WO2025035256A1 - 无线通信方法及装置、终端设备、网络设备 - Google Patents

无线通信方法及装置、终端设备、网络设备 Download PDF

Info

Publication number
WO2025035256A1
WO2025035256A1 PCT/CN2023/112522 CN2023112522W WO2025035256A1 WO 2025035256 A1 WO2025035256 A1 WO 2025035256A1 CN 2023112522 W CN2023112522 W CN 2023112522W WO 2025035256 A1 WO2025035256 A1 WO 2025035256A1
Authority
WO
WIPO (PCT)
Prior art keywords
model
information
terminal device
time
network device
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/CN2023/112522
Other languages
English (en)
French (fr)
Inventor
陈文洪
田文强
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangdong Oppo Mobile Telecommunications Corp Ltd
Original Assignee
Guangdong Oppo Mobile Telecommunications Corp Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guangdong Oppo Mobile Telecommunications Corp Ltd filed Critical Guangdong Oppo Mobile Telecommunications Corp Ltd
Priority to CN202380101132.0A priority Critical patent/CN121713607A/zh
Priority to PCT/CN2023/112522 priority patent/WO2025035256A1/zh
Publication of WO2025035256A1 publication Critical patent/WO2025035256A1/zh
Priority to US19/419,652 priority patent/US20260106806A1/en
Priority to MX2025015518A priority patent/MX2025015518A/es
Anticipated expiration legal-status Critical
Pending legal-status Critical Current

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/16Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L1/00Arrangements for detecting or preventing errors in the information received
    • H04L1/12Arrangements for detecting or preventing errors in the information received by using return channel
    • H04L1/16Arrangements for detecting or preventing errors in the information received by using return channel in which the return channel carries supervisory signals, e.g. repetition request signals
    • H04L1/18Automatic repetition systems, e.g. Van Duuren systems
    • H04L1/1812Hybrid protocols; Hybrid automatic repeat request [HARQ]
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W72/00Local resource management
    • H04W72/04Wireless resource allocation
    • H04W72/044Wireless resource allocation based on the type of the allocated resource
    • H04W72/0446Resources in time domain, e.g. slots or frames

