or distributed device. In a distributed device, a program module can be located in a local and remote storage medium.

[0210] The program code for executing the disclosed method can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device so that when the processor or controller executes the program code, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed completely on the machine, partially on the machine, as an independent software package, partially on the machine and partially on a remote machine, or completely on a remote machine or server.

[0211] The above program code can be embodied in a machine-readable medium, which can be any tangible medium that can contain or store a program for use by or in connection with an instruction execution system, device or equipment. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium can include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0212] In addition, although each operation is depicted in a specific order, this should not be understood as requiring such operations to be performed in the specific order shown or in sequence, or requiring the execution of all illustrated operations to achieve the desired result. In some scenarios, multitasking and parallel processing may be advantageous. Similarly, although several specific embodiment details are included in the above discussion, these should not be interpreted as limitations on the scope of the present disclosure, but should be interpreted as descriptions of features that may be specific to a particular embodiment. Certain features described in the context of separate embodiments may also be implemented in combination in a single embodiment. On the contrary, the various features described in the context of a single embodiment may also be implemented separately in multiple embodiments or in any suitable sub-combination.

[0213] Although the disclosure has been described in language specific to structural features and / or methodological acts, it should be understood that the disclosure defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

[0214] In summary, the embodiments of the present disclosure may provide the following solutions.

[0215] A communication method includes: receiving first information about a line of sight (LOS) state between a first device and a second device; determining whether the LOS state is within a first state range based on the first information; and in response to determining that the LOS state is within the first state range, selecting a first model classified for the first state range.

[0216] In one embodiment, selecting the first model includes selecting the first model from at least two models including the first model and a second model classified for a second state range, the second state range being at least partially different from the first state range.

[0217] In one embodiment, one of the at least two models includes at least one of the following: a first artificial intelligence (AI) or machine learning (ML) model for channel state information (CSI) compression; a second AI or ML model for beam management (BM).

[0218] In one embodiment, the first model differs from the second model in at least one of the following: the type of AI or ML model; the number of layers of the AI ​​or ML model; the branches of the AI ​​or ML model; the real-valued parameters of the AI ​​or ML model; the complex-valued parameters of the AI ​​or ML model; the number of compression bits of the AI ​​or ML model; the input data size of the AI ​​or ML model; the output data size of the AI ​​or ML model; the input type of the AI ​​or ML model; or the output type of the AI ​​or ML model.

[0219] In one embodiment, the first state range is specific to a LOS communication environment, and wherein the second state range is specific to a non-LOS (NLOS) communication environment, and wherein the at least two models further include a third model classified for a third state range, the third state range being specific to a predefined default communication environment.

[0220] In one embodiment, when a first model and a second model are used for CSI compression, a first number of compressed bits output by the first model is less than a second number of compressed bits output by the second model.

[0221] In one embodiment, the method further includes sending a CSI report to the second device, the CSI report including at least one of the first number of compressed bits or the second number of compressed bits.

[0222] In one embodiment, at least one of the following: a first CSI report including a first number of compressed bits also includes a first indication indicating that the first number of compressed bits is determined by a model classified for a LOS communication environment; or a second CSI report including a second number of compressed bits also includes a second indication indicating that the second number of compressed bits is determined by a model classified for a NLOS communication environment.

[0223] In one embodiment, a first size of a first input for a first model is smaller than a second size of a second input for a second model.

[0224] In one embodiment, one of the first size or the second size includes at least one of: the number of subbands; the number of rows or columns of the channel matrix; the bandwidth of the subbands; the number of beams; the number of discrete Fourier transform (DFT) vectors; the number of eigenvalues ​​or eigenvectors; or the rank value of the channel.

[0225] In some embodiments, a first set of historical channel information used as input for the first model spans a longer period of time than a second set of historical channel information used as input for the second model.

[0226] In one embodiment, a first group of historical channel information includes a first number of latest measurement instances, and a second group of historical channel information includes a second number of latest measurement instances, wherein at least one of the following exists: the first number of latest measurement instances is greater than the second number of latest measurement instances; or a first time interval between adjacent measurement instances in the first group is greater than a second time interval between adjacent measurement instances in the second group.

