WO2024007949A1 - Ai模型处理方法、装置、终端及网络侧设备 - Google Patents

Ai模型处理方法、装置、终端及网络侧设备 Download PDF

Info

Publication number
WO2024007949A1
WO2024007949A1 PCT/CN2023/103909 CN2023103909W WO2024007949A1 WO 2024007949 A1 WO2024007949 A1 WO 2024007949A1 CN 2023103909 W CN2023103909 W CN 2023103909W WO 2024007949 A1 WO2024007949 A1 WO 2024007949A1
Authority
WO
WIPO (PCT)
Prior art keywords
frequency domain
terminal
domain resource
model
channel matrix
Prior art date
Application number
PCT/CN2023/103909
Other languages
English (en)
French (fr)
Inventor
任千尧
Original Assignee
维沃移动通信有限公司
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 维沃移动通信有限公司 filed Critical 维沃移动通信有限公司
Publication of WO2024007949A1 publication Critical patent/WO2024007949A1/zh

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H04L25/022—Channel estimation of frequency response
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H04L25/024—Channel estimation channel estimation algorithms
    • H04L25/0242—Channel estimation channel estimation algorithms using matrix methods
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network

Definitions

  • This application belongs to the field of communication technology, and specifically relates to an AI model processing method, device, terminal and network side equipment.
  • AI Artificial Intelligence
  • Communication services have an impact and affect the communication performance between communication equipment.
  • the embodiments of this application provide an AI model processing method, device, terminal and network-side equipment, which can solve problems in related technologies.
  • the first aspect provides an AI model processing method, including:
  • the terminal determines the first frequency domain resource used for AI model training
  • the terminal performs channel estimation on the first frequency domain resource, obtains a target channel matrix, and performs at least one of the following:
  • the target channel matrix is sent to a network side device, and the network side device is used to train and/or update the AI model based on the target channel matrix.
  • the second aspect provides an AI model processing method, including:
  • the network side device receives a target channel matrix sent by the terminal, where the target channel matrix is a channel matrix obtained by the terminal performing channel estimation on the first frequency domain resource;
  • the network side device trains and/or updates the AI model based on the target channel matrix.
  • an AI model processing device including:
  • Determining module used to determine the first frequency domain resource used for AI model training
  • An execution module configured to perform channel estimation on the first frequency domain resource, obtain a target channel matrix, and perform at least one of the following:
  • the target channel matrix is sent to a network side device, and the network side device is used to train and/or update the AI model based on the target channel matrix.
  • an AI model processing device including:
  • a receiving module configured to receive a target channel matrix sent by the terminal, where the target channel matrix is a channel matrix obtained by the terminal performing channel estimation on the first frequency domain resource;
  • a processing module configured to train and/or update the AI model based on the target channel matrix.
  • a terminal in a fifth aspect, includes a processor and a memory.
  • the memory stores programs or instructions that can be run on the processor.
  • the program or instructions are executed by the processor, the following implementations are implemented: The steps of the AI model processing method described in one aspect.
  • a terminal including a processor and a communication interface, wherein the processor is configured to determine a first frequency domain resource for AI model training, and to perform training on the first frequency domain resource.
  • Channel estimation obtain the target channel matrix, and perform at least one of the following:
  • the target channel matrix is sent to a network side device, and the network side device is used to train and/or update the AI model based on the target channel matrix.
  • a network side device in a seventh aspect, includes a processor and a memory.
  • the memory stores programs or instructions that can be run on the processor.
  • the program or instructions are executed by the processor.
  • a network side device including a processor and a communication interface, wherein the communication interface is used to receive a target channel matrix sent by a terminal, wherein the target channel matrix is the first frequency domain resource of the terminal.
  • the channel matrix obtained by performing channel estimation; the processor is used to train and/or update the AI model based on the target channel matrix.
  • a ninth aspect provides a communication system, including: a terminal and a network side device.
  • the terminal can be used to perform the steps of the AI model processing method described in the first aspect.
  • the network side device can be used to perform the steps of the second aspect. The steps of the AI model processing method described in this aspect.