Definitions

  • the embodiments of the present application relate to the field of mobile communication technology, and specifically to a wireless communication method and apparatus, terminal equipment, and network equipment.
  • AI artificial intelligence
  • AI models When AI models are used for wireless communications, they can be divided into single-end models and dual-end models.
  • the dual-end model needs to be deployed in pairs on the terminal device and network device side.
  • the AI model used for channel state information (CSI) feedback is a typical dual-end model, in which the terminal device can use the obtained channel information (such as feature vectors, beam information, delay information, etc.) as the input of the AI model, and the AI model can infer the corresponding CSI quantization bits.
  • the corresponding AI model on the network device side can use the CSI quantization bits as input and reversely infer the corresponding channel information.
  • the network device can instruct the terminal device to update the model through signaling (for example, replace an AI model or adjust model parameters).
  • the network device also needs to perform corresponding model updates so that the AI models on both sides still match each other. If the updates of the AI models on both sides are not completely synchronized, the output of the receiving AI model will be incorrect, thereby affecting the information transmission between the terminal device and the network device. How to ensure that the model updates between the network device and the terminal device are synchronized is a problem that needs to be solved.
  • the embodiments of the present application provide an information transmission method and apparatus, a terminal device, and a network device.
  • the wireless communication method provided by the embodiment of the present application includes:
  • the terminal device determines the time when the first AI model starts to be applied according to the model application time reported to the network device, or according to the model application time configured by the network device;
  • the terminal device uses the first AI model to communicate with the network device.
  • the wireless communication method provided by the embodiment of the present application includes:
  • the network device determines the time when the first AI model starts to be applied according to the model application time reported by the terminal device, or according to the model application time configured for the terminal device;
  • the network device uses the second AI model corresponding to the first AI model to communicate with the terminal device.
  • the wireless communication device provided in the embodiment of the present application is applied to a terminal device, and the device includes:
  • a first determining unit is configured to determine a time when the first AI model starts to be applied according to the model application time reported to the network device, or according to the model application time configured by the network device;
  • the first communication unit is configured to communicate with the network device using the first AI model after the moment.
  • a wireless communication device provided in an embodiment of the present application is applied to a network device, and the device includes:
  • a second determining unit is configured to determine a time when the first AI model starts to be applied according to the model application time reported by the terminal device, or according to the model application time configured for the terminal device;
  • the second communication unit is configured to communicate with the terminal device using a second AI model corresponding to the first AI model after the moment.
  • an embodiment of the present application provides a terminal device, the terminal device includes a processor and a memory.
  • the memory is used to store a computer program
  • the processor is used to call and run the computer program stored in the memory to execute the above-mentioned wireless communication method.
  • an embodiment of the present application provides a network device, the network device comprising a processor and a memory.
  • the memory is used to store a computer program
  • the processor is used to call and run the computer program stored in the memory to execute the above-mentioned wireless communication method.
  • the chip provided in the embodiment of the present application is used to implement the above-mentioned wireless communication method.
  • the chip includes: a processor, which is used to call and run a computer program from a memory so that the chip installed The device executes the above wireless communication method.
  • the computer-readable storage medium provided in the embodiment of the present application is used to store a computer program, which enables a computer to execute the above-mentioned wireless communication method.
  • the computer program product provided in the embodiment of the present application includes computer program instructions, which enable a computer to execute the above-mentioned wireless communication method.
  • the computer program provided in the embodiment of the present application when executed on a computer, enables the computer to execute the above-mentioned wireless communication method.
  • the terminal device when the corresponding dual-end AI model is used on the terminal device side and the network device side, if the AI model on the terminal device side is updated or reconfigured, the terminal device can determine the time when the updated or reconfigured AI model starts to be applied based on the model application time reported to the network device, or based on the model application time configured by the network device. In this way, it can be ensured that the time when the models on both sides start to be applied is the same, thereby ensuring that the output of the AI model is available, and improving the performance of the communication system.
  • FIG1 is a schematic diagram of a communication architecture provided in an embodiment of the present application.
  • FIG2 is a schematic diagram of a neuron structure provided in an embodiment of the present application.
  • FIG3 is a schematic diagram of a neural network structure provided in an embodiment of the present application.
  • FIG4 is a schematic diagram of a neural network for CSI feedback provided in an embodiment of the present application.
  • FIG5 is a schematic diagram of a wireless communication method provided in an embodiment of the present application.
  • FIG6A is a schematic diagram of a time slot structure provided in an embodiment of the present application.
  • FIG6B is a second schematic diagram of a time slot structure provided in an embodiment of the present application.
  • FIG7A is a third schematic diagram of a time slot structure provided in an embodiment of the present application.
  • FIG7B is a schematic diagram of a time slot structure provided in an embodiment of the present application.
  • FIG8 is a schematic diagram of a time slot structure provided in an embodiment of the present application.
  • FIG9A is a schematic diagram of a time slot structure provided in an embodiment of the present application.
  • FIG9B is a schematic diagram of a time slot structure provided in an embodiment of the present application.
  • FIG9C is a schematic diagram of a time slot structure provided in an embodiment of the present application.
  • FIG10 is a schematic structural diagram of a wireless communication device 1000 provided in an embodiment of the present application.
  • FIG11 is a schematic structural diagram of a wireless communication device 1100 provided in an embodiment of the present application.
  • FIG12 is a schematic structural diagram of a communication device provided in an embodiment of the present application.
  • FIG13 is a schematic structural diagram of a chip according to an embodiment of the present application.
  • FIG. 14 is a schematic block diagram of a communication system provided in an embodiment of the present application.
  • FIG. 1 is a schematic diagram of an application scenario of an embodiment of the present application.
  • the communication system 100 may include a terminal device 110 and a network device 120.
  • the network device 120 may communicate with the terminal device 110 via an air interface.
  • the terminal device 110 and the network device 120 support multi-service transmission.
  • LTE Long Term Evolution
  • TDD LTE Time Division Duplex
  • UMTS Universal Mobile Telecommunication System
  • IoT Internet of Things
  • NB-IoT Narrow Band Internet of Things
  • eMTC enhanced Machine-Type Communications
  • 5G communication system also known as New Radio (NR) communication system
  • NR New Radio
  • the network device 120 may be an access network device that communicates with the terminal device 110.
  • the network device may provide communication coverage for a specific geographical area, and may communicate with a terminal device 110 (eg, UE) located in the coverage area.
  • a terminal device 110 eg, UE
  • the network device 120 can be an evolved base station (Evolutional Node B, eNB or eNodeB) in a Long Term Evolution (LTE) system, or a Next Generation Radio Access Network (NG RAN) device, or a base station (gNB) in an NR system, or a wireless controller in a Cloud Radio Access Network (CRAN), or the network device 120 can be a relay station, an access point, an in-vehicle device, a wearable device, a hub, a switch, a bridge, a router, or a network device in a future evolved Public Land Mobile Network (PLMN), etc.
  • Evolutional Node B, eNB or eNodeB in a Long Term Evolution (LTE) system
  • NG RAN Next Generation Radio Access Network
  • gNB base station
  • CRAN Cloud Radio Access Network
  • PLMN Public Land Mobile Network
  • the terminal device 110 may be any terminal device, including but not limited to a terminal device connected to the network device 120 or other terminal devices by wire or wireless connection.
  • the terminal device 110 may refer to an access terminal, a user equipment (UE), a user unit, a user station, a mobile station, a mobile station, a remote station, a remote terminal, a mobile device, a user terminal, a terminal, a wireless communication device, a user agent, or a user device.
  • UE user equipment
  • the access terminal may be a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, an IoT device, a satellite handheld terminal, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA), a handheld device with wireless communication function, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device, a wearable device, a terminal device in a 5G network, or a terminal device in a future evolution network, etc.
  • SIP Session Initiation Protocol
  • IoT IoT device
  • satellite handheld terminal a Wireless Local Loop (WLL) station
  • PDA Personal Digital Assistant
  • PDA Personal Digital Assistant
  • the terminal device 110 can be used for device to device (Device to Device, D2D) communication.
  • D2D Device to Device
  • the wireless communication system 100 may further include a core network device 130 that communicates with the network device 120.
  • the core network device 130 may be a 5G core network (5G Core, 5GC) device, such as an access and mobility management function (Access and Mobility Management Function, AMF), and another example, an authentication server function (Authentication Server Function, AUSF), and another example, a user plane function (User Plane Function, UPF), and another example, a session management function (Session Management Function, SMF).
  • 5G Core, 5GC 5G Core, 5GC
  • AMF Access and Mobility Management Function
  • AUSF Authentication Server Function
  • UPF User Plane Function
  • SMF Session Management Function
  • the core network device 130 may also be an evolved packet core (Evolved Packet Core, EPC) device of the LTE network, such as a session management function + core network data gateway (Session Management Function+Core Packet Gateway, SMF+PGW-C) device.
  • EPC evolved Packet Core
  • SMF+PGW-C Session Management Function+Core Packet Gateway
  • SMF+PGW-C Session Management Function+Core Packet Gateway
  • SMF+PGW-C Session Management Function+Core Packet Gateway
  • SMF+PGW-C Session Management Function+Core Packet Gateway
  • SMF+PGW-C Session Management Function+Core Packet Gateway
  • the various functional units in the communication system 100 can also establish connections and achieve communication through the next generation network (NG) interface.
  • NG next generation network
  • the terminal device establishes an air interface connection with the access network device through the NR interface for transmitting user plane data and control plane signaling; the terminal device can establish a control plane signaling connection with the AMF through the NG interface 1 (N1 for short); the access network device, such as the next-generation wireless access base station (gNB), can establish a user plane data connection with the UPF through the NG interface 3 (N3 for short); the access network device can establish a control plane signaling connection with the AMF through the NG interface 2 (N2 for short); the UPF can establish a control plane signaling connection with the SMF through the NG interface 4 (N4 for short); the UPF can exchange user plane data with the data network through the NG interface 6 (N6 for short); the AMF can establish a control plane signaling connection with the SMF through the NG interface 11 (N11 for short); the SMF can establish a control plane signaling connection with the PCF through the NG interface 7 (N7 for short).
  • the access network device such as the next-generation wireless access
  • FIG1 exemplarily shows a network device, a core network device and two terminal devices.
  • the wireless communication system 100 may include multiple network devices and each network device may include other number of terminal devices within its coverage area, which is not limited in the embodiments of the present application.
  • FIG. 1 is only an example of the system to which the present application is applicable.
  • the method shown in the embodiment of the present application can also be applied to other systems.
  • system and “network” are often used interchangeably in this article.
  • the term “and/or” in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships.
  • a and/or B can represent: A exists alone, A and B exist at the same time, and B exists alone.
  • the character "/" in this article generally indicates that the associated objects before and after are in an "or” relationship.
  • the "indication" mentioned in the embodiment of the present application can be a direct indication, an indirect indication, or an indication of an association relationship.
  • a indicates B which can mean that A directly indicates B, for example, B can be obtained through A; it can also mean that A indirectly indicates B, for example, A indicates C, B can be obtained through C; it can also mean that A and B have an association relationship.
  • the "correspondence” mentioned in the embodiment of the present application can mean that there is a direct or indirect correspondence relationship between the two, or it can mean that there is an association relationship between the two, or it can mean that there is an indication and being indicated, configuration and being configured, etc.
  • predefined can refer to the definition in the protocol.
  • protocol can refer to a standard protocol in the field of communications, for example, it can include LTE protocol, The NR protocol and related protocols used in future communication systems are not limited in this application.
  • AI models are models that can handle a variety of tasks. They have the ability to self-learn and self-adapt, and can dynamically adjust and make decisions based on changes in the environment. AI models can also be called machine learning (ML) models, and the two are equivalent or interchangeable.
  • ML machine learning
  • AI models can be composed of neural networks.
  • a neural network is a computing model composed of multiple interconnected neuron nodes, where the connection between nodes represents the weighted value from the input signal to the output signal, called the weight; each node performs a weighted summation on different input signals and outputs them through a specific activation function.
  • a1, a2, ..., an and 1 are the inputs of the neuron
  • w1, w2, ..., wn and b represent weights
  • Sum represents the summation function
  • f represents the activation function
  • t is the output result.
  • a simple neural network is shown in Figure 3, which includes an input layer, a hidden layer, and an output layer. Through different connections, weights, and activation functions of multiple neurons, different outputs can be generated, thereby fitting the mapping relationship from input to output.
  • Each upper-level node is connected to all of its lower-level nodes.
  • This fully connected model can also be called DNN, or deep neural network.
  • An AI model can be trained and obtained through the process of data set construction, training, verification and testing. Training can be divided into offline training and online training. A static training result can be obtained by offline training of the data set, which can be called offline training here.
  • the network equipment can continue to collect more data and perform real-time online training to optimize the parameters of the AI model to achieve better inference and prediction results. After obtaining the AI model, by inputting the current information into the AI model, the corresponding model output can be inferred.
  • AI When AI is used in wireless communications, it can be divided into single-end models and dual-end models.
  • the single-end model can be used by deploying it only on one side of the terminal device or network device, and the training of the AI model can also be performed on only one side;
  • the dual-end model needs to be deployed in pairs on the terminal device and network device side, and the models on both sides need to be trained together, that is, the models deployed on both sides are corresponding and cannot be used or updated separately.
  • an AI model for channel state information (CSI) feedback is a typical dual-end model.
  • FIG4 for a schematic diagram of the dual-end AI model structure for CSI feedback.
  • an AI model for encoding (encoder) is deployed on the terminal side
  • a corresponding AI model for decoding (decoder) is deployed on the network side.
  • the terminal device outputs CSI quantization bits based on the encoder model, typically PMI bits, and then feeds back to the network device through the uplink channel (PUSCH/PUCCH).
  • the network device uses the CSI quantization bits (PMI) fed back by the terminal as the input of the decoder model, thereby outputting channel information corresponding to the input on the terminal side, such as the feature vectors of each subband, etc., for downlink precoding.
  • the terminal device also needs to monitor the model performance and report to the network device when the model performance is not good, so that the network device can update the model (for example, update the model structure or update the model parameters). Since the AI models on both sides are matched, if the network device updates the model, it needs to notify the terminal device through signaling so that the terminal device also updates to the corresponding AI model. Otherwise, the output result of the network device will be unusable.
  • the above-mentioned AI model based on dual-end deployment can not only be used for CSI feedback, but also for other processing methods with corresponding operations/structures on the terminal device and network device sides, such as channel coding-channel decoding, modulation-demodulation, pilot generation-channel estimation, transceiver RF signal processing, etc. It is only necessary to train a set of AI models on the corresponding terminal device side and network device side.
  • the network device can instruct the terminal device to update the model through signaling (for example, replace an AI model or adjust model parameters).
  • the network device also needs to perform corresponding model updates so that the AI models on both sides still match each other. If the updates of the AI models on both sides are not completely synchronized, the output of the receiving AI model will be wrong, thereby affecting the information transmission between the terminal device and the network device.
  • an embodiment of the present application provides a wireless communication method, wherein when the corresponding dual-end AI model is used on the terminal device side and the network device side, if the AI model on the terminal device side is updated or reconfigured, the terminal device can determine the time when the AI model configured by the network device starts to be applied based on the model application time reported to the network device, or based on the model application time configured by the network device. In this way, it can be ensured that the model application time on both sides is the same, thereby ensuring that the output of the AI model is available, and improving the performance of the communication system.
  • FIG5 shows a wireless communication method provided in an embodiment of the present application, which may include step S110 and step S120.
  • the terminal device determines the time when the first AI model starts to be applied according to the model application time reported to the network device, or according to the model application time configured by the network device;
  • the terminal device uses the first AI model to communicate with the network device.
  • the network device can determine the time when the first AI model starts to be applied according to the model application time reported by the terminal device, or according to the model application time configured for the terminal device; and after the time, the network device uses the second AI model corresponding to the first AI model to communicate with the terminal device.
  • the terminal device and the network device use corresponding dual-end AI models for communication, wherein the first AI model is a neural network model deployed on the terminal device side, and the second AI model is a neural network model deployed on the network device side.
  • the first AI model can be used in any of the following processes:
  • the first AI model is used for downlink CSI feedback.
  • the terminal device may use the obtained channel information (such as eigenvectors, beam information, delay information, etc.) as the input of the first AI model, and infer the corresponding CSI quantization bits through the first AI model.
  • channel information such as eigenvectors, beam information, delay information, etc.
  • the first AI model is used for channel decoding of downlink data, and the terminal device may use the received encoded data as the input of the first AI model and output the decoded information bits.
  • the first AI model is used for demodulating the downlink signal.
  • the terminal device may use the received downlink modulated signal as the input of the first AI model and output the demodulated information bits.
  • the first AI model is used for downlink channel estimation.
  • the terminal device may use the received downlink channel and pilot signal as input of the first AI model and output a channel estimation result.
  • the first AI model is used for downlink RF signal processing.
  • the terminal device may use the received downlink RF signal as the input of the first AI model and output the downlink RF signal processing result.
  • the first AI model is used for channel coding of uplink data, and the terminal device may use the information bits to be sent as the input of the first AI model and output the encoded data.
  • the first AI model is used for modulation of uplink data
  • the terminal device may use the information bits of the uplink data to be sent as the input of the first AI model and output an uplink modulation signal.
  • the first AI model is used for uplink pilot generation, and the terminal device may use the signal to be sent as the input of the first AI model and output a pilot signal.
  • the first AI model is used for uplink RF signal processing, and the terminal device may use the signal to be sent as the input of the first AI model and output an uplink RF signal.
  • the first AI model can be trained through the processes of data set construction, training, verification and testing.
  • the first AI model can be trained in advance through offline training and/or online training.
  • the second AI model can be used in any of the following processes:
  • the correspondence between the first AI model and the second AI model may mean that the operations implemented by the first AI model and the second AI model are one-to-one corresponding.
  • the second AI model is used for decoding downlink CSI feedback information; or, if the first AI model is used for generating downlink beam information, the second AI model can be used for decoding downlink beam feedback information; if the first AI model is used for channel coding of uplink data, the second AI model can be used for channel decoding of uplink data, etc.
  • the second AI model can be trained through the processes of data set construction, training, verification and testing.
  • the second AI model can be trained in advance through offline training and/or online training.
  • the first AI model deployed on the terminal device side may be configured by the network device for the terminal device. It should be noted that the network device may configure the first AI model for the terminal device when the terminal device initially accesses; or, when the network device detects that the original AI model has poor performance, the network device may configure the first AI model for the terminal device to instruct the terminal device to update the model.
  • the network device may send first information to the terminal device, and correspondingly, the terminal device receives the first information sent by the network device, where the first information is used to configure the first AI model.
  • the first information is downlink information used to configure the first AI model.
  • the first information may be sent to the terminal device via broadcast signaling.
  • the first information may be carried via a physical broadcast channel (PBCH), system information (SIB), group common downlink control information (Group Common DCI), etc., and the embodiments of the present application do not limit this.
  • PBCH physical broadcast channel
  • SIB system information
  • Group Common DCI group common downlink control information
  • Group Common DCI is DCI sent to a group of terminal devices via a common search space (CSS).
  • SCS common search space
  • Group Common DCI is scrambled using a public radio network temporary identifier (RNTI).
  • the first information may be high-level information, which is sent via high-level signaling.
  • the first information may be sent via radio resource control (RRC) signaling, or via media access control (MAC) signaling, such as a MAC control element (MAC CE).
  • RRC radio resource control
  • MAC media access control
  • the first information may also be physical layer information, which is sent via physical layer signaling.
  • the first information may be carried via DCI.
  • the first information may be model recovery information sent by the network device, and the model recovery information includes configuration information of the first AI model.
  • the network device may send corresponding model recovery information for updating the terminal side model.
  • the first information may include one or more of the following:
  • the first information may carry identification information of the first AI model.
  • the terminal device may preconfigure or predefine one or more AI models.
  • the terminal device may determine the first AI model to be applied/updated from one or more AI models based on the identification information of the first AI model carried in the first information.
  • Each AI model may have an identification information corresponding to one of them, and the corresponding relationship may be pre-agreed by the network device and the terminal device, or configured by the network device to the terminal device.
  • the first information may indicate a model function of the first AI model.
  • the model function is the function implemented by the AI/ML model, and different models have different model functions, such as: PMI reporting, RI/PMI/CQI reporting, beam information reporting, RSRP reporting, CSI compression, channel coding and decoding, modulation and demodulation, CSI prediction, CSI prediction and compression, etc.
  • the terminal device can determine the first AI model to be used according to the model function to be implemented by the AI model.
  • model structure of the first AI model carried in the first information may be at least part of the model structure in the first AI model, or in other words, the first information may carry all or part of the model structure of the first AI model.
  • model parameters of the first AI model carried in the first information may be at least part of the model parameters in the first AI model, or in other words, the first information may carry all or part of the model parameters of the first AI model.
  • the network device does not have to configure the entire AI model, but may only configure or update part of the structure or part of the model parameters in the first AI model.
  • the first AI model deployed on the terminal device side may also be a model selected and used by the terminal device from a plurality of preconfigured or predefined AI models. It should be noted that the terminal device may report the first AI model it uses to the network device upon initial access; in addition, the terminal device may also update the model (such as through online training) when it detects that the performance of the original AI model is poor, use the first AI model for communication, and indicate the first AI model to the network device at the same time, so that the network device updates the corresponding model currently deployed on the network device side.
  • the terminal device may send the second information to the network device, and correspondingly, the network device receives the second information sent by the terminal device, and the second information is used to indicate to the network device the first AI model used by the terminal device.
  • the second information is sent to the network device through a channel, typically reported to the network device through a PUSCH or a PUCCH. If the second information is physical layer information, it can be reported to the network device through a PUSCH or a PUCCH, for example, as part of a CSI report; if the second information is high-layer information, it can be reported to the network device through a PUSCH, for example, MAC signaling.
  • the network device may send response information corresponding to the second information, so that the terminal device knows that the network device has correctly received the second information.
  • the response information of the second information may be the corresponding RAR (random access response); if the second information is carried by PUSCH, the second information may be the HARQ-ACK information of the PUSCH.
  • the second information may include one or more of the following:
  • the second information may carry identification information of the first AI model.
  • the network device can determine the first AI model that the terminal device is about to apply/update from one or more AI models based on the identification information of the first AI model carried in the second information, thereby determining the second AI model corresponding to the first AI model.
  • Each AI model may have an identification information corresponding to one of them, and the corresponding relationship may be pre-agreed by the network device and the terminal device, or configured by the network device to the terminal device.
  • the second information may indicate the model function of the first AI model.
  • the function of the model is the function implemented by the AI/ML model, and different models have different model functions, such as: PMI reporting, RI/PMI/CQI reporting, beam information reporting, RSRP reporting, CSI compression, channel coding and decoding, modulation and demodulation, CSI prediction, CSI prediction and compression, etc.
  • the network device can determine the first AI model to be used/updated by the terminal device according to the model function to be implemented by the AI model, thereby determining the second model corresponding to the first AI model.
  • model structure of the first AI model carried in the second information may be at least part of the model structure in the first AI model, or in other words, the second information may carry all or part of the model structure of the first AI model.
  • model parameters of the first AI model carried in the second information may be at least part of the model parameters in the first AI model, or in other words, the second information may carry all or part of the model parameters of the first AI model.
  • the terminal device does not have to indicate the entire first AI model, and may only configure or update part of the structure or part of the model parameters in the first AI model.
  • both the terminal device and the network device need to determine the time when the first AI model starts to be applied, so as to enable the first AI model to communicate with the network device at this time, thereby ensuring that the application time of the corresponding models on the terminal device side and the network device side are synchronized, and avoiding the problem of model mismatch on both sides.
  • the terminal device may determine the moment when the first AI model configured by the network device/indicated by the terminal device starts to be applied based on the model application time reported by the terminal device; accordingly, the network device may determine the moment when the first AI model configured by the network device for the terminal device/indicated by the terminal device starts to be applied based on the model application time reported by the terminal device. In another possible implementation, the terminal device may determine the moment when the first AI model configured by the network device/indicated by the terminal device starts to be applied based on the model application time configured by the network device; accordingly, the network device may determine the moment when the first AI model configured by the network device/indicated by the terminal device starts to be applied based on the model application time configured by the network device.
  • the terminal device determines the time when the first AI model configured by the network device/indicated by the terminal device starts to be applied based on the model application time reported to the network device; correspondingly, the network device can determine the time when the first AI model configured for the terminal device/indicated by the terminal device starts to be applied based on the model application time reported by the terminal device.
  • the terminal device can report the application time of the model it supports to the network device.
  • the terminal device reports the model application time to the network device through UE capability information (eg, UECapabilityInformation).
  • UE capability information eg, UECapabilityInformation
  • the terminal device may report the model application time to the network device before receiving the first information sent by the network device for configuring the first AI model.
  • the network device may receive the model application time reported by the terminal device before sending the first information.
  • the terminal device may report the model application time to the network device before sending the second information indicating the first AI model; accordingly, the network device may receive the model application time reported by the terminal device before receiving the second information.
  • the terminal device may carry the model application time in the second information sent to indicate the first AI model, and the first AI model used may be The AI model and the model application time are reported to the network device together. In this way, after receiving the second information, the network device can also determine the model application time at the same time, and can apply the model faster.
  • the model application time may be any one of the following:
  • the time interval between when the terminal device receives the first information and when the first AI model is applied is applied
  • the output information of the model does not need to be reported to the first AI model of the network device, such as the first AI model used for channel decoding of downlink data, demodulation of downlink channels, downlink channel estimation, downlink RF signal processing, channel coding of uplink data, modulation of uplink data, uplink pilot generation, and uplink RF signal processing.
  • the model application time may be the time interval between the time when the terminal device receives the first information and the time when the first AI model is applied, or the time interval between the time when the terminal device sends the HARQ-ACK information of the first information and the time when the first AI model is applied.
  • the model application time may be the time interval between the time when the terminal device receives the first information and the time when the terminal device sends feedback information obtained based on the first AI model, or the time interval between the time when the terminal device sends the HARQ-ACK information of the first information and the time when the terminal device sends the feedback information obtained based on the first AI model.
  • the model application time may be any one of the following:
  • the model application time can be the time interval between the time when the terminal device sends the second information and the time when the first AI model is applied, or the time interval between the time when the terminal device receives the response information of the second information and the time when the first AI model is applied.
  • the model application time may be the time interval between the time when the terminal device sends the second information and the time when the terminal device sends feedback information obtained based on the first AI model, or the time interval between the time when the terminal device receives the response information of the second information and the time when the terminal device sends feedback information obtained based on the first AI model.
  • the moment when the terminal device sends feedback information obtained based on the first AI model involved in the model application time in the above two scenarios may refer to the moment when the terminal device sends feedback information obtained based on the first AI model for the first time.
  • the model application time may be: the time interval between the time when the terminal device receives the first information and the time when the terminal device sends the CSI feedback information obtained based on the first AI model (for the first time), or the time interval between the time when the terminal device sends the HARQ-ACK information of the first information and the time when the terminal device sends the CSI feedback information obtained based on the first AI model (for the first time).
  • the model application time may also be: the time interval between the time when the terminal device sends the second information and the time when the terminal device sends the CSI feedback information obtained based on the first AI model (for the first time), or the time interval between the time when the terminal device receives the response information of the second information and the time when the terminal device sends the CSI feedback information obtained based on the first AI model (for the first time).
  • the model application time may be: the time interval between the time the terminal device receives the first information and the time the terminal device sends the downlink beam information obtained based on the first AI model (for the first time), or the time interval between the time the terminal device sends the HARQ-ACK information of the first information and the time the terminal device sends the downlink beam information obtained based on the first AI model (for the first time).
  • model application time may also be the time interval between the time the terminal device sends the second information and the time the terminal device sends the downlink beam information obtained based on the first AI model (for the first time), or The time interval between the time when the terminal device receives the response information of the second information and the time when the terminal device sends the downlink beam information obtained based on the first AI model (for the first time).