[0227] In one embodiment, the first model is trained based on first data samples labeled with a LOS communication environment, wherein the second model is trained based on second data samples labeled with a NLOS communication environment.

[0228] In one embodiment, at least one of the following exists: a first quantization step size for the first model is greater than a second quantization step size for the second model; or a second quantization bit number for the second model is greater than a first quantization bit number for the first model.

[0229] In one embodiment, at least one of the following exists: a first criterion for a first loss function of a first model is looser than a second criterion for a second loss function of a second model; or a first constraint for the first model is looser than a second constraint for the second model.

[0230] In one embodiment, at least one of the following exists: an association between at least one of the model or parameters associated with the model and the LOS state range is predefined; the association is received from the second device; or the association is received in response to a request for association sent by the first device.

[0231] In one embodiment, selecting the first model includes: determining at least one of the following based on the first information and the association: a first ID for the first model; a number of compression bits; a compression ratio; an input data size; a pattern of a beam set for beam measurement; a number of predicted beams; or a payload size of a beam report.

[0232] In one embodiment, selecting the first model includes: determining at least one of the following based on the first information and the scaling value: the number of compression bits; the compression ratio; the input data size; the pattern of the beam set used for beam measurement; the number of predicted beams; or the payload size of the beam report.

[0233] In one embodiment, the method further comprises: determining a first beam set for the first model and a second beam set for the second model when the first model and the second model are used for BM, the first beam set and the second beam set being used for beam measurement.

[0234] In one embodiment, the number of the second beams in the second beam set is greater than the number of the first beams in the first beam set.

[0235] In one embodiment, the beams in the first beam set are distributed into a plurality of adjacent beams, wherein the beams in the second beam set are randomly or uniformly distributed in the spatial domain.

[0236] In one embodiment, the first beam set is a subset of a third beam set, and wherein the second beam set is different from a fourth beam set, the third beam set being one beam set used for beam prediction, and the fourth beam set being another beam set used for beam prediction.

[0237] In one embodiment, a first reference signal received power (RSRP) threshold for a first beam set is greater than a second RSRP for a second beam set.

[0238] In one embodiment, a first model outputs a first number of predictions for a first number of future time instances, and each of the first number of predictions corresponds to a corresponding future time instance in the first number of future time instances, wherein a second model outputs a second number of predictions for a second number of future time instances, and each of the second number of predictions corresponds to a corresponding future time instance in the second number of future time instances, a prediction in the predictions includes at least one prediction beam, and wherein at least one of the following: the first number of predictions is greater than the second number of predictions; the third time interval between adjacent time instances in the first number of future time instances is greater than the fourth time interval between adjacent time instances in the second number of future time instances; or a first lifetime for the first number of predictions is greater than a second lifetime for the second number of predictions.

[0239] In one embodiment, the method further includes determining first reporting information for the first beam set, the first reporting information comprising at least one of: a first beam ID and a first RSRP for a beam in the first beam set; or a second beam ID for a receive beam in the first beam set.

[0240] In one embodiment, the method also includes determining second reporting information for a second beam set, the second reporting information including at least one of: a third beam ID for a beam in the second beam set, a second RSRP, and a beam orientation of the beam; a third beam ID and a signal to interference plus noise ratio (SINR) associated with the beam; or a fourth beam ID for a transmit beam in the second beam set.

[0241] In one embodiment, the first device is a terminal device and the second device is a network device, wherein the method further includes at least one of the following: sending a first BM report to the second device, the first BM report including first report information; or sending a second BM report to the second device, the second BM report including second report information.

[0242] In one embodiment, the first device is a terminal device and the second device is a network device, wherein the method further includes at least one of the following: sending a third BM report to the second device, the third BM report indicating a third number of beams predicted by the first model in a third beam set and report information associated with the third number of beams, and sending a fourth BM report to the second device, the fourth BM report indicating a fourth number of beams predicted by the second model in a fourth beam set and report information associated with the fourth number of beams, the fourth number being greater than the third number.

[0243] In one embodiment, the first information is received from at least one of: a second device; a location server; or an AI or ML model.

[0244] In one embodiment, the method further includes sending the first report to at least one of: a second device; a location server; or an AI or ML model.