  • a readable storage medium In a tenth aspect, a readable storage medium is provided. Programs or instructions are stored on the readable storage medium. When the programs or instructions are executed by a processor, the steps of the AI model processing method as described in the first aspect are implemented. , or implement the steps of the AI model processing method described in the second aspect.
  • a chip in an eleventh aspect, includes a processor and a communication interface.
  • the communication interface is coupled to the processor.
  • the processor is used to run programs or instructions to implement the method described in the first aspect. AI model processing method, or implement the AI model processing method as described in the second aspect.
  • a computer program/program product is provided, the computer program/program product being stored in In the storage medium, the computer program/program product is executed by at least one processor to implement the AI model processing method as described in the first aspect, or to implement the AI model processing method as described in the second aspect.
  • the terminal can perform channel estimation based on specific first frequency domain resources to obtain the target channel matrix, that is, it can train and/or update the AI model through specific frequency domain resources to avoid the terminal's
  • the training and/or updating of the AI model will occupy too many frequency domain resources, which can avoid affecting other communication services of the terminal and ensure the communication performance between the terminal and the network side equipment.
  • Figure 1 is a block diagram of a wireless communication system applicable to the embodiment of the present application.
  • Figure 2 is a flow chart of an AI model processing method provided by an embodiment of the present application.
  • Figure 3 is a flow chart of another AI model processing method provided by an embodiment of the present application.
  • Figure 4 is a structural diagram of an AI model processing device provided by an embodiment of the present application.
  • FIG. 5 is a structural diagram of another AI model processing device provided by an embodiment of the present application.
  • Figure 6 is a structural diagram of a communication device provided by an embodiment of the present application.
  • Figure 7 is a structural diagram of a terminal provided by an embodiment of the present application.
  • Figure 8 is a structural diagram of a network side device provided by an embodiment of the present application.
  • first, second, etc. in the description and claims of this application are used to distinguish similar objects and are not used to describe a specific order or sequence. It is to be understood that the terms so used are interchangeable under appropriate circumstances so that the embodiments of the present application can be practiced in sequences other than those illustrated or described herein, and that "first" and “second” are distinguished objects It is usually one type, and the number of objects is not limited.
  • the first object can be one or multiple.
  • “and/or” in the description and claims indicates at least one of the connected objects, and the character “/" generally indicates that the related objects are in an "or” relationship.
  • LTE Long Term Evolution
  • LTE-Advanced, LTE-A Long Term Evolution
  • LTE-A Long Term Evolution
  • CDMA Code Division Multiple Access
  • TDMA Time Division Multiple Access
  • FDMA Frequency Division Multiple Access
  • OFDMA Orthogonal Frequency Division Multiple Access
  • SC-FDMA Single-carrier Frequency Division Multiple Access
  • NR New Radio
  • FIG. 1 shows a block diagram of a wireless communication system to which embodiments of the present application are applicable.
  • the wireless communication system includes a terminal 11 and a network side device 12.
  • the terminal 11 can be a mobile phone, a tablet computer (Tablet Personal Computer), a laptop computer (Laptop Computer), or a notebook computer, a personal digital assistant (Personal Digital Assistant, PDA), a handheld computer, a netbook, or a super mobile personal computer.
  • Tablet Personal Computer Tablet Personal Computer
  • laptop computer laptop computer
  • PDA Personal Digital Assistant
  • PDA Personal Digital Assistant
  • UMPC ultra-mobile personal computer
  • UMPC mobile Internet device
  • MID mobile Internet device
  • augmented reality augmented reality, AR
  • VR virtual reality
  • robots wearable devices
  • Vehicle user equipment VUE
  • pedestrian terminal pedestrian terminal
  • PUE pedestrian terminal
  • smart home home equipment with wireless communication functions, such as refrigerators, TVs, washing machines or furniture, etc.
  • game consoles personal computers (personal computer, PC), teller machine or self-service machine and other terminal-side devices.
  • Wearable devices include: smart watches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart bracelets, smart rings, smart necklaces, smart anklets) bracelets, smart anklets, etc.), smart wristbands, smart clothing, etc.
  • the network side equipment 12 may include access network equipment or core network equipment, where the access network equipment may also be called wireless access network equipment, radio access network (Radio Access Network, RAN), radio access network function or wireless access network unit.