  • time interval can be in units of time slots or OFDM symbols.
  • time interval can also be in units of micro-time slots, OFDM symbol sets, or absolute time (e.g., N milliseconds, N seconds, N>0), which is not limited in the embodiments of the present application.
  • the model application time reported by the terminal device to the network device may include one or more of the following:
  • model application time corresponding to one or more AI models supported by the terminal device; wherein the one or more AI models supported by the terminal device include the first AI model;
  • the model application time corresponding to one or more model update methods supported by the terminal device; wherein the one or more model update methods supported by the terminal device include an update method of the first AI model;
  • the model application time corresponding to one or more AI model functions supported by the terminal device; wherein the one or more model functions supported by the terminal device include the function of the first AI model;
  • the model application time corresponding to one or more model types supported by the terminal device; wherein the one or more model types supported by the terminal device include the type of the first AI model.
  • the terminal device may support one or more AI models, which may be preconfigured or predefined. It is understandable that different AI models have different model structures and model parameters, so the model application time of different AI models is also different.
  • the terminal device may report the model application time of each of the one or more AI models it supports. It is understandable that the terminal device may report different model application times for different models.
  • the terminal device may support one or more model update modes.
  • the model update mode includes: updating model parameters and/or updating model structure.
  • the terminal device can report the corresponding model application time for one or more AI model update methods it supports. It is understandable that the terminal device can report different model application times for different model update methods. Exemplarily, the application time required to update only the model parameters is t1, and the application time required to update the model structure and model parameters is t2, and t1 ⁇ t2.
  • updating the model structure and model parameters at the same time can be considered as updating the entire AI model.
  • the terminal device may support one or more model types.
  • the model types supported by the terminal device may include high-precision models and low-precision models; or known models and unknown models.
  • the complexity of the high-precision model is higher than that of the low-precision model, so the model application time required is also longer.
  • the high-precision model can be a model with more output bits, or a model with more neural network layers, or a model with a more complex neural network structure.
  • the known model may be a model that has been deployed on the terminal device
  • the unknown model may be a model that has not been deployed on the terminal device.
  • the application time required for the known model is shorter than that required for the unknown model.
  • model application time reported by the terminal device can be determined based on the processing power or hardware configuration of the terminal device, as well as the time required for reasoning of different AI models/different AI model types.
  • the model application time of the AI model using the CPU of the terminal device can be longer than the model application time of the AI model using the GPU.
  • the terminal device and the network device determine the time when the first AI model (the first AI model may be configured by the network device through the first information, or indicated by the terminal device through the second information) starts to be applied according to the reported model application time, which can be achieved by one or more of the following:
  • the terminal device may report that the model application time corresponding to AI model 1 is T1, and the model application time corresponding to AI model 2 is T2. If the network device configures model 1 for the terminal device through the first information, the terminal device and the network device may determine the time when the AI model 1 configured by the first information starts to be applied according to the model application time T1 corresponding to AI model 1. Alternatively, if the terminal device indicates model 2 through the second information, the terminal device and the network device may determine the time when the AI model 2 indicated by the second information starts to be applied according to the model application time T2 corresponding to AI model 2.
  • the terminal device may report that the model application time corresponding to model function 1 (such as CSI compression function) is T3, and the model The model application time corresponding to model function 2 (such as channel coding and decoding function) is T4.
  • the network device configures model function 1 for the terminal device through the first information
  • the terminal device and the network device can determine the time when the AI model configured by the first information starts to be applied according to the model application time T3 corresponding to model function 1.
  • the terminal device indicates model function 2 through the second information
  • the terminal device and the network device can determine the time when the AI model indicated by the second information starts to be applied according to the model application time T4 corresponding to model function 2.
  • the terminal device may report the model application time T5 corresponding to the high-precision model and the model application time T6 corresponding to the low-precision model. If the AI model configured for the terminal device by the network device through the first information is a low-precision model, the terminal device and the network device may determine the time when the AI model configured by the first information starts to be applied based on the model application time T6 of the low-precision model. If the terminal device indicates that the type of AI model is a high-precision model through the second information, the terminal device and the network device may determine the time when the AI model indicated by the second information starts to be applied based on the application time T5 of the high-precision model.
  • the terminal device reports the model application time T7 corresponding to the known model and the model application time T8 corresponding to the unknown model.
  • the AI model 1 configured for the terminal device by the network device through the first information is an unknown model (for example, AI model 1 is a model other than the preconfigured or predefined candidate model of the terminal device)
  • the terminal device and the network device can determine the time when the AI model 1 configured by the first information starts to be applied based on the model application time T8 of the unknown model.
  • the terminal device can determine the start time of application of the first AI model based on the longest (or shortest) model application time.
  • the terminal device determines the moment when the first AI model configured by the network device/indicated by the terminal device starts to be applied based on the model application time configured by the network device; correspondingly, the network device can determine the moment when the first AI model configured by the network device/indicated by the terminal device starts to be applied based on the model application time configured for the terminal device.
  • the network device can directly configure the model application time for the terminal device.
  • the network device can send the third information to the terminal device, and the terminal device can receive the third information sent by the network device, wherein the third information is used to configure the model application time.
  • the network device configures the model application time for the terminal device through the third information.
  • the third information may be sent to the terminal device via broadcast signaling.
  • the third information may be carried via PBCH, SIB, Group Common DCI, etc., which is not limited in the embodiments of the present application.
  • the third information may be high-level information, which is sent via high-level signaling.
  • the third information may be sent via RRC signaling, or MAC CE signaling.
  • the third information may also be physical layer information, which is sent via physical layer signaling.
  • the third information may be carried via DCI signaling.
  • the third information may include the model application time of the first AI model configured in the first information.
  • the third information can be indicated to the terminal device together with the first information, so that the terminal device can also determine the application time of the model at the same time after receiving the configuration information of the model, and can apply the model more quickly.
  • the network device may configure the model application time of the first AI model for the terminal device according to a preset rule or a preset mapping relationship.
  • the network device may configure the application time of the first AI model for the terminal device based on factors such as the type of the first AI model configured for the terminal device and the update method of the first AI model.
  • the network device may pre-store the model application time corresponding to each AI model (the mapping relationship between the AI model and the model application time).
  • the model application time corresponding to the first AI model may be configured for the terminal device based on the mapping relationship.
  • the network device may receive the model application time reported by the terminal device, and configure the actual model application time for the terminal device according to the model application time reported by the terminal device.
  • the network device can determine the model application time configured for the terminal device based on the model application time reported by multiple terminal devices within the coverage area.
  • the network device may use the maximum value of the model application time reported by multiple terminal devices as the model application time actually used by each terminal device within the coverage area, and configure the model application time to these terminal devices.
  • the duration of the model application time configured by the network device for the terminal device is greater than or equal to the duration of the model application time reported by the terminal device.
  • the network device can configure the model application time for multiple different terminal devices within its coverage at the same time, so that Network devices and different terminal devices can update models synchronously, thereby ensuring that the model output at the receiving end is available and improving communication performance.
  • the network device may determine the model application time of the first AI model indicated by the second information after receiving the second information, and indicate the model application time to the terminal device through the third information. That is, after receiving the second information, the network device may send the third information to the terminal device to configure the model application time of the first AI model indicated in the second information for the terminal device.
  • the network device can configure the model application time of the first AI model for the terminal device according to a preset rule or a preset mapping relationship.
  • the network device can configure the actual model application time for the terminal device according to the model application time reported by the terminal device.
  • the network device can use the maximum value of the model application time reported by multiple terminal devices as the model application time actually used by each terminal device within the coverage area, and configure the model application time to these terminal devices.
  • the network device may also send third information to the terminal device before the second information.
  • the third information may include one or more of the following:
  • model application time corresponding to one or more AI models respectively; wherein the one or more AI models include the first AI model;
  • model application time corresponding to one or more model functions respectively; wherein the one or more model functions include the function of the first AI model;
  • Model application time corresponding to one or more model update methods; wherein the one or more model update methods include an update method of the first AI model;
  • Model application time corresponding to one or more model types respectively; wherein the one or more model types include the type of the first AI model.
  • the terminal device can determine the actual model application time of the first AI model indicated by the second information in combination with the first AI model indicated by the second information, and the model application time corresponding to each AI model, each model function, each model update method, or each model type configured by the third information.
  • model application time configured in the third information can be determined by the network device according to the model application time reported by the terminal device. It is understandable that the network device can determine the model application time configured for the terminal device according to the model application time reported by multiple terminal devices within the coverage area. In some embodiments, the network device can use the maximum value of the model application time reported by multiple terminal devices as the model application time actually used by each terminal device within the coverage area, and configure the model application time to these terminal devices.
  • the third information may configure the model application time corresponding to AI model 1 (or model function 1/model update mode 1/model type 1) and AI model 2 (or model function 2/model update mode 2/model type 2), respectively.
  • the model application time of AI model 1 (or model function 1/model update mode 1/model type 1) configured in the third information may be the maximum value of the model application time corresponding to AI model 1 (or model function 1/model update mode 1/model type 1) reported by multiple terminal devices
  • the model application time of AI model 2 (or model function 2/model update mode 2/model type 2) configured in the third information may be the maximum value of the model application time corresponding to AI model 2 (or model function 2/model update mode 2/model type 2) reported by multiple terminal devices.
  • the network device can configure the model application time for multiple different terminal devices within its coverage at the same time.
  • the network device and different terminal devices can update the model synchronously, thereby ensuring that the model output at the receiving end is available and improving communication performance.
  • the terminal device after the terminal device determines the time when the first AI model starts to be applied based on the above method, the terminal device can further use the first AI model to transmit information with the network device after the time.
  • the network device can use the second model corresponding to the first AI model to transmit information with the terminal device after the time.
  • the terminal device may use the third AI model to transmit information with the network device, or the terminal device may not transmit information with the network device based on the AI model. Accordingly, before the terminal device applies the first AI model, the network device may use the fourth AI model corresponding to the third AI model to transmit information with the terminal device, or the network device may not transmit information with the terminal device based on the AI model, for example, the network device receives information transmitted by the terminal device based on a traditional non-AI method.
  • the third AI model may be a predefined AI model or a historical AI model.
  • the predefined AI model is an AI model agreed upon in advance by the terminal device and the network device.
  • the historical AI model may be an AI model configured by the network device for the terminal device before the first AI model starts to be applied, or an AI model indicated by the terminal device.
  • the historical AI model here may be an AI model with a model structure and model parameters completely different from those of the first AI model, or may be an AI model with the same model structure and model parameters as the first AI model.
  • the AI models with different model parameters may also be AI models with different model functions, which is not limited in the embodiments of the present application.
  • the fourth AI model may be an AI model deployed on the network device side corresponding to the third AI model.
  • the terminal device can use the previously used third AI model to transmit information with the network device.
  • the terminal device can use the third model corresponding to the previous functionality to transmit information with the network device.
  • the function can also be a function ID.
  • the terminal device can use the previously adopted model structure and model parameters to transmit information with the network device.
  • the terminal device can use the previously adopted model parameters to transmit information with the network device.
  • the terminal device does not transmit information with the network device based on the AI model, which means that the terminal device uses traditional non-AI methods to transmit information with the network device.
  • the terminal device when the corresponding dual-end AI model is used on the terminal device side and the network device side, if the AI model on the terminal device side is updated or reconfigured, the terminal device can determine the time when the AI model configured by the network device starts to be applied based on the model application time reported to the network device, or based on the model application time configured by the network device. In this way, it can be ensured that the model application time on both sides is the same, thereby ensuring that the output of the AI model is available, and improving the performance of the communication system.
  • the moment when the first AI model determined by the terminal device and the network device starts to be applied is related to the first information.
  • the following is an explanation of the moment when the first AI model starts to be applied in the scenario where the network device configures the first AI model for the terminal device through the first information through methods #1 to #3.
  • Method #1 The time when the first AI model starts to be applied can be:
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the uplink channel carrying the HARQ-ACK information of the first information starts k OFDM symbols;
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the downlink channel carrying the first information starts k OFDM symbols; wherein k is the model application time.
  • the time interval represented by the model application time in mode #1 may be in units of OFDM symbols.
  • the model application time may be k OFDM symbols, or in other words, the time interval represented by the model application time is k OFDM symbols, where k is an integer greater than or equal to 1.
  • k can be reported by the terminal device to the network device, or can be configured by the network device.
  • the terminal device may use the first time slot or the first OFDM symbol after k OFDM symbols from the last OFDM symbol of the downlink channel carrying the first information as the time when the first AI model starts to be applied.
  • the terminal device may use the first time slot or the first OFDM symbol after k OFDM symbols from the last OFDM symbol of the downlink channel carrying the first information as the time when the first AI model starts to be applied.
  • the downlink channel here may be PBCH, PDSCH, PDCCH, etc.
  • the downlink channel when the first information is broadcast information, the downlink channel may be PBCH; when the first information is high-layer information, the downlink channel may be PDSCH; when the first information is physical layer information, the downlink channel may be PDCCH.
  • k OFDM symbols in the embodiment of the present application may also be k-1 or k+1 OFDM symbols, and the embodiment of the present application does not impose any restrictions on this.
  • the terminal device can determine that the time when the first AI model starts to be applied is the first time slot after the symbol where symbol m+k is located.
  • the terminal device can determine that the moment when the first AI model starts to apply is the first OFDM symbol after the symbol m+k, that is, the symbol m+k+1 shown in Figure 6B.
  • the terminal device may use the first time slot or the first OFDM symbol after the last OFDM symbol of the uplink channel carrying the HARQ-ACK information of the first information k OFDM symbols as the time when the first AI model starts to be applied.
  • the terminal device may use the first time slot or the first OFDM symbol after the last OFDM symbol of the uplink channel carrying the HARQ-ACK information of the first information k OFDM symbols as the time when the first AI model starts to be applied. The moment to start applying.
  • the terminal device can send HARQ-ACK information of the first information to the network device.
  • the terminal device can feed back HARQ-ACK information of the first information to the network device.
  • the uplink channel carrying the HARQ-ACK information of the first information may be a PUCCH.
  • k OFDM symbols in the embodiment of the present application may also be k-1 or k+1 OFDM symbols, and the embodiment of the present application does not impose any restrictions on this.
  • the terminal device can determine that the time when the first AI model starts to apply is the first time slot after the symbol where symbol m+k is located.
  • the terminal device can determine that the moment when the first AI model starts to apply is the first OFDM symbol after the symbol m+k, that is, the symbol m+k+1 shown in FIG7B .
  • using the first OFDM symbol after k OFDM symbols as the moment to start applying the first AI model can shorten the time interval for applying the first AI model, and at the same time improve the accuracy of the model application time, compared with using the first time slot after k OFDM symbols as the moment to start applying the first AI model.
  • Method #2 The time when the first AI model starts to be applied can be:
  • the first time slot after k time slots starts from the time slot in which the terminal device receives the first information, where k is the model application time.
  • the time interval represented by the model application time in mode #2 may be in time slots.
  • the model application time may be k time slots, or in other words, the time interval represented by the model application time is k time slots, where k is an integer greater than or equal to 1.
  • k can be reported by the terminal device to the network device, or can be configured by the network device.
  • the terminal device can use the first time slot after k time slots from the time slot of receiving the first information as the time when the first AI model starts to be applied.
  • the terminal device can use the first time slot after k time slots from the time slot of receiving the first information as the time when the first AI model starts to be applied.
  • k time slots in the embodiment of the present application may also be k-1 or k+1 time slots, and the embodiment of the present application does not impose any limitation on this.
  • the terminal device may determine that the moment when the first AI model starts to be applied is the first time slot after time slot n+k, that is, time slot n+k+1.
  • Method #3 The time when the first AI model starts to be applied can be:
  • the first time slot after k time slots, (k+3N) time slots, or max(k,3N) time slots from the time slot where the uplink channel of the HARQ-ACK information carrying the first information is located, where k is the application time of the model and N is the number of time slots contained in 1ms.
  • the time interval represented by the model application time in mode #3 may be in time slots.
  • the model application time may be k time slots, or in other words, the time interval represented by the model application time is k time slots, where k is an integer greater than or equal to 1.
  • k can be reported by the terminal device to the network device, or can be configured by the network device.
  • the terminal device may use the first time slot after k time slots from the time slot where the uplink channel of the HARQ-ACK information carrying the first information is located as the time when the first AI model starts to be applied. In other words, the terminal device may use the first time slot after k time slots from the time slot where the uplink channel of the HARQ-ACK information carrying the first information is located as the time when the first AI model starts to be applied.
  • the first information is high-level information.
  • the first information is sent via RRC signaling.
  • k time slots in the embodiment of the present application may also be k-1 or k+1 time slots, and the embodiment of the present application does not impose any restrictions on this.
  • the time when the first AI model begins to be applied is the first time slot after time slot n+k, that is, time slot n+k+1.
  • the terminal device may use the first time slot after (k+3N) time slots from the time slot where the uplink channel of the HARQ-ACK information carrying the first information is located as the time when the first AI model starts to be applied. In other words, the terminal device may use the first time slot after (k+3N) time slots from the time slot where the uplink channel of the HARQ-ACK information carrying the first information is located as the time when the first AI model starts to be applied.
  • the first information is high-layer information.
  • the first information is sent via MAC layer signaling, such as the signaling defined in 3GPP TS38.321.
  • 3N time slots (ie, 3 ms) is the minimum time required for the MAC layer signaling to take effect, and the MAC layer signaling is ensured to have sufficient time to take effect by spacing k+3N time slots.
  • k time slots in the embodiment of the present application may also be k-1 or k+1 time slots, and the embodiment of the present application does not impose any restrictions on this.
  • the time when the first AI model begins to apply is the first time slot after time slot n+k+3N, that is, time slot n+k+3N+1.
  • the terminal device may use the first time slot after the time slot of the uplink channel carrying the HARQ-ACK information of the first information that is located max(k,3N) time slots as the time when the first AI model starts to be applied. In other words, the terminal device may use the first time slot after the time slot of the uplink channel carrying the HARQ-ACK information of the first information that is located max(k,3N) time slots as the time when the first AI model starts to be applied.
  • the first information is high-level information.
  • the first information is transmitted through MAC layer signaling, such as the signaling defined in 3GPP TS38.321.
  • 3N time slots (ie, 3 ms) is the minimum time required for the MAC layer signaling to take effect. By taking the larger value of k and 3N, it is ensured that the MAC layer signaling has sufficient time to take effect.
  • k time slots in the embodiment of the present application may also be k-1 or k+1 time slots, and the embodiment of the present application does not impose any restrictions on this.
  • the time when the first AI model begins to apply is the first time slot after time slot n+3N, that is, time slot n+3N+1.
  • the terminal device and the network device can agree on one of the methods #1 to #3, or the network device can specify one of the methods #1 to #3 for the terminal device to determine the time when the first AI model starts to be applied, thereby ensuring that the time when the corresponding AI models on both sides start to be applied is the same.
  • the moment when the first AI model starts to be applied determined by the terminal device and the network device may be related to the second information.
  • the following is an explanation of the moment when the first AI model starts to be applied in the scenario where the terminal device indicates the first AI model to the network device through the second information through methods #1' to #3'.
  • Method #1 the time when the first AI model starts to be applied is:
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the uplink channel carrying the second information starts k OFDM symbols;
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the response information of the uplink channel carrying the second information starts k OFDM symbols; wherein k is the model application time.
  • the time interval represented by the model application time in mode #1' can be in units of OFDM symbols.
  • the model application time can be k OFDM symbols, or in other words, the time interval represented by the model application time is k OFDM symbols. Wherein k is an integer greater than or equal to 1.
  • the terminal device may use the first time slot or the first OFDM symbol after k OFDM symbols from the last OFDM symbol of the uplink channel carrying the second information as the time when the first AI model starts to be applied.
  • the terminal device may use the first time slot or the first OFDM symbol after k OFDM symbols from the last OFDM symbol of the uplink channel carrying the second information as the time when the first AI model starts to be applied.
  • the uplink channel here may be PUSCH or PUCCH.
  • the uplink channel may be PUSCH; if the second information is physical layer information, the uplink channel may be PUCCH.
  • k OFDM symbols in the embodiment of the present application may also be k-1 or k+1 OFDM symbols, and the embodiment of the present application does not impose any restrictions on this.
  • the terminal device may use the first time slot or the first OFDM symbol after k OFDM symbols from the last OFDM symbol of the downlink channel of the response information carrying the second information as the time when the first AI model starts to be applied.
  • the terminal device may use the first time slot or the first OFDM symbol after k OFDM symbols from the last OFDM symbol of the downlink channel of the response information carrying the second information as the time when the first AI model starts to be applied.
  • the network device can send response information corresponding to the second information, so that the terminal device knows that the network device has correctly received the second information.
  • the response information of the second information can be the corresponding RAR; if the second information is carried by PUSCH, the second information can be the HARQ-ACK information of the PUSCH.
  • k OFDM symbols may also be k-1 or k+1 OFDM symbols, and the embodiment of the present application does not impose any limitation on this.
  • Method #2' the time when the first AI model starts to be applied is:
  • the first time slot after k time slots start from the time slot in which the terminal device sends the second information; the second information is uplink information used to indicate the first AI model, and k is the model application time.
  • the time interval represented by the model application time in mode #2' can be in time slots.
  • the model application time can be k time slots, or in other words, the time interval represented by the model application time is k time slots. Where k is an integer greater than or equal to 1.
  • k can be reported by the terminal device to the network device, or can be configured by the network device.
  • the terminal device can use the first time slot after k time slots from the time slot for sending the second information as the time when the first AI model starts to be applied.
  • the terminal device can use the first time slot after k time slots from the time slot for sending the second information as the time when the first AI model starts to be applied.
  • k time slots in the embodiment of the present application may also be k-1 or k+1 time slots, and the embodiment of the present application does not impose any limitation on this.
  • Method #3 the first AI model starts to be applied at:
  • the downlink channel of the response information carrying the second information starts from k time slots, or (k+3N) time slots, or the first time slot after max(k,3N) time slots, wherein the second information is the uplink information used to indicate the first AI model, k is the model application time, and N is the number of time slots contained in 1ms.
  • the time interval represented by the model application time in mode #3' can be in time slots.
  • the model application time can be k time slots, or in other words, the time interval represented by the model application time is k time slots. Where k is an integer greater than or equal to 1.
  • k can be reported by the terminal device to the network device, or can be configured by the network device.
  • the terminal device may use the first time slot after k time slots from the time slot where the downlink channel carrying the response information of the second information is located as the time when the first AI model starts to be applied. In other words, the terminal device may use the first time slot after k time slots from the time slot where the downlink channel carrying the response information of the second information is located as the time when the first AI model starts to be applied.
  • the second information is high-layer information.
  • the second information is sent via RRC signaling.
  • k time slots in the embodiment of the present application may also be k-1 or k+1 time slots, and the embodiment of the present application does not impose any restrictions on this.
  • the terminal device may use the first time slot after (k+3N) time slots from the time slot where the downlink channel carrying the response information of the second information is located as the time when the first AI model starts to be applied. In other words, the terminal device may use the first time slot after (k+3N) time slots from the time slot where the downlink channel carrying the response information of the second information is located as the time when the first AI model starts to be applied.
  • the second information is high-level information.
  • the second information is sent via MAC layer signaling, such as the signaling defined in 3GPP TS38.321.
  • 3N time slots (ie, 3 ms) is the minimum time required for the MAC layer signaling to take effect, and the MAC layer signaling is ensured to have sufficient time to take effect by spacing k+3N time slots.
  • k time slots in the embodiment of the present application may also be k-1 or k+1 time slots, and the embodiment of the present application does not impose any restrictions on this.
  • the terminal device may use the first time slot after the time slot of the downlink channel carrying the response information of the second information that is located max(k,3N) time slots as the time point when the first AI model starts to be applied. In other words, the terminal device may use the first time slot after the time slot of the downlink channel carrying the response information of the second information that is located that is located max(k,3N) time slots as the time point when the first AI model starts to be applied.
  • the second information is high-level information.
  • the second information is sent via MAC layer signaling, such as the signaling defined in 3GPP TS38.321.
  • 3N time slots (ie, 3 ms) is the minimum time required for the MAC layer signaling to take effect. By taking the larger value of k and 3N, it is ensured that the MAC layer signaling has sufficient time to take effect.
  • k time slots in the embodiment of the present application may also be k-1 or k+1 time slots, and the embodiment of the present application does not impose any restrictions on this.
  • the terminal device and the network device can agree on one of the methods #1’ to #3’, or the network device can specify one of the methods #1’ to #3’ for the terminal device to determine the time when the first AI model starts to be applied, thereby ensuring that the time when the corresponding AI models on both sides start to be applied is the same.
  • the moment when the first AI model starts to be applied and/or the model application time can be determined according to any one of the following:
  • the subcarrier spacing of the downlink channel carrying the response information of the second information is the subcarrier spacing of the downlink channel carrying the response information of the second information
  • the subcarrier spacing of the BWP activated on the carrier to which the first AI model is applied is applied.
  • time when the first AI model starts to be applied and the model application time are parameters in units of time slots or OFDM symbols.
  • the actual corresponding time length/time (or absolute time length/time slot) needs to be determined based on the subcarrier spacing. Different subcarrier spacings correspond to different time lengths.
  • Table 1 for the relationship between the subcarrier spacing, the number of time slots, and the time slot length.
  • the subcarrier spacing is 30kHz, and the length of each time slot is 0.5ms. If the model application time is k time slots, the length of k time slots is 0.5k ms. Similarly, the subcarrier spacing is 60kHz, and the length of each time slot is 0.25ms. If the model application time is k time slots, the length of k time slots is 0.25k ms.
  • the moment when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing related to the first AI model.
  • the moment when the first AI model starts to be applied and/or the model application time can be determined according to any one of the following:
  • the subcarrier spacing of the BWP activated on the carrier to which the first AI model is applied is applied.
  • the moment when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the downlink channel carrying the first information.
  • the time when the first AI model starts to be applied and/or the model application time is determined according to the subcarrier spacing of PBCH. If the first information is carried via Group Common DCI, the time when the first AI model starts to be applied and/or the model application time is determined according to the subcarrier spacing of PDCCH carrying the DCI.
  • the time when the model starts to be applied is time slot n+k+1.
  • the terminal device can obtain the time length between time slot n+k+1 and time slot n based on the subcarrier spacing carrying the first information, thereby determining the time for model application.
  • the moment when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the uplink channel of the HARQ-ACK information carrying the first information.
  • the time when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the uplink channel where the HARQ-ACK information of the PDSCH carrying the high-level signaling is located.
  • the terminal device can obtain the time length between time slot n+k+3N+1 and time slot n based on the determined subcarrier spacing, thereby determining the time for model application.
  • the moment when the first AI model starts to be applied and/or the model application time may be based on the subcarrier spacing of the BWP activated on the carrier to which the first AI model is applied.
  • the time when the first AI model starts to be applied and/or the model application time can be determined according to the first subcarrier spacing.
  • the first subcarrier spacing is any subcarrier spacing of the BWPs respectively activated on multiple carriers to which the first AI model is applied.
  • the first subcarrier spacing can be agreed upon by the network device and the terminal device.
  • the first subcarrier spacing may be the smallest subcarrier spacing among the BWPs respectively activated on multiple carriers to which the first AI model is applied.
  • the carriers to which the first AI model is applied are CC1, CC2, and CC3, respectively, wherein the BWPs activated on the three CCs are BWP1, BWP2, and BWP3, respectively, and the subcarrier spacing configured for BWP1 is less than the subcarrier spacing configured for BWP2 and less than the subcarrier spacing configured for BWP3.
  • the moment when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing configured for BWP1.