[0245] In one embodiment, sending the first information includes sending the first information according to at least one of a transmission path, a beam, an antenna path, a resource set, a CSI-reference signal (RS) resource indicator (CSI), a transmit-receive point (TRP), a cell, a bandwidth part (BWP), and a spectrum band.

[0246] In one embodiment, the first information further indicates at least one of: a location of the first device; a speed of the first device; or a trajectory of the first device.

[0247] In one embodiment, the first information is used as input data for a first model.

[0248] In one embodiment, the first device includes at least one of a terminal device or a network device.

[0249] A communication method includes: receiving, at a second device, a first channel state information (CSI) report including a plurality of compressed bits from a first device; receiving first information about a line of sight (LOS) state between the first device and the second device; determining, based on the first information, whether the LOS state is within a first state range; and in response to determining that the LOS state is within the first state range, selecting a first model for decompressing the first CSI report, the first model being classified for use in the first state range.

[0250] In one embodiment, one of the at least two models includes at least one of the following: an artificial intelligence (AI) or machine learning (ML) model for channel state information (CSI) decompression.

[0251] In one embodiment, the first information is received by receiving a first CSI report.

[0252] In one embodiment, selecting the first model includes selecting the first model from at least two models including the first model and a second model classified for a second state range, the second state range being at least partially different from the first state range.

[0253] In one embodiment, the first state range is specific to a LOS communication environment, and wherein the second state range is specific to a non-LOS (NLOS) communication environment, and wherein the at least two models further include a third model classified for a third state range, the third state range being specific to a predefined default communication environment.

[0254] In one embodiment, a first number of compression bits for a first model is less than a second number of compression bits for a second model.

[0255] In one embodiment, the first CSI report including the first number of compression bits further includes a first indication that the first number of compression bits is determined by a model classified for a LOS communication environment.

[0256] In one embodiment, the second CSI report including the second number of compressed bits further includes a second indication that the second number of compressed bits is determined by a model classified for a NLOS communication environment.

[0257] In one embodiment, the first information is received from at least one of: a first device; a location server; or an AI or ML model.

[0258] In one embodiment, the method further includes sending at least one of the following to the first information: a first device; a location server; or an AI or ML model.

[0259] In one embodiment, the first information further indicates at least one of: a location of the first device; a speed of the first device; or a trajectory of the first device.

[0260] In one embodiment, the second device includes a network device or a terminal device.

[0261] A terminal device comprises: a processor; and a memory, the memory being coupled to the processor and storing instructions, the instructions causing the terminal device to execute a method according to the above communication method when executed by the processor.

[0262] A network device comprises: a processor; and a memory, the memory being coupled to the processor and storing instructions, the instructions causing the network device to perform a method according to the above communication method when executed by the processor.

[0263] A computer readable medium stores instructions thereon, which when executed on at least one processor cause the at least one processor to perform a method according to the above communication method.

[0264] Without any limitation, the terms and corresponding descriptions related to the present disclosure are shown in Table 3 below.

[0265] Table 3

[0266]

[0267]

Claims

1. A method of communication, comprising: receiving, at a first device, first information regarding a line of sight (LOS) status between the first device and a second device; determining whether the LOS state is within a first state range based on the first information; as well as In response to determining that the LOS state is within the first state range, a first model classified for the first state range is selected.

2. The method of claim 1 , wherein selecting the first model comprises: The first model is selected from at least two models, the at least two models including the first model and a second model classified for a second state range, the second state range being at least partially different from the first state range.

3. The method according to claim 2, wherein one of the at least two models comprises at least one of the following: A first artificial intelligence (AI) or machine learning (ML) model for channel state information (CSI) compression; A second AI or ML model for beam management (BM).

4. The method of claim 2, wherein the first model and the second model are different in at least one of the following: The type of AI or ML model; The number of layers in the AI ​​or ML model; Branching of AI or ML models; Real-valued parameters of AI or ML models; Complex-valued parameters of AI or ML models; The number of bits of compression for the AI ​​or ML model; The size of the input data to the AI ​​or ML model; The output data size of the AI ​​or ML model; The type of input to the AI ​​or ML model; or The output type of the AI ​​or ML model.

5. The method of claim 2, wherein the first state range is specific to a LOS communication environment, and wherein the second state range is specific to a non-LOS (NLOS) communication environment, and The at least two models further include a third model classified for a third state range, the third state range being specific to a predefined default communication environment.