  • Access network equipment can include base stations, Wireless Local Area Network (WLAN) access points or Wireless Fidelity (WiFi) nodes, etc.
  • WLAN Wireless Local Area Network
  • WiFi Wireless Fidelity
  • the base station can be called Node B, Evolved Node B (eNB), Access Point Entry point, Base Transceiver Station (BTS), radio base station, radio transceiver, Basic Service Set (BSS), Extended Service Set (ESS), home B node, home evolution Type B node, Transmitting Receiving Point (TRP) or some other appropriate terminology in the field, as long as the same technical effect is achieved, the base station is not limited to specific technical terms. It should be noted that in this application, In the embodiment, only the base station in the NR system is taken as an example for introduction, and the specific type of the base station is not limited.
  • Channel State Information (Channel State Information, CSI) is crucial to channel capacity.
  • the transmitter can optimize signal transmission based on CSI to better match the channel status.
  • CQI Channel Quality Indicator
  • MCS Modulation and Coding Scheme
  • PMI Precoding Matrix Indicator
  • Eigen beamforming Eigen beamforming
  • the base station sends a channel state information reference signal on certain time-frequency resources in a certain time slot.
  • CSI-RS Channel State Information Reference Signal
  • the terminal performs channel estimation based on CSI-RS, calculates the channel information in this slot, and feeds back the PMI to the base station through the codebook.
  • the base station combines the channel information based on the codebook information fed back by the terminal. , before the next CSI report, the base station uses this to perform data precoding and multi-user scheduling.
  • the terminal can change the PMI reported on each subband to report PMI based on delay. Since the channels in the delay domain are more concentrated, PMI with fewer delays can approximately represent the PMI of all subbands. That is, the delay field information will be compressed before reporting.
  • the base station can precode the CSI-RS in advance and send the coded CSI-RS to the terminal. What the terminal sees is the channel corresponding to the coded CSI-RS. The terminal only needs to Just select several ports with greater strength among the indicated ports and report the coefficients corresponding to these ports.
  • terminals and network-side devices can use neural networks or machine learning methods to transmit channel information.
  • the terminal uses an AI model to compress and encode the channel information
  • the base station uses a corresponding AI model to decode the compressed content, thereby restoring the channel information.
  • the AI model for decoding on the base station side and the AI model for encoding on the terminal side need to be jointly trained to achieve a reasonable degree of matching.
  • the AI model is data-driven, it requires a large amount of data to complete the feature learning process. This also causes the training process of the AI model to occupy a large amount of time-frequency resources, causing an impact on other communication services and affecting communication between communication devices. performance.
  • embodiments of this application propose an AI model processing method.
  • Figure 2 is a flow chart of an AI model processing method provided by an embodiment of the present application. As shown in Figure 2, the method includes the following steps:
  • Step 201 The terminal determines the first frequency domain resource used for AI model training.
  • the first frequency domain resource may be a frequency domain resource indicated by the network side device.
  • the network side device indicates a specific subband or a specific physical resource block (Physical Resource Block, PRB) for AI model training.
  • PRB Physical Resource Block
  • the terminal may determine the first frequency domain resource used for AI model training based on instructions from the network side device.
  • the first frequency domain resource may also be agreed in a protocol, agreed in advance, or set in advance.
  • the first frequency domain resource may be determined by the terminal itself.
  • the terminal may select certain subbands or PRBs for AI model training.
  • Step 202 The terminal performs channel estimation on the first frequency domain resource, obtains a target channel matrix, and performs at least one of the following:
  • the target channel matrix is sent to a network side device, and the network side device is used to train and/or update the AI model based on the target channel matrix.
  • the terminal after the terminal determines the first frequency domain resource for AI model training, it performs channel estimation on the first frequency domain resource to obtain a target channel matrix, and performs AI optimization based on the target channel matrix.
  • Modeling Training and/or updating and/or the terminal may send the target channel matrix to the network side device, and the network side device trains and/or updates the AI model based on the received target channel matrix.
  • the terminal and/or the network side device can perform channel estimation based on the specific first frequency domain resource to obtain the target channel matrix, that is, the AI model can be trained and/or updated through the specific frequency domain resource.
  • the terminal may train and/or update all AI models on the terminal side based on the channel matrix, or it may also target specific AI models are trained and/or updated.