  • the first AI may be determined according to any of the following:
  • the subcarrier spacing of the downlink channel carrying the response information of the second information is the subcarrier spacing of the downlink channel carrying the response information of the second information
  • the subcarrier spacing of the BWP activated on the carrier to which the first AI model is applied is applied.
  • the moment when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the uplink channel carrying the second information.
  • the time when the first AI model starts to be applied and/or the model application time is determined according to the subcarrier spacing of the PUCCH of the UCI. If the second information is carried by high-level signaling, the time when the first AI model starts to be applied and/or the model application time is determined according to the subcarrier spacing of the PUSCH carrying the high-level signaling. If the second information is carried by PRACH, the time when the first AI model starts to be applied and/or the model application time is determined according to the subcarrier spacing of the PRACH.
  • the moment when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the downlink channel of the response information carrying the second information.
  • the time when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the downlink channel where the HARQ-ACK information of the PUSCH carrying the high-level signaling is located. If the second information is carried by PRACH, the time when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the downlink channel where the corresponding RAR is located.
  • the moment when the first AI model starts to be applied and/or the model application time may be based on the subcarrier spacing of the BWP activated on the carrier to which the first AI model is applied.
  • the time when the first AI model starts to be applied and/or the model application time can be determined according to the first subcarrier spacing.
  • the first subcarrier spacing is any subcarrier spacing of the BWPs respectively activated on multiple carriers to which the first AI model is applied.
  • the first subcarrier spacing can be agreed upon by the network device and the terminal device.
  • the first subcarrier spacing may be the smallest subcarrier spacing among the BWPs respectively activated on multiple carriers to which the first AI model is applied.
  • the carriers to which the first AI model is applied are CC1, CC2, and CC3, respectively, wherein the BWPs activated on the three CCs are BWP1, BWP2, and BWP3, respectively, and the subcarrier spacing configured for BWP1 is less than the subcarrier spacing configured for BWP2 and less than the subcarrier spacing configured for BWP3.
  • the moment when the first AI model starts to be applied and/or the model application time can be determined according to the subcarrier spacing configured for BWP1.
  • the terminal device and the network device cannot determine the actual length of time, which may result in different time lengths determined by both sides.
  • Embodiment 1 is a diagrammatic representation of Embodiment 1:
  • the network device may configure the AI model for the terminal devices within its coverage through broadcast signaling (that is, the first information is carried through the broadcast signaling).
  • This embodiment may include the following steps:
  • the terminal device reports the supported model application time through UE capability information.
  • model application time is the time interval between the terminal device receiving the broadcast signaling for configuring the AI model and applying the AI model, where the time interval is in time slots.
  • the model application time reported by the terminal device may include:
  • the model application time corresponding to one or more AI models supported by the terminal device
  • Model application time corresponding to one or more model update modes supported by the terminal device
  • the model application time corresponding to one or more model types supported by the terminal device.
  • the network device indicates the model application time to the terminal device according to the UE capabilities reported by the terminal device.
  • the network device needs to consider the model application time reported by multiple terminal devices within the coverage area to determine the indicated model application time.
  • the network device can use the maximum value of the model application time reported by multiple terminal devices as the model application time used by each terminal within the coverage area, and indicate it to these terminal devices.
  • the model application time may be indicated by broadcast signaling.
  • the broadcast signaling may be PBCH, SIB, or Group Common DCI.
  • the broadcast signaling indicating the model application time may be the same signaling as the broadcast signaling indicating the AI model, or may be different signaling, and the embodiments of the present application do not limit this.
  • the model application time may also be indicated by high-layer signaling or physical layer signaling.
  • the signaling may be a MAC layer signaling indication for a model failure recovery response, or a MAC layer signaling for a model update.
  • the terminal device receives the broadcast signaling sent by the network device for configuring the AI model.
  • the terminal device can receive at least one of the model ID, functionality, model structure and model parameters of the AI model configured by the network device through broadcast signaling.
  • the network device does not have to configure the entire AI model, but may only configure or update part of the structure or part of the model parameters in the AI model.
  • the time when the AI model starts to be applied is the time when the updated model starts to be applied.
  • the terminal device can use the previously adopted model to transmit information with the network device.
  • the time when the AI model starts to be applied is the time when the updated model starts to be applied.
  • the terminal device can use the model corresponding to the previous function to transmit information with the network device.
  • the function here can also be a function ID.
  • the time when the AI model starts to be applied is the time when the updated model structure and model parameters start to be applied.
  • the terminal device can use the previously adopted model structure and model parameters to transmit information with the network device.
  • the time when the AI model starts to be applied is the time when the updated model parameters start to be applied.
  • the terminal device can use the previously used model parameters to transmit information with the network device.
  • a network device can indicate the AI model used to a group of terminal devices through Group Common DCI, which is transmitted through the common search space CSS and encrypted using a common RNTI.
  • the network device while the network device configures the AI model through broadcast signaling, it can also indicate the corresponding model application time through the same signaling.
  • the network device can deploy the same AI model to a group of terminal devices and indicate the same model application time, and this group of terminal devices can apply the AI model at the same time.
  • this group of terminal devices can apply the AI model at the same time.
  • only one AI model needs to be deployed on the network side as the corresponding encoder/decoder model, and there is no need to deploy a corresponding model for each terminal device, which significantly reduces the implementation complexity of the network device and also reduces the overhead of downlink signaling.
  • the terminal device determines the time to start applying the AI model based on the model application time configured by the network device.
  • the moment when the AI model starts to be applied is: the first time slot after k time slots from the time slot when the terminal device receives the broadcast signaling used to configure the AI model; k is the model application time indicated by the network device (in time slots).
  • the time when the AI model starts to be applied is the first time slot after time slot n+k, that is, it starts to be applied in time slot n+k+1.
  • the moment when the AI model starts to be applied and/or the model application time are parameters in time slots, and the corresponding time length/moment needs to be determined according to the subcarrier spacing, where different subcarrier spacings correspond to different time lengths.
  • the time when the AI model configured by the network device starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the channel carrying the broadcast signaling.
  • the terminal device uses the AI model configured by broadcast signaling to transmit information with the network device.
  • the terminal device may use a predefined AI model to transmit information with the network device, or the terminal device may not transmit information with the network device based on the AI model.
  • the predefined AI model is an AI model agreed upon in advance by the terminal device and the network device. The terminal device does not transmit information with the network device based on the AI model, that is, the terminal device uses a traditional non-AI method to transmit information with the network device.
  • the terminal device may use the AI model for any of the following processes:
  • the network device uses the network side model corresponding to the AI model configured by the above broadcast signaling to transmit information between the terminal device.
  • the network device may use the network side model corresponding to the configured AI model for any of the following processes:
  • modules of the network equipment application model correspond one-to-one to the modules of the AI model applied on the terminal side, such as channel coding-channel decoding, modulation-demodulation, pilot generation-channel estimation, transmitting RF signal processing-receiving RF signal processing, etc.
  • the network device may configure the AI model for the terminal device through high-level signaling (that is, the first information is carried through high-level signaling).
  • This embodiment may include the following steps:
  • the terminal device reports the supported model application time through UE capabilities.
  • model application time is the time interval between the terminal device feeding back the HARQ-ACK information of the high-level signaling used to configure the AI model and the application of the AI model.
  • the time interval is in time slots or OFDM symbols.
  • the above-mentioned high-level signaling may be RRC signaling or MAC layer signaling (MAC CE).
  • MAC CE MAC layer signaling
  • the HARQ-ACK information of the above-mentioned high-layer signaling is the HARQ-ACK information of the PDSCH carrying the first information.
  • the terminal device receives the AI model configured by the network device through high-level signaling.
  • the terminal device determines the time when the AI model configured by the network device starts to be applied based on the model application time reported in S1.
  • the time when the AI model configured by the high-level signaling starts to be applied may be: the first time slot after k time slots from the time slot where the HARQ-ACK information of the PDSCH carrying the high-level signaling is located.
  • k is the model application time, which can be determined by the model application time supported by the terminal device through UE capability reporting.
  • the typical high-layer signaling in this implementation manner is RRC signaling.
  • the time when the AI model configured by the high-level signaling starts to be applied may be: the first time slot after (k+3N) time slots from the time slot where the HARQ-ACK information of the PDSCH carrying the high-level signaling is located.
  • k is the application time of the model
  • N is the number of time slots contained in 1 ms, that is, 3N time slots are 3 ms.
  • the typical high-level signaling may be MAC layer signaling.
  • 3N time slots ie, 3 ms is the minimum time required for the MAC layer signaling to take effect, and the MAC layer signaling is ensured to have sufficient time to take effect by spacing k+3N time slots.
  • the time when the AI model configured by the high-level signaling starts to be applied may be: the first time slot after the time slot where the HARQ-ACK information of the PDSCH carrying the high-level signaling is located p time slots.
  • p is the larger value of k and 3N
  • k is the model application time (in time slots)
  • N is the number of time slots contained in 1ms, that is, 3N time slots are 3ms.
  • typical high-level signaling may be MAC layer signaling, and 3N time slots (ie, 3 ms) is the minimum time required for the MAC layer signaling to take effect. By taking the larger value of k and 3N, it is ensured that the MAC layer signaling has sufficient time to take effect.
  • the moment when the AI model starts to be applied and/or the model application time are parameters in time slots, and the corresponding time length/moment needs to be determined based on the subcarrier spacing, where different subcarrier spacings correspond to different time lengths.
  • the time when the AI model configured by the network device starts to be applied and/or the model application time can be based on the carrier
  • the subcarrier spacing of the uplink channel where the HARQ-ACK information of the PDSCH of the high-layer signaling is located is determined.
  • the time when the AI model configured by the network device starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the BWP activated on the carrier of the AI model.
  • the moment when the AI model starts to be applied and/or the model application time are determined according to the smallest subcarrier spacing among the BWPs activated on the multiple carriers. In this way, even in a multi-carrier scenario, the network device and the terminal device can determine a unique moment to start application.
  • the terminal device uses the AI model configured by the above-mentioned high-level signaling to transmit information with the network device.
  • the network device uses the network side model corresponding to the AI model configured by the above-mentioned high-level signaling to transmit information between the terminal device.
  • the network device may configure the AI model for the terminal device through the DCI (that is, the first information is carried through the DCI).
  • This embodiment may include the following steps:
  • the terminal device reports the supported model application time through UE capabilities.
  • model application time is the time interval between the terminal device feeding back the HARQ-ACK information of the DCI signaling used to configure the AI model and the application of the AI model.
  • the time interval is in time slots or OFDM symbols.
  • the terminal device receives the AI model configured by the network device through DCI signaling.
  • the network device can configure the AI model for the terminal device through the DCI signaling transmitted in the UE-specific search space.
  • the DCI signaling can be a DCI sent by the network device for model failure recovery response, or a DCI for scheduling PDSCH or PUSCH.
  • the transmission of PDSCH or PUSCH can be based on the AI model (not necessarily the configured AI model).
  • the terminal device determines the time when the AI model configured by the network device starts to be applied based on the model application time reported in S1.
  • the AI model configured by the DCI signaling may start to be applied at the first time slot at least k OFDM symbols after the last OFDM symbol of the uplink channel carrying the HARQ-ACK information of the DCI, where k is the model application time (in units of OFDM symbols).
  • the time when the AI model starts to be applied is the first time slot after the time slot where symbol m+k is located.
  • the time when the AI model configured by the DCI signaling starts to be applied may be: the first OFDM symbol at least k OFDM symbols after the last OFDM symbol of the uplink channel carrying the HARQ-ACK information of the DCI, wherein k is the model application time (in units of OFDM symbols).
  • the moment when the AI model starts to apply is the first symbol after symbol m+k, that is, symbol m+k+1.
  • the moment when the AI model starts to be applied and/or the model application time are parameters in units of time slots/OFDM, and the corresponding time length/moment needs to be determined based on the subcarrier spacing, where different subcarrier spacings correspond to different time lengths.
  • the time when the AI model configured by the network device starts to be applied and/or the model application time can be determined according to the subcarrier spacing of the BWP activated on the carrier of the AI model.
  • the moment when the AI model starts to be applied and/or the model application time are determined according to the smallest subcarrier spacing among the BWPs activated on the multiple carriers. In this way, even in a multi-carrier scenario, the network device and the terminal device can determine a unique moment to start application.
  • the terminal device uses the AI model configured by the above-mentioned DCI signaling to transmit information between the terminal device and the network device.
  • the network device uses the network side model corresponding to the AI model configured by the above DCI signaling to transmit information between the terminal device.
  • the wireless communication method provided in the embodiment of the present application can ensure that the model application time on both sides is the same, and the network device side is The model can be updated synchronously between the equipment and different terminals, thus ensuring that the model output of the other end is available.
  • the size of the sequence number of each process does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.
  • downlink indicates that the transmission direction of the signal or data
  • uplink is used to indicate that the transmission direction of the signal or data is the second direction sent from the user equipment of the cell to the site
  • side is used to indicate that the transmission direction of the signal or data is the third direction sent from user equipment 1 to user equipment 2.
  • downlink signal indicates that the transmission direction of the signal is the first direction.
  • the term "and/or” is only a description of the association relationship of the associated objects, indicating that three relationships can exist. Specifically, A and/or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character “/" in this article generally indicates that the front and back associated objects are in an "or" relationship.
  • FIG. 10 is a schematic diagram of the structure of a wireless communication device 1000 provided in an embodiment of the present application, which is applied to a terminal device. As shown in FIG. 10 , the wireless communication device 1000 includes:
  • the first determining unit 1001 is configured to determine a time point when the first AI model starts to be applied according to the model application time reported to the network device, or according to the model application time configured by the network device;
  • the first communication unit 1002 is configured to communicate with the network device using the first AI model after the moment.
  • the model application time is any one of the following:
  • the time interval between when the terminal device receives the first information and when the first AI model is applied is applied
  • the first information is downlink information used to configure the first AI model, and the time interval is in units of time slots or OFDM symbols.
  • the time when the first AI model starts to be applied is any one of the following:
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the uplink channel of the HARQ-ACK information carrying the first information starts k OFDM symbols;
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the downlink channel carrying the first information starts k OFDM symbols;
  • the first time slot after the time slot where the terminal device receives the first information starts with k time slots
  • the first time slot after k time slots, (k+3N) time slots, or max(k,3N) time slots from the time slot where the uplink channel of the HARQ-ACK information carrying the first information is located;
  • the first information is downlink information used to configure the first AI model
  • k is the model application time
  • N is the number of time slots contained in 1ms.
  • the wireless communication device 1000 further includes a receiving unit configured to receive first information, where the first information is downlink information for configuring the first AI model, wherein the first information includes one or more of the following:
  • the model application time is any one of the following:
  • the terminal device After the terminal device sends the second information, the terminal device sends feedback information obtained based on the first AI model the time interval between moments;
  • the second information is uplink information used to indicate the first AI model, and the time interval is in units of time slots or OFDM symbols.
  • the time when the first AI model starts to be applied is any one of the following:
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the uplink channel carrying the second information starts k OFDM symbols;
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the response information of the uplink channel carrying the second information starts k OFDM symbols;
  • the first time slot after k time slots from the time slot where the terminal device sends the second information
  • the first time slot after k time slots, (k+3N) time slots, or max(k,3N) time slots from the time slot where the downlink channel carrying the response information of the second information is located;
  • the second information is uplink information used to indicate the first AI model
  • k is the model application time
  • N is the number of time slots contained in 1ms.
  • the wireless communication device 1000 further includes a sending unit configured to send second information, where the second information is uplink information for indicating the first AI model;
  • the second information includes one or more of the following:
  • the model application time reported to the network device includes one or more of the following:
  • model application time corresponding to one or more AI models supported by the terminal device; wherein the one or more AI models supported by the terminal device include the first AI model;
  • the model application time corresponding to the one or more model functions supported by the terminal device; wherein the one or more model functions supported by the terminal device include the function of the first AI model;
  • the model application time corresponding to the one or more model update modes supported by the terminal device; wherein the one or more model update modes supported by the terminal device include the update mode of the first AI model;
  • the model application time corresponding to the one or more model types supported by the terminal device; wherein the one or more model types supported by the terminal device include the type of the first AI model.
  • the model updating method includes: updating model parameters and/or updating model structure.
  • the model types include:
  • High-precision models and low-precision models or, known models and unknown models.
  • the moment when the first AI model starts to be applied and/or the model application time are determined according to any one of the following:
  • the subcarrier spacing of the downlink channel carrying the response information of the second information is the subcarrier spacing of the downlink channel carrying the response information of the second information
  • the first information is downlink information used to configure the first AI model
  • the second information is uplink information used to indicate the first AI model
  • the number of carriers to which the first AI model is applied is multiple,
  • the moment when the first AI model starts to be applied and/or the model application time is determined according to a first subcarrier spacing, where the first subcarrier spacing is the smallest subcarrier spacing among the BWPs respectively activated on multiple carriers to which the first AI model is applied.
  • the terminal device before the terminal device applies the first AI model, transmits information with the network device based on a third AI model, or the terminal device transmits information with the network device not based on an AI model;
  • the third AI model is a predefined AI model or a historical AI model.
  • FIG. 11 is a schematic diagram of the structure of a wireless communication device 1100 provided in an embodiment of the present application, which is applied to a network device. As shown in FIG. 11 , the wireless communication device 1100 includes:
  • the second determining unit 1101 is configured to determine the time when the first AI model starts to be applied according to the model application time reported by the terminal device, or according to the model application time configured for the terminal device;
  • the second communication unit 1102 is configured to communicate with the terminal device using a second AI model corresponding to the first AI model after the moment.
  • the model application time is any one of the following:
  • the time interval between when the terminal device receives the first information and when the first AI model is applied is applied
  • the first information is downlink information used to configure the first AI model, and the time interval is in units of time slots or OFDM symbols.
  • the time when the first AI model starts to be applied is any one of the following:
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the uplink channel of the HARQ-ACK information carrying the first information starts k OFDM symbols;
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the downlink channel carrying the first information starts k OFDM symbols;
  • the first time slot after the time slot where the terminal device receives the first information starts with k time slots
  • the first time slot after k time slots, (k+3N) time slots, or max(k,3N) time slots from the time slot where the uplink channel of the HARQ-ACK information carrying the first information is located;
  • the first information is downlink information used to configure the first AI model
  • k is the model application time
  • N is the number of time slots contained in 1ms.
  • the wireless communication device 1100 further includes a sending unit configured to send first information to the terminal device, where the first information is a downlink for configuring the first AI model, wherein the first information includes one or more of the following:
  • the model application time is any one of the following:
  • the second information is uplink information used to indicate the first AI model, and the time interval is in units of time slots or OFDM symbols.
  • the time when the first AI model starts to be applied is any one of the following:
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the uplink channel carrying the second information starts k OFDM symbols;
  • the first time slot or the first OFDM symbol after the last OFDM symbol of the response information of the uplink channel carrying the second information starts k OFDM symbols;
  • the first time slot after k time slots from the time slot where the terminal device sends the second information
  • the first time slot after k time slots, (k+3N) time slots, or max(k,3N) time slots from the time slot where the downlink channel carrying the response information of the second information is located;
  • the second information is uplink information for indicating the first AI model, k is the model application time, and N is 1ms. The number of timeslots included.
  • the wireless communication device 1100 further includes a receiving unit configured to receive second information sent by the terminal device, where the second information is uplink information indicating the first AI model;
  • the second information includes one or more of the following:
  • the model application time reported by the terminal device includes one or more of the following:
  • model application time corresponding to one or more AI models supported by the terminal device; wherein the one or more AI models supported by the terminal device include the first AI model;
  • the model application time corresponding to the one or more model functions supported by the terminal device; wherein the one or more model functions supported by the terminal device include the function of the first AI model;
  • the model application time corresponding to the one or more model update modes supported by the terminal device; wherein the one or more model update modes supported by the terminal device include the update mode of the first AI model;
  • the model application time corresponding to the one or more model types supported by the terminal device; wherein the one or more model types supported by the terminal device include the type of the first AI model.
  • the model updating method includes: updating model parameters and/or updating model structure.
  • the model types include:
  • High-precision models and low-precision models or, known models and unknown models.
  • the model application time configured by the network device for the terminal device is determined based on the model application time reported by the terminal device.
  • the duration of the model application time configured by the network device for the terminal device is greater than or equal to the duration of the model application time reported by the terminal device.
  • the moment when the first AI model starts to be applied and/or the model application time are determined according to any one of the following:
  • the subcarrier spacing of the downlink channel carrying the response information of the second information is the subcarrier spacing of the downlink channel carrying the response information of the second information
  • the first information is downlink information used to configure the first AI model
  • the second information is uplink information used to indicate the first AI model
  • the number of carriers to which the first AI model is applied is multiple;
  • the moment when the first AI model starts to be applied and/or the model application time is determined according to a first subcarrier spacing, where the first subcarrier spacing is the smallest subcarrier spacing among the BWPs respectively activated on multiple carriers to which the first AI model is applied.
  • the network device before the network device applies the first AI model, transmits information with the terminal device based on a fourth AI model corresponding to the third AI model, or the network device transmits information with the terminal device not based on the AI model;
  • the third AI model is a predefined model or a historical AI model.
  • FIG12 is a schematic structural diagram of a communication device 1200 provided in an embodiment of the present application.
  • the communication device may be a terminal device or a network device.
  • the communication device 1200 shown in FIG12 includes a processor 1210, which may call and run a computer program from a memory to implement the method in the embodiment of the present application.
  • the communication device 1200 may further include a memory 1220.
  • the processor 1210 may call and run a computer program from the memory 1220 to implement the method in the embodiment of the present application.
  • the memory 1220 may be a separate device independent of the processor 1210 , or may be integrated into the processor 1210 .
  • the communication device 1200 may further include a transceiver 1230 , and the processor 1210 may control the transceiver 1230 to communicate with other devices, specifically, may send information or data to other devices, or receive information or data sent by other devices.
  • the transceiver 1230 may include a transmitter and a receiver.
  • the transceiver 1230 may further include an antenna, and the number of antennas may be one or more.
  • the communication device 1200 may specifically be a network device of an embodiment of the present application, and the communication device 1800 may implement corresponding processes implemented by the network device in each method of the embodiment of the present application, which will not be described in detail here for the sake of brevity.
  • the communication device 1200 may specifically be a mobile terminal/terminal device of an embodiment of the present application, and the communication device 1200 may implement the corresponding processes implemented by the mobile terminal/terminal device in each method of the embodiment of the present application, which will not be described again for the sake of brevity.
  • Fig. 13 is a schematic structural diagram of a chip according to an embodiment of the present application.
  • the chip 1300 shown in Fig. 13 includes a processor 1910, and the processor 1310 can call and run a computer program from a memory to implement the method according to the embodiment of the present application.
  • the chip 1300 may further include a memory 1320.
  • the processor 1310 may call and run a computer program from the memory 1320 to implement the method in the embodiment of the present application.
  • the memory 1320 may be a separate device independent of the processor 1310 , or may be integrated into the processor 1310 .
  • the chip 1300 may further include an input interface 1330.
  • the processor 1310 may control the input interface 1330 to communicate with other devices or chips, and specifically, may obtain information or data sent by other devices or chips.
  • the chip 1300 may further include an output interface 1340.
  • the processor 1310 may control the output interface 1340 to communicate with other devices or chips, and specifically, may output information or data to other devices or chips.
  • the chip can be applied to the network device in the embodiments of the present application, and the chip can implement the corresponding processes implemented by the network device in each method of the embodiments of the present application. For the sake of brevity, they will not be repeated here.
  • the chip can be applied to the mobile terminal/terminal device in the embodiments of the present application, and the chip can implement the corresponding processes implemented by the mobile terminal/terminal device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.
  • the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.
  • the embodiments of the present application also provide a computer storage medium, which stores one or more programs.
  • the one or more programs can be executed by one or more processors to implement the method in the embodiments of the present application.
  • FIG14 is a schematic block diagram of a communication system 1400 provided in an embodiment of the present application.
  • the communication system 1400 includes a terminal device 1410 and a network device 1420 .
  • the terminal device 1410 can be used to implement the corresponding functions implemented by the terminal device in the above method
  • the network device 1420 can be used to implement the corresponding functions implemented by the network device in the above method.
  • the terminal device 1410 can be used to implement the corresponding functions implemented by the terminal device in the above method
  • the network device 1420 can be used to implement the corresponding functions implemented by the network device in the above method.
  • the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capabilities.
  • each step of the above method embodiment can be completed by the hardware integrated logic circuit in the processor or the instruction in the form of software.
  • the above processor can be a general processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.
  • DSP Digital Signal Processor
  • ASIC Application Specific Integrated Circuit
  • FPGA Field Programmable Gate Array
  • the methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed.
  • the general processor can be a microprocessor or the processor can also be any conventional processor, etc.
  • the steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to execute, or the hardware and software modules in the decoding processor can be executed.
  • the software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc.
  • the storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.
  • the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories.
  • the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory.
  • the volatile memory can be a random access memory (RAM), which is used as an external cache.
  • RAM Direct Rambus RAM
  • SRAM Static RAM
  • DRAM Dynamic RAM
  • SDRAM Synchronous DRAM
  • DDR SDRAM Double Data Rate SDRAM
  • ESDRAM Enhanced SDRAM
  • SLDRAM Synchlink DRAM
  • DR RAM Direct Rambus RAM
  • the memory in the embodiment of the present application may also be a static random access memory (static RAM, SRAM), a dynamic random access memory (dynamic RAM, DRAM), a synchronous dynamic random access memory (SDRAM), or a synchronous dynamic random access memory (SDRAM).
  • the memory in the embodiments of the present invention includes synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct RAM bus random access memory (DR RAM).
  • SDRAM synchronous DRAM
  • DDR SDRAM double data rate synchronous dynamic random access memory
  • ESDRAM enhanced synchronous dynamic random access memory
  • SLDRAM synchronous link dynamic random access memory
  • DR RAM direct RAM bus random access memory
  • An embodiment of the present application also provides a computer-readable storage medium for storing a computer program.
  • the computer-readable storage medium can be applied to the network device in the embodiments of the present application, and the computer program enables the computer to execute the corresponding processes implemented by the network device in the various methods of the embodiments of the present application. For the sake of brevity, they are not repeated here.
  • the computer-readable storage medium can be applied to the mobile terminal/terminal device in the embodiments of the present application, and the computer program enables the computer to execute the corresponding processes implemented by the mobile terminal/terminal device in the various methods of the embodiments of the present application. For the sake of brevity, they are not repeated here.
  • An embodiment of the present application also provides a computer program product, including computer program instructions.
  • the computer program product can be applied to the network device in the embodiments of the present application, and the computer program instructions enable the computer to execute the corresponding processes implemented by the network device in the various methods of the embodiments of the present application. For the sake of brevity, they are not repeated here.
  • the computer program product can be applied to the mobile terminal/terminal device in the embodiments of the present application, and the computer program instructions enable the computer to execute the corresponding processes implemented by the mobile terminal/terminal device in the various methods of the embodiments of the present application. For the sake of brevity, they are not repeated here.
  • the embodiment of the present application also provides a computer program.
  • the computer program can be applied to the network device in the embodiments of the present application.
  • the computer program runs on a computer, the computer executes the corresponding processes implemented by the network device in the various methods in the embodiments of the present application. For the sake of brevity, they are not described here.
  • the computer program can be applied to the mobile terminal/terminal device in the embodiments of the present application.
  • the computer program runs on the computer, the computer executes the corresponding processes implemented by the mobile terminal/terminal device in the various methods of the embodiments of the present application. For the sake of brevity, they are not repeated here.
  • the disclosed systems, devices and methods can be implemented in other ways.
  • the device embodiments described above are only schematic.
  • the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.
  • Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
  • each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.
  • the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art.
  • the computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