6. The method of claim 5, wherein when the first model and the second model are used for CSI compression, a first number of compressed bits output by the first model is less than a second number of compressed bits output by the second model.

7. The method according to claim 6, further comprising: A CSI report is sent to the second device, the CSI report comprising at least one of the first number of compressed bits or the second number of compressed bits.

8. The method of claim 6, wherein at least one of the following is present: The first CSI report including the first number of compressed bits also includes a first indication indicating that the first number of compressed bits is determined by a model classified for the LOS communication environment; or The second CSI report including the second number of compressed bits also includes a second indication that the second number of compressed bits is determined by a model classified for the NLOS communication environment.

9. The method of claim 5, wherein a first size of a first input for the first model is smaller than a second size of a second input for the second model.

10. The method of claim 5, wherein a first set of historical channel information as input for the first model spans a longer period of time than a second set of historical channel information as input for the second model.

11. The method of claim 10, wherein the first set of historical channel information includes a first number of latest measurement instances, the second set of historical channel information includes a second number of latest measurement instances, and at least one of the following exists: The first number of latest measurement instances is greater than the second number of latest measurement instances; or A first time interval between adjacent measurement instances in the first group is greater than a second time interval between adjacent measurement instances in the second group.

12. The method of claim 5, wherein at least one of the following is present: an association between at least one of a model or a parameter associated with the model and the LOS state range is predefined; The association is received from the second device; or The association is received in response to a request for the association sent by the first device.

13. The method according to claim 5, further comprising: In a case where the first model and the second model are used for BM, a first beam set for the first model and a second beam set for the second model are determined, and the first beam set and the second beam set are used for beam measurement. The method according to claim 13 , wherein the number of second beams in the second beam set is greater than the number of first beams in the first beam set.

15. The method of claim 13, wherein the beams in the first beam set are distributed into a plurality of adjacent beams, and wherein the beams in the second beam set are randomly or uniformly distributed in a spatial domain.

16. The method of claim 13, wherein the first beam set is a subset of a third beam set, and wherein the second beam set is different from a fourth beam set, the third beam set being one beam set used for beam prediction, and the fourth beam set being another beam set used for beam prediction.

17. The method of claim 13, wherein the first model outputs a first number of predictions for a first number of future time instances, and each of the first number of predictions corresponds to a respective one of the first number of future time instances, wherein the second model outputs a second number of predictions for a second number of future time instances, and each of the second number of predictions corresponds to a respective one of the second number of future time instances, one of the predictions comprising at least one prediction beam, and wherein at least one of the following is present: the first number of predictions is greater than the second number of predictions; a third time interval between adjacent time instances in the first number of future time instances is greater than a fourth time interval between adjacent time instances in the second number of future time instances; or A predicted first lifetime for the first quantity is greater than a predicted second lifetime for the second quantity.

18. The method according to claim 17, further comprising determining first reporting information for the first beam set, the first reporting information comprising at least one of the following: a first beam ID and a first RSRP for a beam in the first beam set; or A second beam ID for a receive beam in the first beam set.

19. The method according to claim 18, further comprising determining second reporting information for the second beam set, the second reporting information comprising at least one of the following: a third beam ID for a beam in the second beam set, a second RSRP, and a beam orientation of the beam; the third beam ID and a signal to interference plus noise ratio (SINR) associated with the beam; or A fourth beam ID for the transmit beam in the second beam set.

20. The method according to claim 19, wherein the first device is a terminal device, and the second device is a network device, and wherein the method further comprises at least one of the following: Sending a first BM report to the second device, where the first BM report includes the first report information; or A second BM report is sent to the second device, where the second BM report includes the second report information.

21. The method according to claim 17, wherein the first device is a terminal device, and the second device is a network device, and wherein the method further comprises at least one of the following: sending a third BM report to the second device, the third BM report indicating a third number of beams predicted by the first model in the third beam set and reporting information associated with the third number of beams, and A fourth BM report is sent to the second device, the fourth BM report indicating a fourth number of beams predicted by the second model in the fourth beam set and reporting information associated with the fourth number of beams, the fourth number being greater than the third number.

Images