  • the channel matrix obtained by the terminal performing channel estimation on the first frequency domain resource is the target channel matrix; or, in other embodiments, the channel matrix obtained by the terminal performing channel estimation on the first frequency domain resource is A part of the target channel matrix. This situation will be illustrated in subsequent embodiments and will not be described in detail here.
  • the terminal may train and/or update the matching first AI model and the second AI model based on the target channel matrix, where the first AI model and the second AI model are two paired AIs.
  • Model for example, the first AI model is suitable for the terminal side and is used to encode channel information.
  • the terminal sends the encoding information output by the first AI model to the network side device, and the network side device is used to encode the encoding information through the second AI model. Decode to recover channel information.
  • the terminal After completing the training and/or updating of the first AI model and the second AI model, the terminal sends the trained and/or updated second AI model to the network side device, and then the network side device can directly use the training and/or Updated second AI model.
  • the terminal may also send the target channel matrix to the network side device, and the network side device trains and/or updates the paired first AI model and the second AI model based on the target channel matrix, and then trains and/or updates the paired first AI model and the second AI model. /or the updated first AI model is sent to the terminal, and the terminal can directly use the trained and/or updated first AI model.
  • the terminal determines the first frequency domain resource used for AI model training, including:
  • the terminal receives first indication information sent by the network side device, where the first indication information is used to indicate the first frequency domain resource for AI model training;
  • the terminal determines the first frequency domain resource based on the first indication information.
  • the network side device indicates that subband A is used for AI model training through the first indication information, and then the terminal can use subband A as a frequency domain resource for AI model training based on the first indication information.
  • Band A performs channel estimation to obtain the channel matrix, and trains the AI model based on the channel matrix and/or updates the existing AI model.
  • the terminal can also send the channel matrix obtained by channel estimation in subband A to the network side.
  • the network side device performs training and/or updating of the AI model based on the channel matrix.
  • the first frequency domain resource used for AI model training is indicated through the network side device, so as to avoid the training and/or updating of the AI model from occupying too many frequency domain resources, thereby ensuring the smooth communication between the terminal and the network side device. communication performance.
  • the first indication information is also used to indicate a first time domain resource.
  • the terminal performs channel estimation on the first frequency domain resource to obtain a target channel matrix, including:
  • the terminal performs channel estimation on the first frequency domain resource based on the first time domain resource and obtains a target channel moment. Array.
  • the network side device indicates the first frequency domain resource and the first time domain resource through the first indication information.
  • the first frequency domain resource is subband A and the first time domain resource is slots 0 to 4, then the terminal uses the first indication information to indicate the first frequency domain resource and the first time domain resource.
  • the first instruction information is to perform channel estimation in subband A in slots 0 to 4 to obtain the target channel matrix for AI model training and/or updating.
  • the network side device indicates the first time domain resource and the first frequency domain resource for AI model training, which in turn enables the terminal to perform channel estimation on the specific time domain resource and frequency domain resource to obtain user information.
  • the target channel matrix for AI model training can avoid AI model training from occupying too many time domain resources and frequency domain resources and affecting other services of the terminal and network side equipment to ensure communication between the terminal and network side equipment. performance.
  • the first frequency domain resource corresponds to one frequency domain resource block
  • the terminal performs channel estimation on the first frequency domain resource to obtain a target channel matrix, including:
  • the terminal receives the first signaling sent by the network side device
  • the terminal activates the frequency domain resource block configured by the network side device based on the first signaling, and performs channel estimation on the frequency domain resource block to obtain a target channel matrix.
  • frequency domain resource blocks is at least one, that is, there may be multiple frequency domain resource blocks used for AI model training, or there may be multiple first frequency domain resources.
  • the network side device can pre-configure PRBs 0-3 and PRBs 28-31 as frequency domain resource blocks for AI model training, that is, one frequency domain resource block.
  • the network side device is pre-configured with two frequency domain resource blocks (that is, PRB No. 0-3 is a frequency domain resource block, and PRB No. 28-31 is a frequency domain resource block); when the network side device sends a request to the terminal Send the first signaling for activating the frequency domain resource block, then after receiving the first signaling, the terminal activates PRB Nos. 0-3 and PRB Nos. 28-31 and performs channel estimation to obtain the results for AI Target channel matrix for model training and/or updating.