一种无线通信方法及装置、终端设备、网络设备,该方法包括:终端设备根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定第一AI模型开始应用的时刻;终端设备在该时刻之后,采用第一AI模型与网络设备进行通信。

Description

无线通信方法及装置、终端设备、网络设备 技术领域
本申请实施例涉及移动通信技术领域,具体涉及一种无线通信方法及装置、终端设备、网络设备。
背景技术
鉴于人工智能(Artificial Intelligence,AI)技术在计算机视觉、自然语言处理等方面取得了巨大的成功,通信领域开始尝试利用AI技术来寻求新的技术思路来解决传统方法受限的技术难题。
AI模型用于无线通信时,可以分为单端模型和双端模型。双端模型需要在终端设备和网络设备侧成对部署。示例性的,用于信道状态信息(Channel State Information,CSI)反馈的AI模型就是一种典型的双端模型,其中,终端设备可以将得到的信道信息(例如特征向量,波束信息,时延信息等)作为AI模型的输入,经AI模型推理出相应的CSI量化比特。相应的,网络设备侧对应的AI模型,可以将CSI量化比特作为输入,反向推理得到相应的信道信息。
对于双端模型,如果当前终端设备侧的AI模型性能较差,网络设备可以通过信令指示终端设备进行模型更新(例如更换一个AI模型或者调整模型参数)。此时,网络设备也需要进行相应的模型更新,以使两侧的AI模型仍然是互相匹配的。如果两侧AI模型的更新不是完全同步的,会导致收端AI模型的输出是错误的,从而影响终端设备和网络设备之间的信息传输。如何保证网络设备与终端设备之间同步进行模型更新是需要解决的问题。
发明内容
本申请实施例提供一种信息传输方法及装置、终端设备、网络设备。
第一方面,本申请实施例提供的无线通信方法,包括:
终端设备根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
所述终端设备在所述时刻之后,采用所述第一AI模型与所述网络设备进行通信。
第二方面,本申请实施例提供的无线通信方法,包括:
网络设备根据终端设备上报的模型应用时间,或者,根据为所述终端设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
所述网络设备在所述时刻之后,采用所述第一AI模型对应的第二AI模型与所述终端设备进行通信。
第三方面,本申请实施例提供的无线通信装置,应用于终端设备,所述装置包括:
第一确定单元,配置为根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
第一通信单元,配置为在所述时刻之后,采用所述第一AI模型与所述网络设备进行通信。
第四方面,本申请实施例提供的无线通信装置,应用于网络设备,所述装置包括:
第二确定单元,配置为根据终端设备上报的模型应用时间,或者,根据为所述终端设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
第二通信单元,配置为在所述时刻之后,采用所述第一AI模型对应的第二AI模型与所述终端设备进行通信。
第五方面,本申请实施例提供的终端设备,该终端设备包括处理器和存储器。该存储器用于存储计算机程序,该处理器用于调用并运行该存储器中存储的计算机程序,执行上述的无线通信方法。
第六方面,本申请实施例提供的网络设备,该网络设备包括处理器和存储器。该存储器用于存储计算机程序,该处理器用于调用并运行该存储器中存储的计算机程序,执行上述的无线通信方法。
本申请实施例提供的芯片,用于实现上述的无线通信方法。
具体地,该芯片包括:处理器,用于从存储器中调用并运行计算机程序,使得安装有该芯片的 设备执行上述的无线通信方法。
本申请实施例提供的计算机可读存储介质,用于存储计算机程序,该计算机程序使得计算机执行上述的无线通信方法。
本申请实施例提供的计算机程序产品,包括计算机程序指令,该计算机程序指令使得计算机执行上述的无线通信方法。
本申请实施例提供的计算机程序,当其在计算机上运行时,使得计算机执行上述的无线通信方法。
本申请实施例提供的无线通信方法中,在终端设备侧和网络设备侧采用相对应的双端AI模型时,如果终端设备侧的AI模型发生更新或重配置,终端设备可以根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定更新或重配置后的AI模型开始应用的时刻。这样,可以确保两侧的模型开始应用的时刻是相同的,从而保证AI模型的输出是可用的,提高了通信系统的性能。
附图说明
此处所说明的附图用来提供对本申请的进一步理解,构成本申请的一部分,本申请的示意性实施例及其说明用于解释本申请,并不构成对本申请的不当限定。在附图中:
图1是本申请实施例提供的一种通信架构示意图;
图2是本申请实施例提供的一种神经元结构示意图;
图3是本申请实施例提供的一种神经网络结构示意图;
图4是本申请实施例提供的一种用于CSI反馈的神经网络示意图;
图5是本申请实施例提供的一种无线通信方法流程示意图;
图6A是本申请实施例提供的一种时隙结构示意图一;
图6B是本申请实施例提供的一种时隙结构示意图二;
图7A是本申请实施例提供的一种时隙结构示意图三;
图7B是本申请实施例提供的一种时隙结构示意图四;
图8是本申请实施例提供的一种时隙结构示意图五;
图9A是本申请实施例提供的一种时隙结构示意图六;
图9B是本申请实施例提供的一种时隙结构示意图七;
图9C是本申请实施例提供的一种时隙结构示意图八;
图10是本申请实施例提供的一种无线通信装置1000的结构示意图;
图11是本申请实施例提供的一种无线通信装置1100的结构示意图;
图12是本申请实施例提供的一种通信设备示意性结构图;
图13是本申请实施例的芯片的示意性结构图;
图14是本申请实施例提供的一种通信系统的示意性框图。
具体实施方式
下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行描述,显然,所描述的实施例是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。
图1是本申请实施例的一个应用场景的示意图。
如图1所示,通信系统100可以包括终端设备110和网络设备120。网络设备120可以通过空口与终端设备110通信。终端设备110和网络设备120之间支持多业务传输。
应理解,本申请实施例仅以通信系统100进行示例性说明,但本申请实施例不限定于此。也就是说,本申请实施例的技术方案可以应用于各种通信系统,例如:长期演进(Long Term Evolution,LTE)系统、LTE时分双工(Time Division Duplex,TDD)、通用移动通信系统(Universal Mobile Telecommunication System,UMTS)、物联网(Internet of Things,IoT)系统、窄带物联网(Narrow Band Internet of Things,NB-IoT)系统、增强的机器类型通信(enhanced Machine-Type Communications,eMTC)系统、5G通信系统(也称为新无线(New Radio,NR)通信系统),或未来的通信系统等。
在图1所示的通信系统100中,网络设备120可以是与终端设备110通信的接入网设备。接入 网设备可以为特定的地理区域提供通信覆盖,并且可以与位于该覆盖区域内的终端设备110(例如UE)进行通信。
网络设备120可以是长期演进(Long Term Evolution,LTE)系统中的演进型基站(Evolutional Node B,eNB或eNodeB),或者是下一代无线接入网(Next Generation Radio Access Network,NG RAN)设备,或者是NR系统中的基站(gNB),或者是云无线接入网络(Cloud Radio Access Network,CRAN)中的无线控制器,或者该网络设备120可以为中继站、接入点、车载设备、可穿戴设备、集线器、交换机、网桥、路由器,或者未来演进的公共陆地移动网络(Public Land Mobile Network,PLMN)中的网络设备等。
终端设备110可以是任意终端设备,其包括但不限于与网络设备120或其它终端设备采用有线或者无线连接的终端设备。
例如,所述终端设备110可以指接入终端、用户设备(User Equipment,UE)、用户单元、用户站、移动站、移动台、远方站、远程终端、移动设备、用户终端、终端、无线通信设备、用户代理或用户装置。接入终端可以是蜂窝电话、无绳电话、会话启动协议(Session Initiation Protocol,SIP)电话、IoT设备、卫星手持终端、无线本地环路(Wireless Local Loop,WLL)站、个人数字处理(Personal Digital Assistant,PDA)、具有无线通信功能的手持设备、计算设备或连接到无线调制解调器的其它处理设备、车载设备、可穿戴设备、5G网络中的终端设备或者未来演进网络中的终端设备等。
终端设备110可以用于设备到设备(Device to Device,D2D)的通信。
无线通信系统100还可以包括与网络设备120进行通信的核心网设备130,该核心网设备130可以是5G核心网(5G Core,5GC)设备,例如,接入与移动性管理功能(Access and Mobility Management Function,AMF),又例如,认证服务器功能(Authentication Server Function,AUSF),又例如,用户面功能(User Plane Function,UPF),又例如,会话管理功能(Session Management Function,SMF)。可选地,核心网络设备130也可以是LTE网络的分组核心演进(Evolved Packet Core,EPC)设备,例如,会话管理功能+核心网络的数据网关(Session Management Function+Core Packet Gateway,SMF+PGW-C)设备。应理解,SMF+PGW-C可以同时实现SMF和PGW-C所能实现的功能。在网络演进过程中,上述核心网设备也有可能叫其它名字,或者通过对核心网的功能进行划分形成新的网络实体,对此本申请实施例不做限制。
通信系统100中的各个功能单元之间还可以通过下一代网络(next generation,NG)接口建立连接实现通信。
例如,终端设备通过NR接口与接入网设备建立空口连接,用于传输用户面数据和控制面信令;终端设备可以通过NG接口1(简称N1)与AMF建立控制面信令连接;接入网设备例如下一代无线接入基站(gNB),可以通过NG接口3(简称N3)与UPF建立用户面数据连接;接入网设备可以通过NG接口2(简称N2)与AMF建立控制面信令连接;UPF可以通过NG接口4(简称N4)与SMF建立控制面信令连接;UPF可以通过NG接口6(简称N6)与数据网络交互用户面数据;AMF可以通过NG接口11(简称N11)与SMF建立控制面信令连接;SMF可以通过NG接口7(简称N7)与PCF建立控制面信令连接。
图1示例性地示出了一个网络设备、一个核心网设备和两个终端设备,可选地,该无线通信系统100可以包括多个网络设备并且每个网络设备的覆盖范围内可以包括其它数量的终端设备,本申请实施例对此不做限定。
需要说明的是,图1只是以示例的形式示意本申请所适用的系统,当然,本申请实施例所示的方法还可以适用于其它系统。此外,本文中术语“系统”和“网络”在本文中常被可互换使用。本文中术语“和/或”,仅仅是一种描述关联对象的关联关系,表示可以存在三种关系,例如,A和/或B,可以表示:单独存在A,同时存在A和B,单独存在B这三种情况。另外,本文中字符“/”,一般表示前后关联对象是一种“或”的关系。还应理解,在本申请的实施例中提到的“指示”可以是直接指示,也可以是间接指示,还可以是表示具有关联关系。举例说明,A指示B,可以表示A直接指示B,例如B可以通过A获取;也可以表示A间接指示B,例如A指示C,B可以通过C获取;还可以表示A和B之间具有关联关系。还应理解,在本申请的实施例中提到的“对应”可表示两者之间具有直接对应或间接对应的关系,也可以表示两者之间具有关联关系,也可以是指示与被指示、配置与被配置等关系。还应理解,在本申请的实施例中提到的“预定义”或“预定义规则”可以通过在设备(例如,包括终端设备和网络设备)中预先保存相应的代码、表格或其他可用于指示相关信息的方式来实现,本申请对于其具体的实现方式不做限定。比如预定义可以是指协议中定义的。还应理解,本申请实施例中,所述“协议”可以指通信领域的标准协议,例如可以包括LTE协议、 NR协议以及应用于未来的通信系统中的相关协议,本申请对此不做限定。
为便于理解本申请实施例的技术方案,以下对本申请实施例的相关技术进行说明,以下相关技术作为可选方案与本申请实施例的技术方案可以进行任意结合,其均属于本申请实施例的保护范围。
AI模型是一种可以胜任多种任务处理的模型,它们具备自我学习和自我适应的能力,可以根据环境的变化来进行动态调整和决策。AI模型也可以称为机器学习(Machine Learning,ML)模型,两者等同或者可替换。
实际应用中,AI模型可以由神经网络构成。神经网络是一种由多个神经元节点相互连接构成的运算模型,其中节点间的连接代表从输入信号到输出信号的加权值,称为权重;每个节点对不同的输入信号进行加权求和,并通过特定的激活函数输出。参考图2所示的神经元结构示意图,其中,a1、a2、……、an和1为神经元的输入,w1、w2、……、wn和b表示权重,Sum表示求和函数,f表示激活函数,t为输出结果。
一个简单的神经网络如图3所示,包含输入层、隐藏层和输出层,通过多个神经元不同的连接方式,权重和激活函数,可以产生不同的输出,进而拟合从输入到输出的映射关系。每一个上一级节点都与其全部的下一级节点相连。此全连接模型也可以叫做DNN,即深度神经网络。
通过数据集的构建,训练,验证和测试等过程可以训练并得到一个AI模型。训练可以分为离线训练和在线训练。可以通过数据集离线训练的方式得到一个静态的训练结果,这里可以称之为离线训练。在网络设备或终端设备对AI模型的使用过程中,随着终端设备的进一步测量和/或上报,网络设备可以继续收集更多的数据,进行实时的在线训练来优化AI模型的参数,达到更好的推断和预测结果。在得到AI模型后,通过将当前得到的信息输入到AI模型中,可以推理得到相应的模型输出。
AI用于无线通信时,可以分为单端模型和双端模型。其中,单端模型只在终端设备或网络设备一侧部署即可使用,AI模型的训练也可以只在单侧进行即可;双端模型需要在终端设备和网络设备侧成对部署,两侧的模型需要一起进行训练,也就是说两侧部署的模型是对应的,不能单独使用或者更新。
示例性的,用于信道状态信息(Channel State Information,CSI)反馈的AI模型就是一种典型的双端模型。参考图4所示的一种用于CSI反馈的双端AI模型结构示意图。其中,在终端侧部署用于编码(encoder)的AI模型,在网络侧部署与之相对应的用于解码(decoder)的AI模型。终端设备基于encoder模型输出CSI量化比特,典型的如PMI比特,再通过上行信道(PUSCH/PUCCH)反馈给网络设备。网络设备将终端反馈的CSI量化比特(PMI)作为decoder模型的输入,从而输出与终端侧输入对应的信道信息,如各个子带的特征向量等,用于下行的预编码。同时,终端设备还需要进行模型性能监测,在模型性能不好时上报给网络设备,让网络设备进行模型的更新(例如模型结构的更新或者模型参数的更新)。由于两侧的AI模型是匹配的,如果网络设备进行模型更新,需要通过信令通知终端设备,令终端设备也更新成对应的AI模型,否则网络设备输出的结果就是不可用的。
需要说明的是,上述基于双端部署的AI模型不仅可以用于CSI反馈,也可以用于终端设备和网络设备侧具有对应操作/结构的其他处理方式,例如信道编码-信道解码,调制-解调,导频生成-信道估计,收发射频信号处理等,只需要训练出对应的终端设备侧和网络设备侧的一套AI模型即可。
可以理解的,针对双端模型,如果当前终端设备侧的模型性能不好,网络设备可以通过信令指示终端设备进行模型更新(例如更换一个AI模型或者调整模型参数)。此时,网络设备也需要进行相应的模型更新,以使两侧的AI模型仍然是互相匹配的。如果两侧AI模型的更新不是完全同步的,会导致收端AI模型的输出是错误的,从而影响终端设备和网络设备之间的信息传输。
另外,由于不同终端设备的处理能力不同,部署AI模型所需要的时间也不同,如何保证网络设备与不同终端设备之间能够同步进行模型更新是需要解决的问题。
基于此,本申请实施例提供一种无线通信方法,其中,在终端设备侧和网络设备侧采用相对应的双端AI模型时,如果终端设备侧的AI模型发生更新或重配置,终端设备可以根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定网络设备配置的AI模型开始应用的时刻。这样,可以确保两侧的模型应用时刻是相同的,从而保证AI模型的输出是可用的,提高了通信系统的性能。
为便于理解本申请实施例的技术方案,以下通过具体实施例详述本申请的技术方案。以上相关技术作为可选方案与本申请实施例的技术方案可以进行任意结合,其均属于本申请实施例的保护范围。本申请实施例包括以下内容中的至少部分内容。
图5示出了本申请实施例提供的一种无线通信方法,该方法可以包括步骤S110和步骤S120。
S110、终端设备根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
S120、终端设备在时刻之后,采用第一AI模型与网络设备进行通信。
相应的,网络设备可以根据终端设备上报的模型应用时间,或者,根据为终端设备配置的模型应用时间,确定第一AI模型开始应用的时刻;以及,网络设备在时刻之后,采用第一AI模型对应的第二AI模型与所述终端设备进行通信。
需要说明的是,终端设备和网络设备采用相对应的双端AI模型进行通信。其中,第一AI模型为部署于终端设备侧的神经网络模型,第二AI模型为部署于网络设备侧的神经网络模型。
示例性的,第一AI模型可以用于以下任一过程:
下行CSI反馈;或者,
下行波束反馈;或者,
下行数据的信道解码;或者,
下行信号的解调;或者,
下行信道估计;或者,
下行射频信号处理;或者,
上行数据的信道编码;或者,
上行数据的调制;或者,
上行导频生成;或者,
上行射频信号处理。
其中,第一AI模型用于下行CSI反馈,可以是终端设备将得到的信道信息(例如特征向量,波束信息,时延信息等)作为第一AI模型的输入,经第一AI模型推理出相应的CSI量化比特。
第一AI模型用于下行数据的信道解码,可以是终端设备将接收到的编码后的数据作为第一AI模型的输入,输出解码后的信息比特。
第一AI模型用于下行信号的解调,可以是终端设备将接收到的下行调制信号作为第一AI模型的输入,输出解调后的信息比特。
第一AI模型用于下行信道估计,可以是终端设备将接收到的下行信道和导频信号作为第一AI模型的输入,输出信道估计结果。
第一AI模型用于下行射频信号处理,可以是终端设备将接收到的下行射频信号作为第一AI模型的输入,输出下行射频信号处理结果。
第一AI模型用于上行数据的信道编码,可以是终端设备将待发送的信息比特作为第一AI模型的输入,输出编码后的数据。
第一AI模型用于上行数据的调制,可以是终端设备将待发送的上行数据的信息比特作为第一AI模型的输入,输出上行调制信号。
第一AI模型用于上行导频生成,可以是终端设备将待发送的信号作为第一AI模型的输入,输出导频信号。
第一AI模型用于上行射频信号处理,可以是终端设备将待发送的信号作为第一AI模型的输入,输出上行射频信号。
可以理解的,第一AI模型可以通过数据集的构建、训练、验证和测试等过程训练得到的。
可以理解的,第一AI模型可以提前通过离线训练和/或在线训练的方式训练得到的。
示例性的,第二AI模型可以用于以下任一过程:
下行CSI反馈信息的解码;或者,
下行波束反馈信息的解码;或者,
上行数据的信道解码;或者,
上行信号的解调;或者,
上行信道估计;或者,
上行射频信号处理;或者,
下行数据的信道编码;或者,
下行数据的调制;或者,
下行导频生成;或者,
下行射频信号处理。
需要说明的是,第一AI模型和第二AI模型对应,可以是指第一AI模型和第二AI模型实现操作是一一对应的。示例性的,第一AI模型用于下行CSI反馈,则第二AI模型用于下行CSI反馈信息的解码;或者,第一AI模型用于下行波束信息的生成,则第二AI模型可以用于下行波束反馈信息的解码;第一AI模型用于上行数据的信道编码,则第二AI模型可以用于上行数据的信道解码等。
可以理解的,第二AI模型可以通过数据集的构建、训练、验证和测试等过程训练得到的。
可以理解的,第二AI模型可以提前通过离线训练和/或在线训练的方式训练得到的。
在一些实施例中,终端设备侧部署的第一AI模型可以是网络设备为终端设备配置的。需要说明的是,网络设备可以在终端设备初始接入时为其配置第一AI模型;或者,网络设备可以在终端设备监测到原AI模型性能较差时,为其配置第一AI模型,以指示终端设备进行模型更新。
在一些实施例中,网络设备可以向终端设备发送第一信息,相应的,终端设备接收网络设备发送的第一信息,第一信息用于配置第一AI模型。换句话说,第一信息为用于配置第一AI模型的下行信息。
在一些实施例中,第一信息可以通过广播信令发送给终端设备。示例性的,第一信息可以通过物理广播信道(Physical Broadcast Channel,PBCH),系统信息(System Information Bits,SIB),组公用下行控制信息(Group Common DCI)等承载,本申请实施例对此不做限制。需要说明的是,Group Common DCI是通过公共搜索空间(Common Search Space,CSS)发给一组终端设备的DCI。其中,Group Common DCI采用公共的无线网临时标识(Radio Network Temporary Indentifier,RNTI)进行加扰。在一些实施例中,第一信息可以是高层信息,通过高层信令发送。示例性的,第一信息可以通过无线资源控制(Radio Resource Control,RRC)信令发送,也可以通过媒体接入控制(Media Access Control,MAC)信令发送,如MAC控制单元(MAC Control Element,MAC CE)。在一些实施例中,第一信息也可以是物理层信息,通过物理层信令发送。示例性的,第一信息可以通过DCI承载。
在一种实施方式中,所述第一信息可以是网络设备发送的模型恢复信息,所述模型恢复信息中包含第一AI模型的配置信息。网络设备在接收到终端设备发送的模型失败上报信息后,可以发送相应的模型恢复信息,用于终端侧模型的更新。
在一些实施例中,第一信息可以包括以下中的一项或多项:
第一AI模型的标识信息;
第一AI模型的模型功能(functionality);
第一AI模型的模型结构;以及,
第一AI模型的模型参数。
在一些实施例中,第一信息中可以携带第一AI模型的标识信息。应理解,终端设备可以预配置或者预定义一个或多个AI模型。终端设备可以根据第一信息中携带的第一AI模型的标识信息,从一个或多个AI模型中确定即将应用/更新的第一AI模型。其中,每个AI模型都可以有与之一一对应的一个标识信息,所述对应关系可以由网络设备与终端设备预先约定好,或者由网络设备配置给终端设备。
在一些实施例中,第一信息中可以指示第一AI模型的模型功能。其中,模型功能即通过AI/ML模型所实现的功能,不同的模型具有不同的模型功能,例如:PMI上报,RI/PMI/CQI上报,波束信息上报,RSRP上报,CSI压缩,信道编码解码,调制解调,CSI预测,CSI预测及压缩等。终端设备接收到第一信息后,可以根据AI模型所要实现的模型功能,确定所使用的第一AI模型。
需要说明的是,第一信息中携带的第一AI模型的模型结构可以是第一AI模型中的至少部分模型结构,或者说,第一信息可以携带全部或部分第一AI模型的模型结构。类似的,第一信息中携带的第一AI模型的模型参数可以是第一AI模型中的至少部分模型参数,或者说,第一信息可以携带全部或部分第一AI模型的模型参数。可以理解的,网络设备不一定要配置整个AI模型,可以只配置或更新第一AI模型中的部分结构或者部分模型参数。
在另一些实施例中,终端设备侧部署的第一AI模型也可以是终端设备从预配置或者预定义的多个AI模型中自行选择并使用的模型。需要说明的是,终端设备可以在初始接入时,向网络设备上报其使用的第一AI模型;另外,终端设备也可以在监测到原AI模型性能较差时,进行模型更新(如通过在线训练),使用第一AI模型进行通信,同时向网络设备指示第一AI模型,以使网络设备对网络设备侧当前部署的对应的模型进行更新。
其中,终端设备可以向网络设备发送第二信息,相应的,网络设备接收终端设备发送的第二信息,第二信息用于向网络设备指示终端设备使用的第一AI模型。此时,第二信息可以通过物理层信 道发送给网络设备,典型的,通过PUSCH或者PUCCH上报给网络设备。如果所述第二信息是物理层信息,则可以通过PUSCH或者PUCCH上报给网络设备,例如作为CSI上报的一部分;如果所述第二信息是高层信息,则可以通过PUSCH上报给网络设备,例如MAC信令。
网络设备收到第二信息后,可以发送第二信息对应的响应信息,以使终端设备知道网络设备正确接收到了所述第二信息。例如,所述第二信息通过PRACH承载,则所述第二信息的响应信息可以是对应的RAR(随机接入响应);所述第二信息通过PUSCH承载,则所述第二信息可以是所述PUSCH的HARQ-ACK信息。
在一些实施例中,第二信息可以包括以下中的一项或多项:
第一AI模型的标识信息;
第一AI模型的模型功能(functionality);
第一AI模型的模型结构;以及,
第一AI模型的模型参数。
在一些实施例中,第二信息中可以携带第一AI模型的标识信息。应理解,网络设备可以根据第二信息中携带的第一AI模型的标识信息,从一个或多个AI模型中确定终端设备即将应用/更新的第一AI模型,从而确定与第一AI模型对应的第二AI模型。其中,每个AI模型都可以有与之一一对应的一个标识信息,所述对应关系可以由网络设备与终端设备预先约定好,或者由网络设备配置给终端设备。
在一些实施例中,第二信息中可以指示第一AI模型的模型功能。其中,模型的功能即通过AI/ML模型所实现的功能,不同的模型具有不同的模型功能,例如:PMI上报,RI/PMI/CQI上报,波束信息上报,RSRP上报,CSI压缩,信道编码解码,调制解调,CSI预测,CSI预测及压缩等。网络设备接收到第二信息后,可以根据AI模型所要实现的模型功能,确定终端设备即将使用的/更新的第一AI模型,从而确定与第一AI模型对应的第二模型。
需要说明的是,第二信息中携带的第一AI模型的模型结构可以是第一AI模型中的至少部分模型结构,或者说,第二信息可以携带全部或部分第一AI模型的模型结构。类似的,第二信息中携带的第一AI模型的模型参数可以是第一AI模型中的至少部分模型参数,或者说,第二信息可以携带全部或部分第一AI模型的模型参数。可以理解的,终端设备不一定要指示整个第一AI模型,可以只配置或更新第一AI模型中的部分结构或者部分模型参数。
可以理解的,在终端设备接收到用于配置第一AI模型的第一信息,或者终端设备向网络设备发送指示第一AI模型的第二信息后,终端设备和网络设备均需要确定第一AI模型开始应用的时刻,从而在该时刻开始启用该第一AI模型与网络设备进行通信,保证终端设备侧和网络设备侧对应的模型的应用时间同步,避免两侧模型不匹配的问题。
在一种可能的实现方式中,终端设备可以根据其上报给网络设备的模型应用时间,确定网络设备配置的/终端设备指示的第一AI模型开始应用的时刻;相应的,网络设备可以根据终端设备上报的模型应用时间,确定网络设备为终端设备配置的/终端设备指示的第一AI模型开始应用的时刻。在另一种可能的实现方式中,终端设备可以根据网络设备配置的模型应用时间,确定网络设备配置的/终端设备指示的第一AI模型开始应用的时刻;相应的,网络设备可以根据其为终端设备配置的模型应用时间,确定网络设备配置的/终端设备指示的第一AI模型开始应用的时刻。
下面通过方式#A和方式#B说明终端设备确定第一AI模型开始应用的时刻的详细流程。
方式#A,终端设备根据上报给网络设备的模型应用时间,确定网络设备配置的/终端设备指示的第一AI模型开始应用的时刻;相应的,网络设备可以根据终端设备上报的模型应用时间,确定其为终端设备配置的/终端设备指示的第一AI模型开始应用的时刻。
需要说明的是,在确定第一AI模型开始应用的时刻之前,终端设备可以向网络设备上报其支持的模型应用时间。
示例性的,终端设备通过UE能力信息(例如UECapabilityInformation),向网络设备上报该模型应用时间。
在一些实施例中,终端设备可以在接收网络设备发送的用于配置第一AI模型的第一信息之前,向网络设备上报模型应用时间。相应的,网络设备可以在发送第一信息之前,可以接收终端设备上报的模型应用时间。
在一些实施例中,终端设备可以在发送用于指示第一AI模型的第二信息之前,向网络设备上报模型应用时间;相应的,网络设备可以在接收第二信息之前,接收终端设备上报的模型应用时间。或者,终端设备可以在发送用于指示第一AI模型的第二信息中,携带模型应用时间,将使用的第一 AI模型和模型应用时间一起上报给网络设备。这样,网络设备接收到第二信息后也能同时确定模型的应用时间,能够更快的应用模型。
在一些实施例中,对于网络设备通过第一信息为终端设备配置第一AI模型的场景,模型应用时间可以是以下中的任意一项:
终端设备接收到第一信息之后到应用第一AI模型的时刻之间的时间间隔;
终端设备发送第一信息的HARQ-ACK信息之后到应用第一AI模型的时刻之间的时间间隔;
终端设备接收到第一信息之后到终端设备发送基于第一AI模型得到的反馈信息的时刻之间的时间间隔;
终端设备发送第一信息的HARQ-ACK信息之后到终端设备发送基于第一AI模型得到的反馈信息的时刻之间的时间间隔。
在一种可能的实现方式中,对于模型的输出信息不需要上报网络设备的第一AI模型,例如用于下行数据的信道解码、下行信道的解调、下行信道估计、下行射频信号处理、上行数据的信道编码、上行数据的调制、上行导频生成、以及上行射频信号处理等操作的第一AI模型,其模型应用时间可以是终端设备接收到第一信息之后到应用第一AI模型的时刻之间的时间间隔,或者终端设备发送第一信息的HARQ-ACK信息之后到应用第一AI模型的时刻之间的时间间隔。
在一种可能的实现方式中,对于模型的输出信息需要上报给网络设备的第一AI模型,例如,用于下行CSI反馈或者下行波束信息反馈的第一AI模型,其模型应用时间可以是终端设备接收到第一信息之后到终端设备发送基于第一AI模型得到的反馈信息的时刻之间的时间间隔,或者,终端设备发送第一信息的HARQ-ACK信息之后到终端设备发送基于第一AI模型得到的反馈信息的时刻之间的时间间隔。
在一些实施例中,对于终端设备通过第二信息向网络设备指示第一AI模型的场景,模型应用时间可以为以下中的任意一项:
终端设备发送第二信息之后到应用第一AI模型的时刻之间的时间间隔;
终端设备发送第二信息之后到所述终端设备发送基于第一AI模型得到的反馈信息的时刻之间的时间间隔;
终端设备接收第二信息的响应信息之后到应用第一AI模型的时刻之间的时间间隔;
终端设备接收第二信息的响应信息之后到终端设备发送基于第一AI模型得到的反馈信息的时刻之间的时间间隔。
其中,对于模型的输出信息不需要上报网络设备的第一AI模型,例如用于下行数据的信道解码、下行信道的解调、下行信道估计、下行射频信号处理、上行数据的信道编码、上行数据的调制、上行导频生成、以及上行射频信号处理等操作的第一AI模型,模型应用时间可以是终端设备发送第二信息之后到应用第一AI模型的时刻之间的时间间隔,或者,终端设备接收第二信息的响应信息之后到应用第一AI模型的时刻之间的时间间隔。
另外,对于模型的输出信息需要上报给网络设备的第一AI模型,例如,用于下行CSI反馈或者下行波束信息反馈的第一AI模型,模型应用时间可以是终端设备发送第二信息之后到所述终端设备发送基于第一AI模型得到的反馈信息的时刻之间的时间间隔,或者,终端设备接收第二信息的响应信息之后到终端设备发送基于第一AI模型得到的反馈信息的时刻之间的时间间隔。
需要说明的是,上述两种场景中模型应用时间涉及的终端设备发送基于第一AI模型得到的反馈信息的时刻,可以是指终端设备第一次发送基于第一AI模型得到的反馈信息的时刻。
示例性的,对于用于下行CSI反馈的第一AI模型,模型应用时间可以是:终端设备接收到第一信息之后到终端设备(第一次)发送基于第一AI模型得到的CSI反馈信息的时刻之间的时间间隔,或者终端设备发送第一信息的HARQ-ACK信息之后到终端设备(第一次)发送基于第一AI模型得到的CSI反馈信息的时刻之间的时间间隔。模型应用时间也可以是:终端设备发送第二信息之后到终端设备(第一次)发送基于第一AI模型得到的CSI反馈信息的时刻之间的时间间隔,或者终端设备接收到第二信息的响应信息之后到终端设备(第一次)发送基于第一AI模型得到的CSI反馈信息的时刻之间的时间间隔。
示例性的,对于用于下行波束信息反馈的第一AI模型,模型应用时间可以是:终端设备接收到第一信息之后到终端设备(第一次)发送基于第一AI模型得到的下行波束信息的时刻之间的时间间隔,或者终端设备发送第一信息的HARQ-ACK信息之后到终端设备(第一次)发送基于第一AI模型得到的下行波束信息的时刻之间的时间间隔。另外,模型应用时间也可以是终端设备发送第二信息之后到终端设备(第一次)发送基于第一AI模型得到的下行波束信息的时刻之间的时间间隔,或 者终端设备接收到第二信息的响应信息之后到终端设备(第一次)发送基于第一AI模型得到的下行波束信息的时刻之间的时间间隔。
还需要说明的是,上述时间间隔可以以时隙或OFDM符号为单位。此外,上述时间间隔还可以以微时隙、OFDM符号集合为单位,或者以绝对时间(例如N毫秒,N秒,N>0)为单位,本申请实施例对此不做限制。
在一些实施例中,终端设备上报给网络设备的模型应用时间可以包括以下中的一项或多项:
终端设备支持的一个或多个AI模型分别对应的模型应用时间;其中,终端设备支持的一个或多个AI模型包括第一AI模型;
终端设备支持的一种或多种模型更新方式分别对应的模型应用时间;其中,终端设备支持的一种或多种模型更新方式包括第一AI模型的更新方式;
终端设备支持的一个或多个AI模型功能分别对应的模型应用时间;其中,终端设备支持的一个或多个模型功能包括第一AI模型的功能;
终端设备支持的一个或多个模型类型分别对应的模型应用时间;其中,终端设备支持的一个或多个模型类型包括第一AI模型的类型。
在一种可能的实现方式中,终端设备可以支持一个或多个AI模型,该一个或多个AI模型可以是预配置或者预定义的。可以理解的,不同的AI模型的模型结构和模型参数不同,因此不同的AI模型的模型应用时间也就不同。终端设备可以为其支持一个或多个AI模型分别上报各自的模型应用时间。可以理解的,终端设备可以为不同的模型上报不同的模型应用时间。
在一种可能的实现方式中,终端设备可以支持一种或多个模型更新方式。示例性的,模型更新方式包括:更新模型参数和/或更新模型结构。
其中,终端设备可以为其支持的一种或多种AI模型更新方式上报各自对应的模型应用时间。可以理解的,终端设备可以为不同的模型更新方式上报不同的模型应用时间。示例性的,只更新模型参数需要的应用时间为t1,更新模型结构和模型参数需要的应用时间为t2,且t1<t2。
需要说明的是,同时更新模型结构和模型参数可以认为是更新了整个AI模型。
在一种可能的实现方式中,终端设备可以支持一个或多个模型类型。示例性的,终端设备支持的模型类型可以包括高精度模型和低精度模型;或者,已知模型和未知模型。
其中,高精度模型的复杂度高于低精度模型,所以需要的模型应用时间也较长。这里,相对于低精度模型,高精度模型可以是输出比特数更多的模型,或者是神经网络层数更多的模型,或者是神经网络结构更复杂的模型。
另外,已知模型可以为终端设备已经部署过的模型,未知模型为终端设备尚未部署过的模型。其中,已知模型需要的应用时间短于未知模型需要的应用时间。
需要说明的是,终端设备上报的模型应用时间可以根据终端设备的处理能力或硬件配置,以及不同AI模型/不同AI模型类型推理需要的时间来确定。示例性的,终端设备采用CPU用于AI模型的模型应用时间,可以长于采用GPU用于AI模型的模型应用时间。