  • the terminal will activate the preconfigured frequency domain resource block for channel estimation only when it receives the signaling to activate the frequency domain resource block. If the frequency domain resource block is not activated, it can still Used for data transmission of other services to effectively and fully utilize frequency domain resources.
  • the terminal activates the frequency domain resource block configured by the network side device based on the first signaling, And perform channel estimation on the frequency domain resource block to obtain the target channel matrix, including:
  • the terminal activates the frequency domain resource block configured by the network side device based on the first signaling, and the terminal determines based on the first indication information that the first time domain resource does not need to be based on the frequency domain resource block.
  • the terminal performs channel estimation in the frequency domain resource block based on the first time domain resource to obtain a target channel matrix;
  • the method also includes:
  • the terminal determines that the second time domain resource requires data transmission based on the frequency domain resource block based on the first indication information, the terminal performs data transmission based on the frequency domain resource block, and the third time domain resource requires data transmission based on the frequency domain resource block.
  • the second time domain resource is a time domain resource other than the first time domain resource.
  • the network side device simultaneously indicates the first time domain resource and the first frequency through the first indication information, domain resources (that is, frequency domain resource blocks), that is, indicating that the terminal needs to perform channel estimation in the first time domain resource, then it can be determined that the terminal does not need to transmit data through the frequency domain resource block in the first time domain resource.
  • domain resources that is, frequency domain resource blocks
  • the terminal after the terminal receives the first signaling from the network side device, it activates the frequency domain resource block preconfigured by the network side device for AI model training based on the first signaling, and at the first time Channel estimation is performed on the frequency domain resource block at a time corresponding to the domain resources to obtain a target channel matrix for training and/or updating the AI model.
  • the first indication information indicates that the frequency domain resource block is used for AI model training in the first time domain resource, and then the frequency domain resource block is used in other time domain resources except the first time domain resource (that is, the second time domain resource).
  • Domain resources do not need to be used for AI model training, that is, they can be used for data transmission, and the terminal transmits data through the frequency domain resource block at the moment corresponding to the second time domain resource. In this way, the frequency domain resource blocks can be fully utilized and the waste of frequency domain resources can be avoided.
  • the method may also include:
  • the terminal When the terminal obtains a target service with a priority greater than the channel estimate, the terminal performs data transmission of the target service in the frequency domain resource block based on the first time domain resource.
  • the network side device instructs to perform channel estimation on PRB No. 0-3 at time A to obtain the target channel matrix for AI model training and/or update. If the terminal obtains a target service with a higher priority, then The terminal may perform data transmission of the target service in PRB No. 0-3 at time A, thereby ensuring that the terminal can perform target services with higher priority first.
  • the terminal may also include:
  • the terminal reports to the network side device the number of frequency domain resource blocks that can perform channel estimation simultaneously and the size of the frequency domain resource block.
  • the terminal may simultaneously report to the network side device the number of frequency domain resource blocks that can perform channel estimation simultaneously and the size of the frequency domain resource blocks.
  • the frequency domain resource blocks configured by the network side device for AI model training are PRBs 0-3 and PRBs 28-31
  • the number of frequency domain resource blocks that the terminal needs to report is 2, and the number of frequency domain resource blocks is 2.
  • the size is 4 PRBs. In this way, the network side device can accurately know the number and size of frequency domain resource blocks actually used by the terminal for AI model training.
  • the terminal performs channel estimation on the first frequency domain resource to obtain a target channel matrix, including at least one of the following:
  • the terminal divides the first frequency domain resource into at least One frequency domain resource block, the terminal performs channel estimation on the at least one frequency domain resource block to obtain the target channel matrix;
  • the terminal When the resource quantity of the target frequency domain resource is greater than the resource quantity of the first frequency domain resource, the terminal performs channel estimation on the first frequency domain resource, obtains a first channel matrix, and performs channel estimation based on the first frequency domain resource.
  • the first channel matrix determines the second channel matrix, wherein the target channel matrix includes the first channel matrix and the second channel matrix.
  • the network side device may be pre-configured with the AI model and the resource quantity of the target frequency domain resources. response, that is, the AI model needs to occupy the number of resources of the target frequency domain resources for training and/or use.
  • the network side device pre-configures the working range of the AI model to 4 PRBs, that is, the training and use of the AI model requires 4 PRBs; if the network side device indicates that the AI model training range is The first frequency domain resource is 8 PRBs, then the terminal can divide the first 4 PRBs of these 8 PRBs into one frequency domain resource block, and divide the last 4 PRBs into another frequency domain resource block, and then the terminal can divide these two PRBs into one frequency domain resource block.
  • Channel estimation is performed on each frequency domain resource block respectively to obtain a target channel matrix used for AI model training and/or updating.
  • the frequency domain resources in the divided frequency domain resource blocks should be continuous or at fixed intervals. For example, assuming that the above target frequency domain resources are 4 PRBs and the first frequency domain resources are 8 PRBs, the terminal can use the first 4 PRBs of the 8 PRBs (that is, PRBs 1, 2, 3, and 4). Divide it into a frequency domain resource block, or divide PRB No. 2, 3, 4, and 5 into a frequency domain resource block, etc.
  • the network side device pre-configures the working range of the AI model to 4 PRBs, that is, the training and use of the AI model requires occupying 4 PRBs; if the network side device indicates that it is used for AI
  • the first frequency domain resource for model training is 2 PRBs, that is, the number of target frequency domain resources is greater than the number of first frequency domain resources; in this case, the terminal performs channel estimation on the 2 PRBs indicated by the network side device. , to obtain the first channel matrix, and then copy the first channel matrix to obtain the second channel matrix, that is, the first channel matrix and the second channel matrix are the same channel matrix, and the terminal combines the first channel matrix and the second channel matrix.
  • the two channel matrices form a target channel matrix, and the AI model is trained and/or updated based on the target channel matrix.
  • the terminal can flexibly obtain frequency domain resources for AI model training through corresponding processing methods. .
  • the method further includes:
  • the terminal sends a first request to the network side device
  • the terminal receives a first response signal sent by the network side device in response to the first request;
  • the terminal sends the AI model to the network side device or a model part of the AI model that is suitable for the network side device based on the first response signal.
  • the terminal may send a first request for updating the AI model to the network side device; the network side device receives the first request. After a request, it is decided whether the AI model needs to be updated and the first response signal is sent to the terminal.
  • the network side device determines that the AI model needs to be updated, the network side device sends a first response signal to the terminal that the AI network model needs to be updated. Based on the first response signal, the terminal sends the trained and/or The updated AI model, or the model part of the AI model that is suitable for the network side device is sent to the network side device. In this way, the terminal needs to be confirmed by the network side device before sending the AI model to the network side device to avoid occupying too many resources of the network side device.
  • the AI model trained and/or updated by the terminal includes a first AI model applicable to the terminal side and a second AI model applicable to the network side device, that is, the two AI models are combined on the terminal side. trained and/or updated, the model part of the AI model sent by the terminal to the network side device that is suitable for the network side device is the second AI model.
  • the terminal sends the AI model to the network side device or a model part of the AI model that is suitable for the network side device based on the first response signal, including:
  • the terminal receives second indication information sent by the network side device, where the second indication information is used to indicate a second frequency domain resource;
  • the terminal Based on the first response signal, the terminal sends the AI model to the network side device at the frequency domain position corresponding to the second frequency domain resource or sends the AI model suitable for the network side device. model part.
  • the network side device when the network side device determines that the AI model needs to be updated, the network side device sends a first response signal to the terminal that the AI network model needs to be updated, and at the same time sends the second indication information for indicating the second frequency domain resource to the terminal. , and then the terminal sends the updated and/or trained AI model or the model part of the AI model that is suitable for the network side device to the network side device through the frequency domain position corresponding to the second frequency domain resource (for example, the above-mentioned Second AI model). In this way, the terminal can send the AI model through a specific frequency domain location indicated by the network side device, thereby avoiding occupying the frequency domain resources of other services of the terminal.
  • the method also includes:
  • the terminal receives third indication information sent by the network side device, where the third indication information is used to indicate the first moment;
  • the terminal applies the trained and/or updated AI model at the first moment based on the third indication information.