可以理解的,终端设备和网络设备根据上报的模型应用时间,确定第一AI模型(第一AI模型可以是网络设备通过第一信息配置的,也可以是终端设备通过第二信息指示的)开始应用的时刻,可以通过以下中的一项或多项实现:
根据网络设备配置的/终端设备指示的第一AI模型,以及终端设备上报的一个或多个AI模型分别对应的模型应用时间,确定第一AI模型开始应用的时刻;
根据网络设备配置的/终端设备指示的第一AI模型的功能,以及终端设备上报的一个或多个模型功能分别对应的模型应用时间,确定第一AI模型开始应用的时刻;
根据网络设备配置的/终端设备指示的第一AI模型的模型更新方式,以及终端设备上报的一种或多种模型更新方式分别对应的模型应用时间,确定该第一AI模型开始应用的时刻;
根据网络设备配置的/终端设备指示的第一AI模型的模型类型,以及终端设备上报的一个或多个模型类型分别对应的模型应用时间,确定第一AI模型开始应用的时刻。
示例性的,终端设备可以上报AI模型1对应的模型应用时间为T1,AI模型2对应的模型应用时间为T2。若网络设备通过第一信息为终端设备配置模型1,则终端设备和网络设备可以根据AI模型1对应的模型应用时间T1,确定第一信息配置的AI模型1开始应用的时刻。或者,若终端设备通过第二信息指示模型2,则终端设备和网络设备可以根据AI模型2对应的模型应用时间T2,确定第二信息指示的AI模型2的开始应用的时刻。
示例性的,终端设备可以上报模型功能1(例如CSI压缩功能)对应的模型应用时间为T3,模 型功能2(例如信道编码解码功能)对应的模型应用时间为T4。若网络设备通过第一信息为终端设备配置模型功能1,则终端设备和网络设备可以根据模型功能1对应的模型应用时间T3,确定第一信息配置的AI模型开始应用的时刻。或者,若终端设备通过第二信息指示模型功能2,则终端设备和网络设备可以根据模型功能2对应的模型应用时间T4,确定第二信息指示的AI模型开始应用的时刻。
示例性的,终端设备可以上报高精度模型对应的模型应用时间T5,低精度模型对应的模型应用时间T6。若网络设备通过第一信息为终端设备配置的AI模型为低精度模型,则终端设备和网络设备可以基于低精度模型的模型应用时间T6,确定第一信息配置的AI模型开始应用的时刻。若终端设备通过第二信息指示AI模型的类型为高精度模型,则终端设备和网络设备可以基于高精度模型的应用时间T5,确定第二信息指示的AI模型开始应用的时刻。
示例性的,终端设备上报已知模型对应的模型应用时间T7,未知模型对应的模型应用时间T8。若网络设备通过第一信息为终端设备配置的AI模型1为未知模型(例如AI模型1为终端设备预配置或者预定义的候选模型之外的模型),则终端设备和网络设备可以基于未知模型的模型应用时间T8,确定第一信息配置的AI模型1开始应用的时刻。
需要说明的是,在终端设备上报的第一AI模型对应的模型应用时间,第一AI模型的模型功能对应的模型应用时间,第一AI模型的模型类型对应的模型应用时间,以及第一AI模型的模型更新方式对应的模型应用时间中的两个或两个以上不同的情况下,终端设备可以根据最长(或最短)的模型应用时间,来确定第一AI模型的开始应用时刻。
方式#B,终端设备根据网络设备配置的模型应用时间,确定网络设备配置的/终端设备指示的第一AI模型开始应用的时刻;相应的,网络设备可以根据其为终端设备配置的模型应用时间,确定网络设备配置的/终端设备指示的第一AI模型开始应用的时刻。
需要说明的是,模型应用时间的相关解释说明可以参考上述实施例中的描述,为了简洁,此处不再赘述。
在一些实施例中,网络设备可以为终端设备直接配置模型应用时间。其中,网络设备可以向终端设备发送第三信息,相应的,终端设备可以接收网络设备发送的第三信息,其中,第三信息用于配置模型应用时间。也就是说,网络设备通过第三信息为终端设备配置模型应用时间。
在一些实施例中,第三信息可以通过广播信令发送给终端设备。示例性的,第三信息可以通过PBCH,SIB,Group Common DCI等承载,本申请实施例对此不做限制。在一些实施例中,第三信息可以是高层信息,通过高层信令发送。示例性的,第三信息可以通过RRC信令,或者MAC CE信令发送。在一些实施例中,第三信息也可以是物理层信息,通过物理层信令发送。示例性的,第三信息可以通过DCI信令承载。
针对网络设备通过第一信息为终端设备配置第一AI模型的场景,第三信息可以包括第一信息中所配置的第一AI模型的模型应用时间。
需要说明的是,所述第三信息可以与第一信息一起指示给终端设备,这样终端设备收到模型的配置信息后也能同时确定模型的应用时间,能够更快的应用模型。
在一种可能的实现方式中,网络设备可以根据预设规则或者预设的映射关系,为终端设备配置第一AI模型的模型应用时间。
示例性的,网络设备可以根据为终端设备配置的第一AI模型的类型、第一AI模型的更新方式等因素,为终端设备配置第一AI模型的应用时间。
示例性的,网络设备可以预先存储每种AI模型对应的模型应用时间(AI模型和模型应用时间之间的映射关系)。当网络设备为终端设备配置第一AI模型时,可以基于该映射关系,为终端设备配置第一AI模型对应的模型应用时间。
在另一种可能的实现方式中,网络设备可以接收终端设备上报的模型应用时间,根据终端设备上报的模型应用时间来为终端设备配置实际的模型应用时间。
需要说明的是,网络设备可以根据覆盖范围内多个终端设备分别上报的模型应用时间,来确定为终端设备配置的模型应用时间。
在一些实施例中,网络设备可以采用多个终端设备分别上报的模型应用时间中的最大值,作为覆盖范围内各个终端设备实际使用的模型应用时间,并将该模型应用时间配置给这些终端设备。也就是说,网络设备为终端设备配置的模型应用时间的时长,大于或等于终端设备上报的模型应用时间的时长。
由此可见,网络设备可以为同时其覆盖范围内的多个不同的终端设备配置模型应用时间,这样, 网络设备与不同终端设备能够同步进行模型更新,从而保证接收端的模型输出是可用的,提高通信性能。
针对终端设备通过第二信息指示第一AI模型的场景,在一种可能的实现方式中,网络设备可以在接收到第二信息之后,确定第二信息指示的第一AI模型的模型应用时间,并通过第三信息向终端设备指示该模型应用时间。也就是说,网络设备可以在接收到第二信息后,向终端设备发送第三信息,以为终端设备配置第二信息中所指示的第一AI模型的模型应用时间。
需要说明的是,网络设备可以根据预设规则或者预设的映射关系,为终端设备配置第一AI模型的模型应用时间。或者,网络设备可以根据终端设备上报的模型应用时间来为终端设备配置实际的模型应用时间。例如,网络设备可以采用多个终端设备分别上报的模型应用时间中的最大值,作为覆盖范围内各个终端设备实际使用的模型应用时间,并将该模型应用时间配置给这些终端设备。
针对终端设备通过第二信息指示第一AI模型的场景,在另一种可能的实现方式中,网络设备还可以在第二信息之前,向终端设备发送第三信息。该实现方式中第三信息可以包括以下中的一项或多项:
一个或多个AI模型分别对应的模型应用时间;其中,所述一个或多个AI模型包括所述第一AI模型;
一个或多个模型功能分别对应的模型应用时间;其中,所述一个或多个模型功能包括所述第一AI模型的功能;
一种或多种模型更新方式分别对应的模型应用时间;其中,所述一种或多种模型更新方式包括所述第一AI模型的更新方式;
一个或多个模型类型分别对应的模型应用时间;其中,所述一个或多个模型类型包括所述第一AI模型的类型。
可以理解的,终端设备接收到第三信息后,可以结合第二信息指示的第一AI模型,以及第三信息配置的每种AI模型、每种模型功能、每种模型更新方式、或者每个模型类型对应的模型应用时间,确定第二信息所指示的第一AI模型实际的模型应用时间。
需要说明的是,第三信息中配置的模型应用时间可以是网络设备根据终端设备上报的模型应用时间确定的。可以理解的,网络设备可以根据覆盖范围内多个终端设备分别上报的模型应用时间,来确定为终端设备配置的模型应用时间。在一些实施例中,网络设备可以采用多个终端设备分别上报的模型应用时间中的最大值,作为覆盖范围内各个终端设备实际使用的模型应用时间,并将该模型应用时间配置给这些终端设备。
示例性的,第三信息可以配置AI模型1(或者模型功能1/模型更新方式1/模型类型1)和AI模型2(或者模型功能2/模型更新方式2/模型类型2)分别对应的模型应用时间。其中,第三信息中配置的AI模型1(或者模型功能1/模型更新方式1/模型类型1)的模型应用时间,可以是多个终端设备上报的AI模型1(或者模型功能1/模型更新方式1/模型类型1)对应的模型应用时间中的最大值,第三信息中配置的AI模型2(或者模型功能2/模型更新方式2/模型类型2)的模型应用时间,可以是多个终端设备上报的AI模型2(或者模型功能2/模型更新方式2/模型类型2)对应的模型应用时间中的最大值。
这样,网络设备可以为同时其覆盖范围内的多个不同的终端设备配置模型应用时间。如此,网络设备与不同终端设备能够同步进行模型更新,从而保证接收端的模型输出是可用的,提高通信性能。
本申请实施例中,终端设备基于上述方式确定了第一AI模型开始应用的时刻之后,进一步地,终端设备可以在该时刻后,采用第一AI模型进行与网络设备之间的信息传输。相应的,网络设备可以在该时刻之后,采用与第一AI模型对应的第二模型进行与终端设备之间的信息传输。
在一些实施例中,在终端设备应用第一AI模型之前,终端设备可以采用第三AI模型进行与网络设备之间的信息传输,或者,终端设备不基于AI模型进行与网络设备之间的信息传输。相应的,在终端设备应用第一AI模型之前,网络设备可以采用与第三AI模型对应的第四AI模型进行与终端设备之间的信息传输,或者网络设备不基于AI模型进行与终端设备之间的信息传输,例如,网络设备基于传统的非AI方式来接收终端设备传输的信息。
需要说明的是,第三AI模型可以是预定义的AI模型,或者历史AI模型。其中,预定义的AI模型即终端设备和网络设备预先约定好的一个AI模型。历史AI模型可以是网络设备在第一AI模型开始应用的时刻之前,网络设备为终端设备配置的AI模型或者终端设备指示的AI模型,这里的历史AI模型可以是与第一AI模型的模型结构和模型参数完全不同的AI模型,也可以是模型结构相同 但模型参数不同的AI模型,也可以是模型功能不同的AI模型,本申请实施例对此不做限制。另外,第四AI模型可以是与第三AI模型对应的网络设备侧部署的AI模型。
示例性的,如果网络设备通过第一信息配置的第一AI模型/终端设备通过第二信息指示的第一AI模型为新的模型(即第一AI模型的模型ID与第三AI模型的模型ID不同),则在第一AI模型开始应用的时刻之前,终端设备可以使用之前使用的第三AI模型进行与网络设备之间的信息传输。
示例性的,如果网络设备通过第一信息配置了一个新的模型的功能/终端设备通过第二信息指示了一个新的模型的功能(即第一AI模型的functionality与第三AI模型的functionality不同),则在第一AI模型开始应用的时刻之前,终端设备可以使用之前的functionality对应的第三模型进行与网络设备之间的信息传输。这里的功能也可以是功能ID。
示例性的,如果网络设备通过第一信息更新了模型结构和模型参数/终端设备通过第二信息更新了模型结构和模型参数,在终端设备确定的第一AI模型开始应用的时刻之前,终端设备可以使用之前采用的模型结构和模型参数进行与网络设备之间的信息传输。
示例性的,如果网络设备通过第一信息只更新了模型参数/终端设备通过第二信息只更新了模型参数,在终端设备确定的第一AI模型开始应用的时刻之前,终端设备可以使用之前采用的模型参数进行与网络设备之间的信息传输。
还需要说明的是,终端设备不基于AI模型进行与网络设备之间的信息传输,也就是指终端设备采用传统的非AI的方式进行与网络设备之间的信息传输。
综上所述,本申请实施例中,在终端设备侧和网络设备侧采用相对应的双端AI模型时,如果终端设备侧的AI模型发生更新或重配置,终端设备可以根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定网络设备配置的AI模型开始应用的时刻。这样,可以确保两侧的模型应用时刻是相同的,从而保证AI模型的输出是可用的,提高了通信系统的性能。
针对网络设备通过第一信息为终端设备配置第一AI模型的场景,终端设备和网络设备确定的第一AI模型开始应用的时刻与第一信息有关。下面通过方式#1~方式#3说明,网络设备通过第一信息为终端设备配置第一AI模型的场景中,第一AI模型开始应用的时刻。
方式#1,第一AI模型开始应用的时刻可以为:
承载第一信息的HARQ-ACK信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一OFDM符号;或者,
承载第一信息的下行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一OFDM符号;其中,k为模型应用时间。
需要说明的是,方式#1中模型应用时间所表征的时间间隔可以以OFDM符号为单位。模型应用时间可以是k个OFDM符号,或者说,模型应用时间表征的时间间隔为k个OFDM符号。其中,k为大于或等于1的整数。
可以理解的,如上文所述,k可以是终端设备上报给网络设备的,也可以是网络设备配置的。
在一种可能的实现方式中,终端设备可以将承载第一信息的下行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一OFDM符号作为第一AI模型开始应用的时刻。或者说,终端设备可以将承载第一信息的下行信道的最后一个OFDM符号之后间隔k个OFDM符号的第一个时隙或者第一个OFDM符号作为第一AI模型开始应用的时刻。
需要说明的是,这里的下行信道可以是PBCH,PDSCH,PDCCH等。示例性的,第一信息为广播信息时,下行信道可以是PBCH;第一信息为高层信息时,下行信道可以是PDSCH;第一信息为物理层信息时,下行信道可以是PDCCH。
还需要说明的是,本申请实施例中的k个OFDM符号也可以是k-1或k+1个OFDM符号,本申请实施例对此不做限制。
示例性的,参考图6A所示,假设承载第一信息的下行信道的最后一个OFDM符号为符号m,则终端设备可以确定第一AI模型开始应用的时刻为符号m+k所在符号之后的第一个时隙。
示例性的,参考图6B所示,假设承载第一信息的下行信道的最后一个OFDM符号为符号m,则终端设备可以确定第一AI模型开始应用的时刻为符号m+k所在符号之后的第一个OFDM符号,即图6B所示的符号m+k+1。
在另一种可能的实现方式中,终端设备可以将承载第一信息的HARQ-ACK信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一OFDM符号作为第一AI模型开始应用的时刻。或者说,终端设备可以将承载第一信息的HARQ-ACK信息的上行信道的最后一个OFDM符号之后间隔k个OFDM符号的第一个时隙或者第一个OFDM符号作为第一AI模型 开始应用的时刻。
可以理解的,终端设备接收到第一信息后,可以向网络设备发送第一信息的HARQ-ACK信息。示例性的,第一信息为高层信息或者物理层信息时,终端设备接收到第一信息后,可以向网络设备反馈第一信息的HARQ-ACK信息。
需要说明的是,承载第一信息的HARQ-ACK信息的上行信道可以是PUCCH。
还需要说明的是,本申请实施例中的k个OFDM符号也可以是k-1或k+1个OFDM符号,本申请实施例对此不做限制。
示例性的,参考图7A所示,假设承载第一信息的HARQ-ACK信息的上行信道的最后一个OFDM符号为符号m,则终端设备可以确定第一AI模型开始应用的时刻为符号m+k所在符号之后的第一个时隙。
示例性的,参考图7B所示,假设承载第一信息的HARQ-ACK信息的上行信道的最后一个OFDM符号为符号m,则终端设备可以确定第一AI模型开始应用的时刻为符号m+k所在符号之后的第一个OFDM符号,即图7B所示的符号m+k+1。
可以理解的,将k个OFDM符号之后的第一OFDM符号作为第一AI模型开始应用的时刻,相比于将k个OFDM符号之后的第一时隙作为第一AI模型开始应用的时刻,可以缩短第一AI模型应用的时间间隔,同时还可以提高模型应用时刻的准确度。
方式#2,第一AI模型开始应用的时刻可以为:
终端设备接收到第一信息的时隙开始k个时隙之后的第一个时隙,k为所述模型应用时间。
需要说明的是,方式#2中模型应用时间所表征的时间间隔可以以时隙为单位。模型应用时间可以是k个时隙,或者说,模型应用时间表征的时间间隔为k个时隙。其中,k为大于或等于1的整数。
可以理解的,如上文所述,k可以是终端设备上报给网络设备的,也可以是网络设备配置的。
在该方式#2中,终端设备可以将接收第一信息的时隙开始k个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。或者说,终端设备可以将接收第一信息的时隙开始间隔k个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。
需要说明的是,本申请实施例中的k个时隙也可以是k-1或k+1个时隙,本申请实施例对此不做限制。
示例性的,参考图8所示,假设终端设备在时隙n接收到第一信息,终端设备可以确定第一AI模型开始应用的时刻为时隙n+k之后的第一个时隙,即时隙n+k+1。
方式#3,第一AI模型开始应用的时刻可以为:
承载第一信息的HARQ-ACK信息的上行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙,其中,k为所述模型应用时间,N为1ms包含的时隙数量。
需要说明的是,方式#3中模型应用时间所表征的时间间隔可以以时隙为单位。模型应用时间可以是k个时隙,或者说,模型应用时间表征的时间间隔为k个时隙。其中,k为大于或等于1的整数。
可以理解的,如上文所述,k可以是终端设备上报给网络设备的,也可以是网络设备配置的。
在一种实施方式中,终端设备可以将携带第一信息的HARQ-ACK信息的上行信道所在时隙开始k个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。或者说,终端设备可以将携带第一信息的HARQ-ACK信息的上行信道所在时隙开始,间隔k个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。
需要说明的是,该实施方式中,第一信息为高层信息,典型地,第一信息通过RRC信令发送。
还需要说明的是,本申请实施例中的k个时隙也可以是k-1或k+1个时隙,本申请实施例对此不做限制。
示例性的,参考图9A所示,假设终端设备在时隙n发送携带第一信息的PDSCH的HARQ-ACK信息,则第一AI模型开始应用的时刻为时隙n+k之后的第一个时隙,即时隙n+k+1。
在一种实施方式中,终端设备可以将携带第一信息的HARQ-ACK信息的上行信道所在时隙开始(k+3N)个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。或者说,终端设备可以将携带第一信息的HARQ-ACK信息的上行信道所在时隙开始,间隔(k+3N)个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。
需要说明的是,该实施方式中,第一信息为高层信息,典型地,第一信息通过MAC层信令发送,例如3GPP TS38.321中的定义的信令。
可以理解的,3N个时隙(即3ms)为MAC层信令生效至少需要的时间,通过间隔k+3N个时隙从而保证MAC层信令有足够的生效时间。
还需要说明的是,本申请实施例中的k个时隙也可以是k-1或k+1个时隙,本申请实施例对此不做限制。
示例性的,参考图9B所示,假设终端设备在时隙n发送携带第一信息的PDSCH的HARQ-ACK信息,则第一AI模型开始应用的时刻为时隙n+k+3N之后的第一个时隙,即时隙n+k+3N+1。
在一种实施方式中,终端设备可以将携带第一信息的HARQ-ACK信息的上行信道所在时隙开始max(k,3N)个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。或者说,终端设备可以将携带第一信息的HARQ-ACK信息的上行信道所在时隙开始,间隔max(k,3N)个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。
需要说明的是,该实施方式中,第一信息为高层信息,典型地,第一信息通过MAC层信令,例如3GPP TS38.321中的定义的信令。
可以理解的,3N个时隙(即3ms)为MAC层信令生效至少需要的时间,通过取k和3N中的较大值从而保证MAC层信令有足够的生效时间。
还需要说明的是,本申请实施例中的k个时隙也可以是k-1或k+1个时隙,本申请实施例对此不做限制。
示例性的,参考图9C所示,假设终端设备在时隙n发送携带第一信息的PDSCH的HARQ-ACK信息且k<3N,则第一AI模型开始应用的时刻为时隙n+3N之后的第一个时隙,即时隙n+3N+1。
需要说明的是,终端设备和网络设备可以约定方式#1~方式#3中的一种方式,或者网络设备为终端设备指定方式#1~方式#3中的一种方式,来确定第一AI模型开始应用的时刻,从而确保两侧相对应的AI模型开始应用的时刻是相同的。
另外,针对终端设备通过第二信息向网络设备指示第一AI模型的场景,终端设备和网络设备确定的第一AI模型开始应用的时刻可以与第二信息相关。下面通过方式#1’~方式#3’说明,终端设备通过第二信息向网络设备指示第一AI模型的场景中,第一AI模型开始应用的时刻。
方式#1’,第一AI模型开始应用的时刻为:
承载第二信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;或者,
承载第二信息的上行信道的响应信息的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;其中,k为所述模型应用时间。
需要说明的是,方式#1’中模型应用时间所表征的时间间隔可以以OFDM符号为单位。模型应用时间可以是k个OFDM符号,或者说,模型应用时间表征的时间间隔为k个OFDM符号。其中,k为大于或等于1的整数。
在一种可能的实现方式中,终端设备可以将承载第二信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一OFDM符号作为第一AI模型开始应用的时刻。或者说,终端设备可以将承载第二信息的上行信道的最后一个OFDM符号之后间隔k个OFDM符号的第一个时隙或者第一个OFDM符号作为第一AI模型开始应用的时刻。
需要说明的是,这里的上行信道可以是PUSCH或者PUCCH。示例性的,若第二信息为高层信息,则上行信道可以是PUSCH;若第二信息为物理层信息,则上行信道可以是PUCCH。
还需要说明的是,本申请实施例中的k个OFDM符号也可以是k-1或k+1个OFDM符号,本申请实施例对此不做限制。
在另一种可能的实现方式中,终端设备可以将承载第二信息的响应信息的下行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一OFDM符号作为第一AI模型开始应用的时刻。或者说,终端设备可以将承载第二信息的响应信息的下行信道的最后一个OFDM符号之后间隔k个OFDM符号的第一个时隙或者第一个OFDM符号作为第一AI模型开始应用的时刻。
可以理解的,网络设备收到第二信息后,可以发送第二信息对应的响应信息,以使终端设备知道网络设备正确接收到了所述第二信息。例如,所述第二信息通过PRACH承载,则所述第二信息的响应信息可以是对应的RAR;所述第二信息通过PUSCH承载,则所述第二信息可以是所述PUSCH的HARQ-ACK信息
需要说明的是,上述的k个OFDM符号也可以是k-1或k+1个OFDM符号,本申请实施例对此不做限制。
方式#2’,第一AI模型开始应用的时刻为:
终端设备发送第二信息的时隙开始k个时隙之后的第一个时隙;第二信息为用于指示所述第一AI模型的上行信息,k为所述模型应用时间。
需要说明的是,方式#2’中模型应用时间所表征的时间间隔可以以时隙为单位。模型应用时间可以是k个时隙,或者说,模型应用时间表征的时间间隔为k个时隙。其中,k为大于或等于1的整数。
可以理解的,如上文所述,k可以是终端设备上报给网络设备的,也可以是网络设备配置的。
在该方式#2’中,终端设备可以将发送第二信息的时隙开始k个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。或者说,终端设备可以将发送第二信息的时隙开始间隔k个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。
需要说明的是,本申请实施例中的k个时隙也可以是k-1或k+1个时隙,本申请实施例对此不做限制。
方式#3’,第一AI模型开始应用的时刻为:
承载第二信息的响应信息的下行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙,其中,所述第二信息为用于指示所述第一AI模型的上行信息,k为所述模型应用时间,N为1ms包含的时隙数量。
需要说明的是,方式#3’中模型应用时间所表征的时间间隔可以以时隙为单位。模型应用时间可以是k个时隙,或者说,模型应用时间表征的时间间隔为k个时隙。其中,k为大于或等于1的整数。
可以理解的,如上文所述,k可以是终端设备上报给网络设备的,也可以是网络设备配置的。
在一种实施方式中,终端设备可以将携带第二信息的响应信息的下行信道所在时隙开始k个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。或者说,终端设备可以将携带第二信息的响应信息的下行信道所在时隙开始,间隔k个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。
需要说明的是,该实施方式中,第二信息为高层信息,典型地,第二信息通过RRC信令发送。
还需要说明的是,本申请实施例中的k个时隙也可以是k-1或k+1个时隙,本申请实施例对此不做限制。
在一种实施方式中,终端设备可以将携带第二信息的响应信息的下行信道所在时隙开始(k+3N)个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。或者说,终端设备可以将携带第二信息的响应信息的下行信道所在时隙开始,间隔(k+3N)个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。
需要说明的是,该实施方式中,第二信息为高层信息,典型地,第二信息通过MAC层信令发送,例如3GPP TS38.321中的定义的信令。
可以理解的,3N个时隙(即3ms)为MAC层信令生效至少需要的时间,通过间隔k+3N个时隙从而保证MAC层信令有足够的生效时间。
还需要说明的是,本申请实施例中的k个时隙也可以是k-1或k+1个时隙,本申请实施例对此不做限制。
在一种实施方式中,终端设备可以将携带第二信息的响应信息的下行信道所在时隙开始max(k,3N)个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。或者说,终端设备可以将携带第二信息的响应信息的下行信道所在时隙开始,间隔max(k,3N)个时隙之后的第一个时隙作为第一AI模型开始应用的时刻。
需要说明的是,该实施方式中,第二信息为高层信息,典型地,第二信息通过MAC层信令发送,例如3GPP TS38.321中的定义的信令。
可以理解的,3N个时隙(即3ms)为MAC层信令生效至少需要的时间,通过取k和3N中的较大值从而保证MAC层信令有足够的生效时间。
还需要说明的是,本申请实施例中的k个时隙也可以是k-1或k+1个时隙,本申请实施例对此不做限制。
需要说明的是,终端设备和网络设备可以约定方式#1’~方式#3’中的一种方式,或者网络设备为终端设备指定方式#1’~方式#3’中的一种方式,来确定第一AI模型开始应用的时刻,从而确保两侧相对应的AI模型开始应用的时刻是相同的。
在一些实施例中,所述第一AI模型开始应用的时刻和/或所述模型应用时间,可以根据以下任意一项确定:
承载第一信息的下行信道的子载波间隔;
承载承载第一信息的HARQ-ACK信息的上行信道的子载波间隔;
承载第二信息的上行信道的子载波间隔;
承载第二信息的响应信息的下行信道的子载波间隔;
应用第一AI模型的载波上所激活的BWP的子载波间隔。
需要说明的是,第一AI模型开始应用的时刻和模型应用时间是以时隙或者OFDM符号为单位的参数,需要根据子载波间隔确定实际对应的时间长度/时刻(或者称为绝对时间长度/时隙),不同子载波间隔对应的时间长度不同。
示例性的,参考表1所示所示的子载波间隔与时隙个数和时隙长度之间的关系。
表1
参考表1所示,子载波间隔为30kHz,对应每个时隙的长度为0.5ms。若模型应用时间为k个时隙,则k个时隙的长度为0.5k ms。以此类推,子载波间隔为60kHz,对应每个时隙的长度为0.25ms。若模型应用时间为k个时隙,则k个时隙的长度为0.25k ms。
可以理解的,第一AI模型开始应用的时刻和/或模型应用时间,可以根据与第一AI模型相关的子载波间隔确定。
在一些实施例中,针对网络设备通过第一信息为终端设备配置第一AI模型的场景,第一AI模型开始应用的时刻和/或模型应用时间,可以根据以下中的任意一项确定:
承载第一信息的下行信道的子载波间隔;
承载第一信息的HARQ-ACK信息的上行信道的子载波间隔;
应用所述第一AI模型的载波上所激活的BWP的子载波间隔。
在一种可能的实现方式中,第一AI模型开始应用的时刻和/或模型应用时间,可以根据承载第一信息的下行信道的子载波间隔确定。
示例性的,若第一信息通过PBCH承载,则第一AI模型开始应用的时刻和/或模型应用时间根据PBCH的子载波间隔确定。若第一信息通过Group Common DCI承载,则第一AI模型开始应用的时刻和/或模型应用时间根据承载所述DCI的PDCCH的子载波间隔确定。
示例性的,结合图8所示的时隙结构示意图,模型开始应用的时刻为时隙n+k+1,终端设备可以根据承载第一信息的子载波间隔,得到时隙n+k+1与时隙n之间间隔的时间长度,从而确定模型应用的时间。
在另一种可能的实现方式中,第一AI模型开始应用的时刻和/或模型应用时间,可以根据承载第一信息的HARQ-ACK信息的上行信道的子载波间隔确定。
示例性的,若第一信息通过高层信令承载,则第一AI模型开始应用的时刻和/或模型应用时间可以根据携带该高层信令的PDSCH的HARQ-ACK信息所在的上行信道的子载波间隔确定。
示例性的,结合图9B所示的时隙结构示意图,如果模型开始应用的时刻为时隙n+k+3N+1,终端设备可以根据所确定的子载波间隔,得到时隙n+k+3N+1与时隙n之间间隔的时间长度,从而确定模型应用的时间。
在又一种可能的实现方式中,第一AI模型开始应用的时刻和/或模型应用时间,可以根据应用第一AI模型的载波上所激活的BWP的子载波间隔。
在一些实施例中,当应用第一AI模型的载波的数量为多个时,第一AI模型开始应用的时刻和/或模型应用时间可以根据第一子载波间隔确定。其中,第一子载波间隔为应用第一AI模型的多个载波上分别激活的BWP中任一个子载波间隔。第一子载波间隔可以是网络设备与终端设备约定好的。
示例性的,第一子载波间隔可以为应用第一AI模型的多个载波上分别激活的BWP中最小的子载波间隔。例如,应用第一AI模型的载波分别为CC1,CC2和CC3,其中三个CC上激活的BWP分别为BWP1,BWP2和BWP3,且BWP1配置的子载波间隔<BWP2配置的子载波间隔<BWP3配置的子载波间隔,此时,第一AI模型开始应用的时刻和/或模型应用时间可以根据BWP1配置的子载波间隔确定。
在另一些实施例中,针对终端设备通过第二信息向网络设备指示第一AI模型的场景,第一AI 模型开始应用的时刻和/或模型应用时间,可以根据以下中的任意一项确定:
承载第二信息的上行信道的子载波间隔;
承载第二信息的响应信息的下行信道的子载波间隔;
应用所述第一AI模型的载波上所激活的BWP的子载波间隔。
在一种可能的实现方式中,第一AI模型开始应用的时刻和/或模型应用时间,可以根据承载第二信息的上行信道的子载波间隔确定。
示例性的,若第二信息通过UCI携带,则第一AI模型开始应用的时刻和/或模型应用时间根据该UCI的PUCCH的子载波间隔确定。若第二信息通过高层信令携带,则第一AI模型开始应用的时刻和/或模型应用时间根据承载该高层信令的PUSCH的子载波间隔确定。若第二信息通过PRACH携带,则第一AI模型开始应用的时刻和/或模型应用时间根据该PRACH的子载波间隔确定。
在另一种可能的实现方式中,第一AI模型开始应用的时刻和/或模型应用时间,可以根据承载第二信息的响应信息的下行信道的子载波间隔确定。
示例性的,若第二信息通过高层信令承载,则第一AI模型开始应用的时刻和/或模型应用时间可以根据携带该高层信令的PUSCH的HARQ-ACK信息所在的下行信道的子载波间隔确定。若第二信息通过PRACH承载,则第一AI模型开始应用的时刻和/或模型应用时间可以根据对应的RAR所在的下行信道的子载波间隔确定。
在又一种可能的实现方式中,第一AI模型开始应用的时刻和/或模型应用时间,可以根据应用第一AI模型的载波上所激活的BWP的子载波间隔。
在一些实施例中,当应用第一AI模型的载波的数量为多个时,第一AI模型开始应用的时刻和/或模型应用时间可以根据第一子载波间隔确定。其中,第一子载波间隔为应用第一AI模型的多个载波上分别激活的BWP中任一个子载波间隔。第一子载波间隔可以是网络设备与终端设备约定好的。
示例性的,第一子载波间隔可以为应用第一AI模型的多个载波上分别激活的BWP中最小的子载波间隔。例如,应用第一AI模型的载波分别为CC1,CC2和CC3,其中三个CC上激活的BWP分别为BWP1,BWP2和BWP3,且BWP1配置的子载波间隔<BWP2配置的子载波间隔<BWP3配置的子载波间隔,此时,第一AI模型开始应用的时刻和/或模型应用时间可以根据BWP1配置的子载波间隔确定。
由此可见,通过终端设备与网络设备约定相同的子载波间隔用于确定第一AI模型开始应用的时刻和/或模型应用时间,可以确保终端设备和网络确定的模型应用时间,以及第一AI模型开始应用的时刻是相同的,从而达到同步更新模型的效果。反之,当承载第一信息/第二信息的信道的子载波间隔与应用第一AI模型的载波所用的子载波间隔不同时,若未约定使用哪个子载波间隔确定第一AI模型开始应用的时刻和/或模型应用时间,则终端设备和网络设备无法确定实际的时间长度,可能导致两侧确定的时间长度不同。
以下结合具体应用场景对本申请实施例提供的方法进行详细阐述。
实施例一:
在该实施例一中,网络设备可以通过广播信令为其覆盖范围内的终端设备配置AI模型(即第一信息通过广播信令承载)。该实施例可以包括以下步骤:
S1、终端设备通过UE能力信息上报所支持的模型应用时间。
需要说明的是,模型应用时间为终端设备接收到用于配置AI模型的广播信令到应用该AI模型之间需要的时间间隔。其中,时间间隔以时隙为单位。
在一些实施例中,终端设备上报的模型应用时间可以包括:
终端设备支持的一个或多个AI模型分别对应的模型应用时间;
终端设备支持的一种或多种模型更新方式分别对应的模型应用时间;
终端设备支持的一个或多个模型类型分别对应的模型应用时间。
S2、网络设备根据终端设备上报的UE能力,为终端设备指示模型应用时间。
需要说明的是,网络设备需要考虑覆盖范围内多个终端设备分别上报的模型应用时间,来确定所指示的模型应用时间。示例性的,网络设备可以采用多个终端设备分别上报的模型应用时间中的最大值,作为覆盖范围内各个终端采用的模型应用时间,并指示给这些终端设备。
在一些实施例中,模型应用时间可以通过广播信令来指示。示例性的,广播信令可以是PBCH,SIB或者Group Common DCI。其中,指示模型应用时间的广播信令,与指示AI模型的广播信令可以为同一信令,也可以为不同的信令,本申请实施例对此不做限制。
在一些实施例中,模型应用时间也可以通过高层信令或者物理层信令来指示。示例性的,高层 信令可以是用于模型失败恢复响应MAC层信令指示,或者用于模型更新的MAC层信令。
S3、终端设备接收网络设备发送的用于配置AI模型的广播信令。
在一些实施例中,终端设备可以接收网络设备通过广播信令配置的AI模型的模型ID,功能(functionality),模型结构和模型参数中的至少一项。
可以理解的,网络设备不一定要配置整个AI模型,可以只配置或更新AI模型中的部分结构或者部分模型参数。
在一些实施例中,如果网络设备通过广播信令配置了一个新的模型的模型ID,则所述AI模型开始应用的时刻为所更新的模型开始应用的时刻。在所述时刻之前,终端设备可以使用之前采用的模型进行与网络设备之间的信息传输。
在一些实施例中,如果网络设备通过广播信令配置了一个新的模型的功能(functionality),则所述AI模型开始应用的时刻为所更新的模型开始应用的时刻。在所述时刻之前,终端设备可以使用之前的功能对应的模型进行与网络设备之间的信息传输。这里的功能也可以是功能ID。
在一些实施例中,如果网络设备通过广播信令更新了模型结构和模型参数,则所述AI模型开始应用的时刻为所更新的模型结构和模型参数开始应用的时刻。在所述时刻之前,终端设备可以使用之前采用的模型结构和模型参数进行与网络设备之间的信息传输。
在一些实施例中,如果网络设备通过广播信令只更新了模型参数,则所述AI模型开始应用的时刻为所更新的模型参数开始应用的时刻。在所述时刻之前,终端设备可以使用之前采用的模型参数进行与网络设备之间的信息传输。
需要说明的是,接收到用于配置AI模型的广播信令的终端设备都要应用该广播信令所配置的AI模型。例如,网络设备可以通过Group Common DCI来给一组终端设备指示所用的AI模型,所述Group Common DCI通过公共搜索空间CSS来传输,采用公共的RNTI进行加扰。
在一些实施例中,网络设备通过广播信令配置AI模型的同时,也可以通过相同的信令指示相应的模型应用时间。
可以理解的,网络设备可以给一组终端设备部署相同的AI模型,并指示相同的模型应用时间,这组终端设备就可以同时应用所述AI模型。这样在网络侧只需要部署一个AI模型作为对应的encoder/decoder模型,不需要为每个终端设备分别部署一个对应的模型,从而明显降低了网络设备的实现复杂度,同时也减少了下行信令的开销。
S4、终端设备根据网络设备配置的模型应用时间,确定AI模型开始应用的时刻。
其中,AI模型开始应用的时刻为:终端设备接收到用于配置AI模型的广播信令的时隙开始k个时隙之后的第一个时隙;k为网络设备指示的模型应用时间(以时隙为单位)。
示例性的,参考图8所示,假设终端设备在时隙n接收到用于配置AI模型的广播信令,则该AI模型开始应用的时刻为时隙n+k之后的第一个时隙,即时隙n+k+1开始应用。
需要说明的是,本实施例中AI模型开始应用的时刻和/或模型应用时间是以时隙为单位的参数,需要根据子载波间隔确定对应的时间长度/时刻,其中,不同子载波间隔对应的时间长度不同。
本申请实施例中,网络设备配置的AI模型开始应用的时刻和/或模型应用时间可以根据承载该广播信令的信道的子载波间隔确定。
可以理解的,通过终端设备与网络设备约定相同的子载波间隔用于确定AI模型开始应用的时刻和模型应用时间,可以确保终端设备和网络设备确定的模型应用时间和模型开始应用的时刻是相同的,从而达到同步更新模型的效果。反之,当承载用于配置AI模型的广播信令的信道的子载波间隔与应用该AI模型的载波所用的子载波间隔不同时,如果未确定基于哪个子载波间隔来确定AI模型开始应用的时刻和/或模型应用时间,那么终端设备和网络设备无法确定k个时隙的长度。
S5、在AI模型开始应用的时刻之后,终端设备采用广播信令配置的AI模型进行与网络设备之间的信息传输。
在一些实施例中,在终端设备应用所述AI模型之前,终端设备可以采用预定义的AI模型进行与网络设备之间的信息传输,或者,终端设备不基于AI模型进行与网络设备之间的信息传输。其中,预定义的AI模型即终端设备和网络设备预先约定好的一个AI模型。终端设备不基于AI模型进行与网络设备之间的信息传输,即终端设备采用传统的非AI的方式进行与网络设备之间的信息传输。
在一些实施例中,在广播信令配置的AI模型开始应用的时刻之后,终端设备可以将该AI模型用于如下任一过程:
下行CSI反馈;或者,
下行数据的信道解码;或者,
下行信号的解调;或者,
下行信道估计;或者,
下行射频信号处理;或者,
上行数据的信道编码;或者,
上行数据的调制;或者,
上行导频生成;或者,
上行射频信号处理。
S6、在AI模型开始应用的时刻之后,网络设备采用上述广播信令配置的AI模型对应的网络侧模型进行与终端设备之间的信息传输。
在一些实施例中,在广播信令配置的AI模型开始应用的时刻之后,网络设备可以将配置的AI模型对应的网络侧模型用于如下任一过程:
下行CSI反馈信息的解码;或者,
上行数据的信道解码;或者,
上行信号的解调;或者,
上行信道估计;或者,
上行射频信号处理;或者,
下行数据的信道编码;或者,
下行数据的调制;或者,
下行导频生成;或者,
下行射频信号处理。
需要说明的是,网络设备应用模型的模块与终端侧应用所述AI模型的模块是一一对应的,例如信道编码-信道解码,调制-解调,导频生成-信道估计,发端射频信号处理-收端射频信号处理等。
实施例二
在该实施例二中,网络设备可以通过高层信令为终端设备配置AI模型(即第一信息通过高层信令承载)。该实施例可以包括以下步骤:
S1、终端设备通过UE能力上报所支持的模型应用时间。
需要说明的是,模型应用时间为终端设备反馈用于配置AI模型的高层信令的HARQ-ACK信息之后到应用AI模型之间需要的时间间隔,时间间隔以时隙或OFDM符号为单位。
在一些实施例中,上述高层信令可以是RRC信令,也可以是MAC层信令(MAC CE)。
在一些实施例中,上述高层信令的HARQ-ACK信息即为承载第一信息的PDSCH的HARQ-ACK信息。
S2、终端设备通过高层信令接收网络设备配置的AI模型。
S3、终端设备根据S1中上报的模型应用时间,确定网络设备配置的AI模型开始应用的时刻。
在一种实施方式中,高层信令配置的AI模型开始应用的时刻可以为:携带该高层信令的PDSCH的HARQ-ACK信息所在时隙开始k个时隙之后的第一个时隙。
其中,k为模型应用时间,可以由终端设备通过UE能力上报所支持的模型应用时间确定。
需要说明的是,该实施方式中典型的高层信令为RRC信令。
在一种实施方式中,高层信令配置的AI模型开始应用的时刻可以为:携带该高层信令的PDSCH的HARQ-ACK信息所在时隙开始(k+3N)个时隙之后的第一个时隙。
其中,k为所述模型应用时间,N为1ms包含的时隙数量,即3N个时隙为3ms。
需要说明的是,典型的高层信令可以为MAC层信令。3N个时隙(即3ms)为MAC层信令生效至少需要的时间,通过间隔k+3N个时隙从而保证MAC层信令有足够的生效时间。
在一种实施方式中,高层信令配置的AI模型开始应用的时刻可以为:携带该高层信令的PDSCH的HARQ-ACK信息所在时隙开始p个时隙之后的第一个时隙。
其中,p为k和3N中的较大值,k为模型应用时间(时隙为单位),N为1ms包含的时隙数量,即3N个时隙为3ms。
需要说明的是,典型的高层信令可以为MAC层信令,3N个时隙(即3ms)为MAC层信令生效至少需要的时间,通过取k和3N中的较大值从而保证MAC层信令有足够的生效时间。
还需要说明的是,本实施例中AI模型开始应用的时刻和/或模型应用时间是以时隙为单位的参数,需要根据子载波间隔确定对应的时间长度/时刻,其中,不同子载波间隔对应的时间长度不同。
在一种实施方式中,网络设备配置的AI模型开始应用的时刻和/或模型应用时间可以根据携带 该高层信令的PDSCH的HARQ-ACK信息所在的上行信道的子载波间隔确定。
在另一种实施方式中,网络设备配置的AI模型开始应用的时刻和/或模型应用时间可以根据该AI模型的载波上所激活的BWP的子载波间隔确定。
如果该AI模型应用于多个载波,多个载波上都有激活的BWP,则AI模型开始应用的时刻和/或模型应用时间根据所述多个载波上激活的BWP中最小的子载波间隔来确定。这样即使在多载波的场景中,网络设备和终端设备也能确定唯一的一个开始应用时刻。
可以理解的,通过终端设备与网络设备约定相同的子载波间隔用于确定AI模型开始应用的时刻和模型应用时间,可以确保终端设备和网络设备确定的模型应用时间和模型开始应用的时刻是相同的,从而达到同步更新模型的效果。反之,当承载用于配置AI模型的广播信令的信道的子载波间隔与应用该AI模型的载波所用的子载波间隔不同时,如果未确定基于哪个子载波间隔来确定AI模型开始应用的时刻和/或模型应用时间,那么终端设备和网络设备无法确定k个时隙的长度。
S4、在AI模型开始应用的时刻之后,终端设备采用上述高层信令配置的AI模型进行与网络设备之间的信息传输。
S5、在AI模型开始应用的时刻之后,网络设备采用上述高层信令配置的AI模型对应的网络侧模型进行与终端设备之间的信息传输。
实施例三
在该实施例一中,网络设备可以通过DCI为终端设备配置AI模型(即第一信息通过DCI承载)。该实施例可以包括以下步骤:
S1、终端设备通过UE能力上报所支持的模型应用时间。
需要说明的是,模型应用时间为终端设备反馈用于配置AI模型的DCI信令的HARQ-ACK信息之后到应用AI模型之间需要的时间间隔,时间间隔以时隙或OFDM符号为单位。
S2、终端设备通过DCI信令接收网络设备配置的AI模型。
需要说明的是,网络设备可以通过UE专属搜索空间传输的DCI信令,为终端设备配置AI模型。示例性的,该DCI信令可以为网络设备发送的用于模型失败恢复响应的DCI,或者用于调度PDSCH或PUSCH的DCI。其中,PDSCH或PUSCH的传输可以基于AI模型(不一定是所配置的这个AI模型)。
S3、终端设备根据S1中上报的模型应用时间,确定网络设备配置的AI模型开始应用的时刻。
在一种实施方式中,DCI信令配置的AI模型开始应用的时刻可以为:承载该DCI的HARQ-ACK信息的上行信道的最后一个OFDM符号开始至少k个OFDM符号之后的第一个时隙。其中,k为模型应用时间(OFDM符号为单位)。
示例性的,参考图6A所示,假设承载所述DCI信令的HARQ-ACK信息的上行信道的最后一个OFDM符号为符号m,则AI模型开始应用的时刻为符号m+k所在时隙之后的第一个时隙。
在另一种实施方式中,DCI信令配置的AI模型开始应用的时刻可以为:承载所述DCI的HARQ-ACK信息的上行信道的最后一个OFDM符号开始至少k个OFDM符号之后的第一个OFDM符号。其中,k为模型应用时间(OFDM符号为单位)。
示例性的,参考图6N所示,假设承载所述DCI的HARQ-ACK信息的上行信道的最后一个OFDM符号为符号m,则所述AI模型开始应用的时刻为符号m+k之后的第一个符号,即符号m+k+1。
还需要说明的是,本实施例中AI模型开始应用的时刻和/或模型应用时间是以时隙/OFDM为单位的参数,需要根据子载波间隔确定对应的时间长度/时刻,其中,不同子载波间隔对应的时间长度不同。
其中,网络设备配置的AI模型开始应用的时刻和/或模型应用时间可以根据该AI模型的载波上所激活的BWP的子载波间隔确定。
如果该AI模型应用于多个载波,多个载波上都有激活的BWP,则AI模型开始应用的时刻和/或模型应用时间根据所述多个载波上激活的BWP中最小的子载波间隔来确定。这样即使在多载波的场景中,网络设备和终端设备也能确定唯一的一个开始应用时刻。
S4、在AI模型开始应用的时刻之后,终端设备采用上述DCI信令配置的AI模型进行与网络设备之间的信息传输。
S5、在AI模型开始应用的时刻之后,网络设备采用上述DCI信令配置的AI模型对应的网络侧模型进行与终端设备之间的信息传输。
综上所述,在终端设备侧和网络设备侧采用相对应的双端AI模型时,如果终端设备侧模型发生更新或重配置,本申请实施例提供的无线通信方法可以确保两侧的模型应用时刻是相同的,网络设 备与不同终端之间能够同步进行模型更新,从而保证对端的模型输出是可用的。
以上结合附图详细描述了本申请的优选实施方式,但是,本申请并不限于上述实施方式中的具体细节,在本申请的技术构思范围内,可以对本申请的技术方案进行多种简单变型,这些简单变型均属于本申请的保护范围。例如,在上述具体实施方式中所描述的各个具体技术特征,在不矛盾的情况下,可以通过任何合适的方式进行组合,为了避免不必要的重复,本申请对各种可能的组合方式不再另行说明。又例如,本申请的各种不同的实施方式之间也可以进行任意组合,只要其不违背本申请的思想,其同样应当视为本申请所公开的内容。又例如,在不冲突的前提下,本申请描述的各个实施例和/或各个实施例中的技术特征可以和现有技术任意的相互组合,组合之后得到的技术方案也应落入本申请的保护范围。
还应理解,在本申请的各种方法实施例中,上述各过程的序号的大小并不意味着执行顺序的先后,各过程的执行顺序应以其功能和内在逻辑确定,而不应对本申请实施例的实施过程构成任何限定。此外,在本申请实施例中,术语“下行”、“上行”和“侧行”用于表示信号或数据的传输方向,其中,“下行”用于表示信号或数据的传输方向为从站点发送至小区的用户设备的第一方向,“上行”用于表示信号或数据的传输方向为从小区的用户设备发送至站点的第二方向,“侧行”用于表示信号或数据的传输方向为从用户设备1发送至用户设备2的第三方向。例如,“下行信号”表示该信号的传输方向为第一方向。另外,本申请实施例中,术语“和/或”,仅仅是一种描述关联对象的关联关系,表示可以存在三种关系。具体地,A和/或B可以表示:单独存在A,同时存在A和B,单独存在B这三种情况。另外,本文中字符“/”,一般表示前后关联对象是一种“或”的关系。
图10是本申请实施例提供的无线通信装置1000的结构组成示意图,应用于终端设备,如图10所示,所述无线通信装置1000包括:
第一确定单元1001,配置为根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
第一通信单元1002,配置为在所述时刻之后,采用所述第一AI模型与所述网络设备进行通信。
在一些实施例中,所述模型应用时间为以下中的任意一项:
终端设备接收到第一信息之后到应用所述第一AI模型的时刻之间的时间间隔;
终端设备发送第一信息的HARQ-ACK信息之后到应用第一AI模型的时刻之间的时间间隔;
终端设备接收到第一信息之后到终端设备发送基于第一AI模型得到的反馈信息的时刻之间的时间间隔;
终端设备发送第一信息的HARQ-ACK信息之后到终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
其中,所述第一信息为用于配置所述第一AI模型的下行信息,所述时间间隔以时隙或OFDM符号为单位。
在一些实施例中,所述第一AI模型开始应用的时刻为以下中的任意一项:
承载第一信息的HARQ-ACK信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
承载第一信息的下行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
所述终端设备接收到第一信息的时隙开始k个时隙之后的第一个时隙;
承载第一信息的HARQ-ACK信息的上行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙;
其中,所述第一信息为用于配置所述第一AI模型的下行信息,k为所述模型应用时间,N为1ms包含的时隙数量。
在一些实施例中,所述无线通信装置1000还包括接收单元,配置为接收第一信息,所述第一信息是用于配置所述第一AI模型的下行信息,其中,所述第一信息包括以下中的一项或多项:
所述第一AI模型的标识信息;
所述第一AI模型的功能;
所述第一AI模型的模型结构;以及,
所述第一AI模型的模型参数。
在一些实施例中,所述模型应用时间为以下中的任意一项:
所述终端设备发送第二信息之后到应用所述第一AI模型的时刻之间的时间间隔;
所述终端设备发送第二信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的 时刻之间的时间间隔;
所述终端设备接收第二信息的响应信息之后到应用所述第一AI模型的时刻之间的时间间隔;
所述终端设备接收第二信息的响应信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
其中,所述第二信息为用于指示所述第一AI模型的上行信息,所述时间间隔以时隙或OFDM符号为单位。
在一些实施例中,所述第一AI模型开始应用的时刻为以下中的任意一项:
承载第二信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
承载第二信息的上行信道的响应信息的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
所述终端设备发送第二信息的时隙开始k个时隙之后的第一个时隙;
承载第二信息的响应信息的下行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙;
其中,所述第二信息为用于指示所述第一AI模型的上行信息,k为所述模型应用时间,N为1ms包含的时隙数量。
在一些实施例中,所述无线通信装置1000还包括发送单元,配置为发送第二信息,所述第二信息是用于指示所述第一AI模型的上行信息;
其中,所述第二信息包括以下中的一项或多项:
所述第一AI模型的标识信息;
所述第一AI模型的模型功能;
所述第一AI模型的模型结构;以及,
所述第一AI模型的模型参数。
在一些实施例中,所述上报给网络设备的模型应用时间包括以下中的一项或多项:
所述终端设备支持的一个或多个AI模型分别对应的模型应用时间;其中,所述终端设备支持的一个或多个AI模型包括所述第一AI模型;
所述终端设备支持的一个或多个模型功能分别对应的模型应用时间;其中,所述终端设备支持的一个或多个模型功能包括所述第一AI模型的功能;
所述终端设备支持的一种或多种模型更新方式分别对应的模型应用时间;其中,所述终端设备支持的一种或多种模型更新方式包括所述第一AI模型的更新方式;
所述终端设备支持的一个或多个模型类型分别对应的模型应用时间;其中,所述终端设备支持的一个或多个模型类型包括所述第一AI模型的类型。
在一些实施例中,所述模型更新方式包括:更新模型参数和/或更新模型结构。
在一些实施例中,所述模型类型包括:
高精度模型和低精度模型;或者,已知模型和未知模型。
在一些实施例中,所述第一AI模型开始应用的时刻和/或所述模型应用时间,根据以下任意一项确定:
承载第一信息的下行信道的子载波间隔;
承载第二信息的上行信道的子载波间隔;
承载第一信息的HARQ-ACK信息的上行信道的子载波间隔;
承载第二信息的响应信息的下行信道的子载波间隔;
应用所述第一AI模型的载波上所激活的BWP的子载波间隔;
其中,所述第一信息为用于配置所述第一AI模型的下行信息,所述第二信息为用于指示所述第一AI模型的上行信息。
在一些实施例中,应用所述第一AI模型的载波的数量为多个,
所述第一AI模型开始应用的时刻和/或所述模型应用时间根据第一子载波间隔确定,所述第一子载波间隔为应用所述第一AI模型的多个载波上分别激活的BWP中最小的子载波间隔。
在一些实施例中,在所述终端设备应用所述第一AI模型之前,所述终端设备基于第三AI模型进行与所述网络设备之间的信息传输,或者,所述终端设备不基于AI模型进行与所述网络设备之间的信息传输;
其中,所述第三AI模型为预定义的AI模型,或者历史AI模型。
图11是本申请实施例提供的无线通信装置1100的结构组成示意图,应用于网络设备,如图11所示,所述无线通信装置1100包括:
第二确定单元1101,配置为根据终端设备上报的模型应用时间,或者,根据为所述终端设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
第二通信单元1102,配置为在所述时刻之后,采用所述第一AI模型对应的第二AI模型与所述终端设备进行通信。
在一些实施例中,所述模型应用时间为以下中的任意一项:
所述终端设备接收到第一信息之后到应用所述第一AI模型的时刻之间的时间间隔;
所述终端设备发送第一信息的HARQ-ACK信息之后到应用所述第一AI模型的时刻之间的时间间隔;
所述终端设备接收到第一信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
所述终端设备发送第一信息的HARQ-ACK信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
其中,所述第一信息为用于配置所述第一AI模型的下行信息,所述时间间隔以时隙或OFDM符号为单位。
在一些实施例中,所述第一AI模型开始应用的时刻为以下中的任意一项:
承载第一信息的HARQ-ACK信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
承载第一信息的下行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
所述终端设备接收到第一信息的时隙开始k个时隙之后的第一个时隙;
承载第一信息的HARQ-ACK信息的上行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙;
其中,所述第一信息为用于配置所述第一AI模型的下行信息,k为所述模型应用时间,N为1ms包含的时隙数量。
在一些实施例中,所述无线通信装置1100还包括发送单元,配置为向所述终端设备发送第一信息,所述第一信息为用于配置所述第一AI模型的下行,其中,所述第一信息包括以下中的一项或多项:
所述第一AI模型的标识信息;
所述第一AI模型的功能;
所述第一AI模型的模型结构;以及,
所述第一AI模型的模型参数。
在一些实施例中,所述模型应用时间为以下中的任意一项:
所述终端设备发送第二信息之后到应用所述第一AI模型的时刻之间的时间间隔;
所述终端设备发送第二信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
所述终端设备接收第二信息的响应信息之后到应用所述第一AI模型的时刻之间的时间间隔;
所述终端设备接收第二信息的响应信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
其中,所述第二信息为用于指示所述第一AI模型的上行信息,所述时间间隔以时隙或OFDM符号为单位。
在一些实施例中,所述第一AI模型开始应用的时刻为以下中的任意一项:
承载第二信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
承载第二信息的上行信道的响应信息的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
所述终端设备发送第二信息的时隙开始k个时隙之后的第一个时隙;
承载第二信息的响应信息的下行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙;
其中,所述第二信息为用于指示所述第一AI模型的上行信息,k为所述模型应用时间,N为1ms 包含的时隙数量。
在一些实施例中,所述无线通信装置1100还包括接收单元,配置为接收所述终端设备发送的第二信息,所述第二信息是用于指示所述第一AI模型的上行信息;
其中,所述第二信息包括以下中的一项或多项:
所述第一AI模型的标识信息;
所述第一AI模型的模型功能;
所述第一AI模型的模型结构;以及,
所述第一AI模型的模型参数。
在一些实施例中,所述终端设备上报的模型应用时间包括以下中的一项或多项:
所述终端设备支持的一个或多个AI模型分别对应的模型应用时间;其中,所述终端设备支持的一个或多个AI模型包括所述第一AI模型;
所述终端设备支持的一个或多个模型功能分别对应的模型应用时间;其中,所述终端设备支持的一个或多个模型功能包括所述第一AI模型的功能;
所述终端设备支持的一种或多种模型更新方式分别对应的模型应用时间;其中,所述终端设备支持的一种或多种模型更新方式包括所述第一AI模型的更新方式;
所述终端设备支持的一个或多个模型类型分别对应的模型应用时间;其中,所述终端设备支持的一个或多个模型类型包括所述第一AI模型的类型。
在一些实施例中,所述模型更新方式包括:更新模型参数和/或更新模型结构。
在一些实施例中,所述模型类型包括:
高精度模型和低精度模型;或者,已知模型和未知模型。
在一些实施例中,所述网络设备为所述终端设备配置的模型应用时间,基于所述终端设备上报的模型应用时间确定。
在一些实施例中,所述网络设备为所述终端设备配置的模型应用时间的时长,大于或等于所述终端设备上报的模型应用时间的时长。
在一些实施例中,所述第一AI模型开始应用的时刻和/或所述模型应用时间,根据以下任意一项确定:
承载第一信息的下行信道的子载波间隔;
承载第二信息的上行信道的子载波间隔;
承载第一信息的HARQ-ACK信息的上行信道的子载波间隔;
承载第二信息的响应信息的下行信道的子载波间隔;
应用所述第一AI模型的载波上所激活的BWP的子载波间隔;
其中,所述第一信息为用于配置所述第一AI模型的下行信息,所述第二信息为用于指示所述第一AI模型的上行信息。
在一些实施例中,应用所述第一AI模型的载波的数量为多个;
所述第一AI模型开始应用的时刻和/或所述模型应用时间,根据第一子载波间隔确定,所述第一子载波间隔为应用所述第一AI模型的多个载波上分别激活的BWP中最小的子载波间隔。
在一些实施例中,在所述网络设备应用所述第一AI模型之前,所述网络设备基于第三AI模型对应的第四AI模型进行与所述终端设备之间的信息传输,或者,所述网络设备不基于AI模型进行与所述终端设备之间的信息传输;
其中,所述第三AI模型为预定义的模型,或者历史AI模型。
本领域技术人员应当理解,本申请实施例的上述无线通信装置的相关描述可以参照本申请实施例的无线通信方法的相关描述进行理解。
图12是本申请实施例提供的一种通信设备1200示意性结构图。该通信设备可以终端设备,也可以是网络设备。图12所示的通信设备1200包括处理器1210,处理器1210可以从存储器中调用并运行计算机程序,以实现本申请实施例中的方法。
可选地,如图12所示,通信设备1200还可以包括存储器1220。其中,处理器1210可以从存储器1220中调用并运行计算机程序,以实现本申请实施例中的方法。
其中,存储器1220可以是独立于处理器1210的一个单独的器件,也可以集成在处理器1210中。
可选地,如图12所示,通信设备1200还可以包括收发器1230,处理器1210可以控制该收发器1230与其他设备进行通信,具体地,可以向其他设备发送信息或数据,或接收其他设备发送的信息或数据。
其中,收发器1230可以包括发射机和接收机。收发器1230还可以进一步包括天线,天线的数量可以为一个或多个。
可选地,该通信设备1200具体可为本申请实施例的网络设备,并且该通信设备1800可以实现本申请实施例的各个方法中由网络设备实现的相应流程,为了简洁,在此不再赘述。
可选地,该通信设备1200具体可为本申请实施例的移动终端/终端设备,并且该通信设备1200可以实现本申请实施例的各个方法中由移动终端/终端设备实现的相应流程,为了简洁,在此不再赘述。
图13是本申请实施例的芯片的示意性结构图。图13所示的芯片1300包括处理器1910,处理器1310可以从存储器中调用并运行计算机程序,以实现本申请实施例中的方法。
可选地,如图13所示,芯片1300还可以包括存储器1320。其中,处理器1310可以从存储器1320中调用并运行计算机程序,以实现本申请实施例中的方法。
其中,存储器1320可以是独立于处理器1310的一个单独的器件,也可以集成在处理器1310中。
可选地,该芯片1300还可以包括输入接口1330。其中,处理器1310可以控制该输入接口1330与其他设备或芯片进行通信,具体地,可以获取其他设备或芯片发送的信息或数据。
可选地,该芯片1300还可以包括输出接口1340。其中,处理器1310可以控制该输出接口1340与其他设备或芯片进行通信,具体地,可以向其他设备或芯片输出信息或数据。
可选地,该芯片可应用于本申请实施例中的网络设备,并且该芯片可以实现本申请实施例的各个方法中由网络设备实现的相应流程,为了简洁,在此不再赘述。
可选地,该芯片可应用于本申请实施例中的移动终端/终端设备,并且该芯片可以实现本申请实施例的各个方法中由移动终端/终端设备实现的相应流程,为了简洁,在此不再赘述。
应理解,本申请实施例提到的芯片还可以称为系统级芯片,系统芯片,芯片系统或片上系统芯片等。
本申请实施例还提供了一种计算机存储介质,所述计算机存储介质存储有一个或者多个程序,所述一个或者多个程序可被一个或者多个处理器执行,以实现本申请实施例中的方法。
图14是本申请实施例提供的一种通信系统1400的示意性框图。如图14所示,该通信系统1400包括终端设备1410和网络设备1420。
其中,该终端设备1410可以用于实现上述方法中由终端设备实现的相应的功能,以及该网络设备1420可以用于实现上述方法中由网络设备实现的相应的功能为了简洁,在此不再赘述。
应理解,本申请实施例的处理器可能是一种集成电路芯片,具有信号的处理能力。在实现过程中,上述方法实施例的各步骤可以通过处理器中的硬件的集成逻辑电路或者软件形式的指令完成。上述的处理器可以是通用处理器、数字信号处理器(Digital Signal Processor,DSP)、专用集成电路(Application Specific Integrated Circuit,ASIC)、现成可编程门阵列(Field Programmable Gate Array,FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。可以实现或者执行本申请实施例中的公开的各方法、步骤及逻辑框图。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。结合本申请实施例所公开的方法的步骤可以直接体现为硬件译码处理器执行完成,或者用译码处理器中的硬件及软件模块组合执行完成。软件模块可以位于随机存储器,闪存、只读存储器,可编程只读存储器或者电可擦写可编程存储器、寄存器等本领域成熟的存储介质中。该存储介质位于存储器,处理器读取存储器中的信息,结合其硬件完成上述方法的步骤。
可以理解,本申请实施例中的存储器可以是易失性存储器或非易失性存储器,或可包括易失性和非易失性存储器两者。其中,非易失性存储器可以是只读存储器(Read-Only Memory,ROM)、可编程只读存储器(Programmable ROM,PROM)、可擦除可编程只读存储器(Erasable PROM,EPROM)、电可擦除可编程只读存储器(Electrically EPROM,EEPROM)或闪存。易失性存储器可以是随机存取存储器(Random Access Memory,RAM),其用作外部高速缓存。通过示例性但不是限制性说明,许多形式的RAM可用,例如静态随机存取存储器(Static RAM,SRAM)、动态随机存取存储器(Dynamic RAM,DRAM)、同步动态随机存取存储器(Synchronous DRAM,SDRAM)、双倍数据速率同步动态随机存取存储器(Double Data Rate SDRAM,DDR SDRAM)、增强型同步动态随机存取存储器(Enhanced SDRAM,ESDRAM)、同步连接动态随机存取存储器(Synchlink DRAM,SLDRAM)和直接内存总线随机存取存储器(Direct Rambus RAM,DR RAM)。应注意,本文描述的系统和方法的存储器旨在包括但不限于这些和任意其它适合类型的存储器。
应理解,上述存储器为示例性但不是限制性说明,例如,本申请实施例中的存储器还可以是静态随机存取存储器(static RAM,SRAM)、动态随机存取存储器(dynamic RAM,DRAM)、同步动 态随机存取存储器(synchronous DRAM,SDRAM)、双倍数据速率同步动态随机存取存储器(double data rate SDRAM,DDR SDRAM)、增强型同步动态随机存取存储器(enhanced SDRAM,ESDRAM)、同步连接动态随机存取存储器(synch link DRAM,SLDRAM)以及直接内存总线随机存取存储器(Direct Rambus RAM,DR RAM)等等。也就是说,本申请实施例中的存储器旨在包括但不限于这些和任意其它适合类型的存储器。
本申请实施例还提供了一种计算机可读存储介质,用于存储计算机程序。
可选的,该计算机可读存储介质可应用于本申请实施例中的网络设备,并且该计算机程序使得计算机执行本申请实施例的各个方法中由网络设备实现的相应流程,为了简洁,在此不再赘述。
可选地,该计算机可读存储介质可应用于本申请实施例中的移动终端/终端设备,并且该计算机程序使得计算机执行本申请实施例的各个方法中由移动终端/终端设备实现的相应流程,为了简洁,在此不再赘述。
本申请实施例还提供了一种计算机程序产品,包括计算机程序指令。
可选的,该计算机程序产品可应用于本申请实施例中的网络设备,并且该计算机程序指令使得计算机执行本申请实施例的各个方法中由网络设备实现的相应流程,为了简洁,在此不再赘述。
可选地,该计算机程序产品可应用于本申请实施例中的移动终端/终端设备,并且该计算机程序指令使得计算机执行本申请实施例的各个方法中由移动终端/终端设备实现的相应流程,为了简洁,在此不再赘述。
本申请实施例还提供了一种计算机程序。
可选的,该计算机程序可应用于本申请实施例中的网络设备,当该计算机程序在计算机上运行时,使得计算机执行本申请实施例的各个方法中由网络设备实现的相应流程,为了简洁,在此不再赘述。
可选地,该计算机程序可应用于本申请实施例中的移动终端/终端设备,当该计算机程序在计算机上运行时,使得计算机执行本申请实施例的各个方法中由移动终端/终端设备实现的相应流程,为了简洁,在此不再赘述。
本领域普通技术人员可以意识到,结合本文中所公开的实施例描述的各示例的单元及算法步骤,能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行,取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能,但是这种实现不应认为超出本申请的范围。
所属领域的技术人员可以清楚地了解到,为描述的方便和简洁,上述描述的系统、装置和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。
在本申请所提供的几个实施例中,应该理解到,所揭露的系统、装置和方法,可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,所述单元的划分,仅仅为一种逻辑功能划分,实际实现时可以有另外的划分方式,例如多个单元或组件可以结合或者可以集成到另一个系统,或一些特征可以忽略,或不执行。另一点,所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口,装置或单元的间接耦合或通信连接,可以是电性,机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的,作为单元显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外,在本申请各个实施例中的各功能单元可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。
所述功能如果以软件功能单元的形式实现并作为独立的产品销售或使用时,可以存储在一个计算机可读取存储介质中。基于这样的理解,本申请的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质中,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行本申请各个实施例所述方法的全部或部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(Read-Only Memory,)ROM、随机存取存储器(Random Access Memory,RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述,仅为本申请的具体实施方式,但本申请的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本申请揭露的技术范围内,可轻易想到变化或替换,都应涵盖在本申请的保护范围之内。因此,本申请的保护范围应所述以权利要求的保护范围为准。

Claims (36)

  1. 一种无线通信方法,所述方法包括:
    终端设备根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
    所述终端设备在所述时刻之后,采用所述第一AI模型与所述网络设备进行通信。
  2. 根据权利要求1所述的方法,其中,所述模型应用时间为以下中的任意一项:
    所述终端设备接收到第一信息之后到应用所述第一AI模型的时刻之间的时间间隔;
    所述终端设备发送第一信息的HARQ-ACK信息之后到应用所述第一AI模型的时刻之间的时间间隔;
    所述终端设备接收到第一信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
    所述终端设备发送第一信息的HARQ-ACK信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
    其中,所述第一信息为用于配置所述第一AI模型的下行信息,所述时间间隔以时隙或OFDM符号为单位。
  3. 根据权利要求1或2所述的方法,其中,所述第一AI模型开始应用的时刻为以下中的任一项:
    承载第一信息的HARQ-ACK信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
    承载第一信息的下行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
    所述终端设备接收到第一信息的时隙开始k个时隙之后的第一个时隙;
    承载第一信息的HARQ-ACK信息的上行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙;
    其中,所述第一信息为用于配置所述第一AI模型的下行信息,k为所述模型应用时间,N为1ms包含的时隙数量。
  4. 根据权利要求2或3所述的方法,其中,所述方法还包括:
    所述终端设备接收第一信息,所述第一信息是用于配置所述第一AI模型的下行信息,
    其中,所述第一信息包括以下中的一项或多项:
    所述第一AI模型的标识信息;
    所述第一AI模型的模型功能;
    所述第一AI模型的模型结构;以及,
    所述第一AI模型的模型参数。
  5. 根据权利要求1所述的方法,其中,所述模型应用时间为以下中的任意一项:
    所述终端设备发送第二信息之后到应用所述第一AI模型的时刻之间的时间间隔;
    所述终端设备发送第二信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
    所述终端设备接收第二信息的响应信息之后到应用所述第一AI模型的时刻之间的时间间隔;
    所述终端设备接收第二信息的响应信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
    其中,所述第二信息为用于指示所述第一AI模型的上行信息,所述时间间隔以时隙或OFDM符号为单位。
  6. 根据权利要求1或5所述的方法,其中,所述第一AI模型开始应用的时刻为以下中的任一项:
    承载第二信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
    承载第二信息的上行信道的响应信息的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
    所述终端设备发送第二信息的时隙开始k个时隙之后的第一个时隙;
    承载第二信息的响应信息的下行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙;
    其中,所述第二信息为用于指示所述第一AI模型的上行信息,k为所述模型应用时间,N为1ms包含的时隙数量。
  7. 根据权利要求5或6所述的方法,其中,所述方法还包括:
    所述终端设备发送第二信息,所述第二信息是用于指示所述第一AI模型的上行信息;
    其中,所述第二信息包括以下中的一项或多项:
    所述第一AI模型的标识信息;
    所述第一AI模型的模型功能;
    所述第一AI模型的模型结构;以及,
    所述第一AI模型的模型参数。
  8. 根据权利要求1-7任一项所述的方法,其中,所述上报给网络设备的模型应用时间包括以下中的一项或多项:
    所述终端设备支持的一个或多个AI模型分别对应的模型应用时间;其中,所述终端设备支持的一个或多个AI模型包括所述第一AI模型;
    所述终端设备支持的一个或多个模型功能分别对应的模型应用时间;其中,所述终端设备支持的一个或多个模型功能包括所述第一AI模型的功能;
    所述终端设备支持的一种或多种模型更新方式分别对应的模型应用时间;其中,所述终端设备支持的一种或多种模型更新方式包括所述第一AI模型的更新方式;
    所述终端设备支持的一个或多个模型类型分别对应的模型应用时间;其中,所述终端设备支持的一个或多个模型类型包括所述第一AI模型的类型。
  9. 根据权利要求8所述的方法,其中,所述模型更新方式包括:更新模型参数和/或更新模型结构。
  10. 根据权利要求8或9所述的方法,其中,所述模型类型包括:
    高精度模型和低精度模型;或者,已知模型和未知模型。
  11. 根据权利要求1-10任一项所述的方法,其中,所述第一AI模型开始应用的时刻和/或所述模型应用时间,根据以下任意一项确定:
    承载第一信息的下行信道的子载波间隔;
    承载第二信息的上行信道的子载波间隔;
    承载第一信息的HARQ-ACK信息的上行信道的子载波间隔;
    承载第二信息的响应信息的下行信道的子载波间隔;
    应用所述第一AI模型的载波上所激活的BWP的子载波间隔;
    其中,所述第一信息为用于配置所述第一AI模型的下行信息,所述第二信息为用于指示所述第一AI模型的上行信息。
  12. 根据权利要求11所述的方法,其中,应用所述第一AI模型的载波的数量为多个,
    所述第一AI模型开始应用的时刻和/或所述模型应用时间根据第一子载波间隔确定,所述第一子载波间隔为应用所述第一AI模型的多个载波上分别激活的BWP中最小的子载波间隔。
  13. 根据权利要求1-12任一项所述的方法,其中,在所述终端设备应用所述第一AI模型之前,所述终端设备基于第三AI模型进行与所述网络设备之间的信息传输,或者,所述终端设备不基于AI模型进行与所述网络设备之间的信息传输;
    其中,所述第三AI模型为预定义的AI模型,或者历史AI模型。
  14. 一种无线通信方法,所述方法包括:
    网络设备根据终端设备上报的模型应用时间,或者,根据为所述终端设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
    所述网络设备在所述时刻之后,采用所述第一AI模型对应的第二AI模型与所述终端设备进行通信。
  15. 根据权利要求14所述的方法,其中,所述模型应用时间为以下中的任意一项:
    所述终端设备接收到第一信息之后到应用所述第一AI模型的时刻之间的时间间隔;
    所述终端设备发送第一信息的HARQ-ACK信息之后到应用所述第一AI模型的时刻之间的时间 间隔;
    所述终端设备接收到第一信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
    所述终端设备发送第一信息的HARQ-ACK信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
    其中,所述第一信息为用于配置所述第一AI模型的下行信息,所述时间间隔以时隙或OFDM符号为单位。
  16. 根据权利要求14或15所述的方法,其中,所述第一AI模型开始应用的时刻为以下中的任一项:
    承载第一信息的HARQ-ACK信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
    承载第一信息的下行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
    所述终端设备接收到第一信息的时隙开始k个时隙之后的第一个时隙;
    承载第一信息的HARQ-ACK信息的上行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙;
    其中,所述第一信息为用于配置所述第一AI模型的下行信息,k为所述模型应用时间,N为1ms包含的时隙数量。
  17. 根据权利要求15或16所述的方法,其中,所述方法还包括:
    所述网络设备向所述终端设备发送第一信息,所述第一信息为用于配置所述第一AI模型的下行,其中,所述第一信息包括以下中的一项或多项:
    所述第一AI模型的标识信息;
    所述第一AI模型的功能;
    所述第一AI模型的模型结构;以及,
    所述第一AI模型的模型参数。
  18. 根据权利要求14所述的方法,其中,所述模型应用时间为以下中的任意一项:
    所述终端设备发送第二信息之后到应用所述第一AI模型的时刻之间的时间间隔;
    所述终端设备发送第二信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
    所述终端设备接收第二信息的响应信息之后到应用所述第一AI模型的时刻之间的时间间隔;
    所述终端设备接收第二信息的响应信息之后到所述终端设备发送基于所述第一AI模型得到的反馈信息的时刻之间的时间间隔;
    其中,所述第二信息为用于指示所述第一AI模型的上行信息,所述时间间隔以时隙或OFDM符号为单位。
  19. 根据权利要求14或18所述的方法,其中,所述第一AI模型开始应用的时刻为以下中的任一项:
    承载第二信息的上行信道的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
    承载第二信息的上行信道的响应信息的最后一个OFDM符号开始k个OFDM符号之后的第一个时隙或者第一个OFDM符号;
    所述终端设备发送第二信息的时隙开始k个时隙之后的第一个时隙;
    承载第二信息的响应信息的下行信道所在时隙开始k个时隙,或(k+3N)个时隙,或max(k,3N)个时隙之后的第一个时隙;
    其中,所述第二信息为用于指示所述第一AI模型的上行信息,k为所述模型应用时间,N为1ms包含的时隙数量。
  20. 根据权利要求18或19所述的方法,其中,所述方法还包括:
    所述网络设备接收所述终端设备发送的第二信息,所述第二信息是用于指示所述第一AI模型的上行信息;
    其中,所述第二信息包括以下中的一项或多项:
    所述第一AI模型的标识信息;
    所述第一AI模型的模型功能;
    所述第一AI模型的模型结构;以及,
    所述第一AI模型的模型参数。
  21. 根据权利要求14-20任一项所述的方法,其中,所述终端设备上报的模型应用时间包括以下中的一项或多项:
    所述终端设备支持的一个或多个AI模型分别对应的模型应用时间;其中,所述终端设备支持的一个或多个AI模型包括所述第一AI模型;
    所述终端设备支持的一个或多个模型功能分别对应的模型应用时间;其中,所述终端设备支持的一个或多个模型功能包括所述第一AI模型的功能;
    所述终端设备支持的一种或多种模型更新方式分别对应的模型应用时间;其中,所述终端设备支持的一种或多种模型更新方式包括所述第一AI模型的更新方式;
    所述终端设备支持的一个或多个模型类型分别对应的模型应用时间;其中,所述终端设备支持的一个或多个模型类型包括所述第一AI模型的类型。
  22. 根据权利要求21所述的方法,其中,所述模型更新方式包括:更新模型参数和/或更新模型结构。
  23. 根据权利要求21或22所述的方法,其中,所述模型类型包括:
    高精度模型和低精度模型;或者,已知模型和未知模型。
  24. 根据权利要求14-23任一项所述的方法,其中,
    所述网络设备为所述终端设备配置的模型应用时间,基于所述终端设备上报的模型应用时间确定。
  25. 根据权利要求14-24任一项所述的方法,其中,
    所述网络设备为所述终端设备配置的模型应用时间的时长,大于或等于所述终端设备上报的模型应用时间的时长。
  26. 根据权利要求14-25任一项所述的方法,其中,所述第一AI模型开始应用的时刻和/或所述模型应用时间,根据以下任意一项确定:
    承载第一信息的下行信道的子载波间隔;
    承载第二信息的上行信道的子载波间隔;
    承载第一信息的HARQ-ACK信息的上行信道的子载波间隔;
    承载第二信息的响应信息的下行信道的子载波间隔;
    应用所述第一AI模型的载波上所激活的BWP的子载波间隔;
    其中,所述第一信息为用于配置所述第一AI模型的下行信息,所述第二信息为用于指示所述第一AI模型的上行信息。
  27. 根据权利要求26所述的方法,其中,应用所述第一AI模型的载波的数量为多个;
    所述第一AI模型开始应用的时刻和/或所述模型应用时间,根据第一子载波间隔确定,所述第一子载波间隔为应用所述第一AI模型的多个载波上分别激活的BWP中最小的子载波间隔。
  28. 根据权利要求14-27任一项所述的方法,其中,在所述网络设备应用所述第一AI模型之前,所述网络设备基于第三AI模型对应的第四AI模型进行与所述终端设备之间的信息传输,或者,所述网络设备不基于AI模型进行与所述终端设备之间的信息传输;
    其中,所述第三AI模型为预定义的模型,或者历史AI模型。
  29. 一种无线通信装置,应用于终端设备,所述装置包括:
    第一确定单元,配置为根据上报给网络设备的模型应用时间,或者,根据网络设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
    第一通信单元,配置为在所述时刻之后,采用所述第一AI模型与所述网络设备进行通信。
  30. 一种无线通信装置,应用于网络设备,所述装置包括:
    第二确定单元,配置为根据终端设备上报的模型应用时间,或者,根据为所述终端设备配置的模型应用时间,确定第一AI模型开始应用的时刻;
    第二通信单元,配置为在所述时刻之后,采用所述第一AI模型对应的第二AI模型与所述终端设备进行通信。
  31. 一种终端设备,包括:
    存储器,用于存储计算机可执行指令;
    处理器,与所述存储器连接,用于通过执行所述计算机可执行指令,实现权利要求1至13中任一项所述的方法。
  32. 一种网络设备,包括:
    存储器,用于存储计算机可执行指令;
    处理器,与所述存储器连接,用于通过执行所述计算机可执行指令,实现权利要求14至28中任一项所述的方法。
  33. 一种芯片,所述芯片包括:
    处理器,用于从存储器中调用并运行计算机程序,使得安装有所述芯片的设备执行如权利要求1至13中任一项所述的方法,或者,执行如权利要求14至28中任一项所述的方法。
  34. 一种计算机可读存储介质,所述计算机可读存储介质存储有计算机程序,所述计算机程序被至少一个处理器执行时实现如权利要求1至13中任一项所述的方法,或者,实现如权利要求14至28中任一项所述的方法。
  35. 一种计算机程序产品,所述计算机程序产品包括计算机存储介质,所述计算机存储介质存储计算机程序,所述计算机程序包括能够由至少一个处理器执行的指令,当所述指令由所述至少一个处理器执行时实现权利要求1至13中任一项所述的方法,或者,实现如权利要求14至28中任一项所述的方法。
  36. 一种计算机程序,所述计算机程序使得计算机执行如权利要求1至13中任一项所述的方法,或者,实现如权利要求14至28中任一项所述的方法。
PCT/CN2023/112522 2023-08-11 2023-08-11 无线通信方法及装置、终端设备、网络设备 Pending WO2025035256A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
CN202380101132.0A CN121713607A (zh) 2023-08-11 2023-08-11 无线通信方法及装置、终端设备、网络设备
PCT/CN2023/112522 WO2025035256A1 (zh) 2023-08-11 2023-08-11 无线通信方法及装置、终端设备、网络设备
US19/419,652 US20260106806A1 (en) 2023-08-11 2025-12-15 Wireless communication method and apparatus, terminal device, and network device
MX2025015518A MX2025015518A (es) 2023-08-11 2025-12-18 Metodo y aparato para comunicacion inalambrica, dispositivo terminal y dispositivo de red

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2023/112522 WO2025035256A1 (zh) 2023-08-11 2023-08-11 无线通信方法及装置、终端设备、网络设备

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US19/419,652 Continuation US20260106806A1 (en) 2023-08-11 2025-12-15 Wireless communication method and apparatus, terminal device, and network device

Publications (1)

Publication Number Publication Date
WO2025035256A1 true WO2025035256A1 (zh) 2025-02-20

Family

ID=94632006

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/112522 Pending WO2025035256A1 (zh) 2023-08-11 2023-08-11 无线通信方法及装置、终端设备、网络设备

Country Status (4)

Country Link
US (1) US20260106806A1 (zh)
CN (1) CN121713607A (zh)
MX (1) MX2025015518A (zh)
WO (1) WO2025035256A1 (zh)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110636020A (zh) * 2019-08-05 2019-12-31 北京大学 一种自适应通信系统神经网络均衡方法
CN112997435A (zh) * 2019-09-04 2021-06-18 谷歌有限责任公司 用于无线通信的神经网络形成配置反馈
CN114175051A (zh) * 2019-08-14 2022-03-11 谷歌有限责任公司 关于深度神经网络的基站-用户设备消息传递
CN115485995A (zh) * 2020-04-29 2022-12-16 华为技术有限公司 用于调整神经网络的方法和装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110636020A (zh) * 2019-08-05 2019-12-31 北京大学 一种自适应通信系统神经网络均衡方法
CN114175051A (zh) * 2019-08-14 2022-03-11 谷歌有限责任公司 关于深度神经网络的基站-用户设备消息传递
CN112997435A (zh) * 2019-09-04 2021-06-18 谷歌有限责任公司 用于无线通信的神经网络形成配置反馈
CN115485995A (zh) * 2020-04-29 2022-12-16 华为技术有限公司 用于调整神经网络的方法和装置

Also Published As

Publication number Publication date
MX2025015518A (es) 2026-02-03
US20260106806A1 (en) 2026-04-16
CN121713607A (zh) 2026-03-20

Similar Documents

Publication Publication Date Title
JP7836426B2 (ja) ユーザ装置及びスケジューリングノード
CN109997398B (zh) 指示无线信道状态的系统和方法
WO2021208779A1 (zh) Srs的传输方法、装置、系统、存储介质及电子装置
WO2019170089A1 (zh) 信息传输的方法、装置和通信节点
US20240056888A1 (en) Quality of Service Management for Protocol Data Unit Sets
CN115473614A (zh) 一种csi上报方法及装置、终端设备、网络设备
CN116547935A (zh) 无线通信方法、终端设备和网络设备
WO2022014281A1 (ja) 端末、基地局及び通信方法
EP4340498A1 (en) Method and apparatus for determining repeated transmission, terminal, and network side device
WO2024149386A1 (zh) 通信方法与装置、终端设备、网络设备和芯片
CN114830707B (zh) 通信方法和通信装置
US20240356609A1 (en) Method and device for transmitting and receiving channel state information in wireless communication system
WO2025139843A1 (zh) 通信方法和通信装置
CN114828228A (zh) Pdcch传输方法、装置、终端及网络侧设备
US12587243B2 (en) Apparatus and method for transmitting and receiving signal according to channel state in wireless communication system
WO2024207959A1 (zh) 传输资源确定方法、设备、装置和存储介质
WO2025035256A1 (zh) 无线通信方法及装置、终端设备、网络设备
WO2024131889A1 (zh) 一种通信方法、装置、芯片及模组设备
WO2024140409A1 (zh) 一种信道状态信息csi报告的配置方法及相关装置
CN115119327B (zh) 信息传输方法、装置及存储介质
CN115668840B (zh) 无线通信方法和设备
WO2025024999A1 (zh) 处理时延要求的确定方法及装置、终端设备、网络设备
WO2024055242A1 (zh) 一种通信方法及装置、通信设备、融合组网架构
WO2025015562A1 (zh) 通信方法、装置、设备、芯片、存储介质、产品及程序
CN121692438A (zh) 传输控制方法、装置、终端及接入网设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23948698

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: MX/A/2025/015518

Country of ref document: MX

WWP Wipo information: published in national office

Ref document number: MX/A/2025/015518

Country of ref document: MX

WWE Wipo information: entry into national phase

Ref document number: 2023948698

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE

REG Reference to national code

Ref country code: BR

Ref legal event code: B01A

Ref document number: 112026002736

Country of ref document: BR