  • the terminal uses the trained and/or updated AI model at the first moment based on the first moment indicated by the network side device.
  • the AI model instructs the terminal through the network side device to use the AI model at a specific moment, so as to avoid the use of the AI model from occupying too many time domain resources on the terminal side.
  • the terminal can send the target channel matrix to the network side device, that is, send the complete target channel matrix, so that the network
  • the side device can train and/or update the AI model of the network side device based on the target channel matrix.
  • sending the target channel matrix to the network side device includes:
  • the terminal maps the target channel matrix into a codebook, and sends the codebook to the network side device.
  • the terminal reports the target channel matrix to the network side device through a codebook.
  • the method further includes:
  • the terminal uses the trained and/or updated AI model in a third frequency domain resource.
  • the terminal performs signaling in subband A. For channel estimation, after obtaining the channel matrix corresponding to subband A, the terminal trains and/or updates the AI model based on the channel matrix obtained for subband A, and then applies the trained and/or updated AI model to subband B. In this way, the terminal can train and use the model separately through different frequency domain resources.
  • the method may also include:
  • the terminal performs channel estimation on the third frequency domain resource, obtains a channel matrix, and updates the trained or updated AI model based on the obtained channel matrix.
  • the first frequency domain resource is subband A
  • the third frequency domain resource is subband B.
  • the terminal performs channel estimation in subband A and completes training and/or updating of the AI model based on the obtained channel matrix.
  • the terminal can also perform channel estimation in sub-band B, obtain the corresponding channel matrix, and update the trained and/or updated AI model based on the channel matrix obtained in sub-band B. In this way, the terminal can further update the AI model based on different frequency domain resources to further modify the AI model, making the terminal more flexible in training and updating the AI model.
  • the third frequency domain resource is a frequency domain resource indicated by the network side device.
  • the network side device indicates the third frequency domain resource through indication information.
  • the network side device may also indicate the third frequency domain resource and the third time domain resource at the same time, that is, the terminal can obtain the channel matrix based on the channel estimation of the third frequency domain resource at the time corresponding to the third time domain resource. , update the trained and/or updated AI model.
  • the third frequency domain resource may be multiple frequency domain resource blocks.
  • the first frequency domain resource and the third frequency domain resource cover the entire carrier bandwidth.
  • the terminal uses the trained and/or updated AI model in the third frequency domain resource, including:
  • the terminal divides the third frequency domain resource into at least one frequency domain resource block based on the first frequency domain resource, and the terminal uses the trained and/or updated data in the at least one frequency domain resource block.
  • the AI model is a model that describes the AI model.
  • the terminal can divide these 8 PRBs into two frequency domain resource blocks, of which the first 4 PRBs are is a frequency domain resource block, and the last 4 PRBs are a frequency domain resource block.
  • the terminal uses the trained and/or updated AI model in these two frequency domain resource blocks, and the terminal uses these two frequency domain resources.
  • the AI model used by the blocks is the same.
  • the AI model is an AI model trained based on the first frequency domain resource. In this way, the frequency domain resources used for AI model training and the frequency domain resources used during use are matched to ensure the use of the AI model.
  • the first frequency domain resource matches the cell corresponding to the terminal. That is to say, each cell corresponding to the terminal has a matching first frequency domain resource.
  • the matching first frequency domain resource of each cell may be the same or different.
  • the first frequency domain resource matched by the cell may be a network side device indication or configuration.
  • the cell corresponding to the terminal may refer to a cell in which the terminal can transmit and receive data.
  • the method further includes:
  • the terminal sends the CSI-RS to the network side device in a code division manner.
  • the terminal can send the CSI-RS to the network side device in a code division manner.
  • the terminal performs channel estimation on the first frequency domain resource to obtain a target channel matrix, including:
  • the terminal performs channel estimation on the first frequency domain resources matched in each cell to obtain a target channel matrix.
  • the terminal can perform channel estimation on the first frequency domain resources matched in each cell to obtain a target channel matrix.
  • the target channel The matrix may be a set of channel matrices obtained by performing channel estimation on the matched first frequency domain resources of each cell.
  • the terminal performs channel estimation on the first frequency domain resources matched in each cell to obtain a target channel matrix, including:
  • the terminal performs channel estimation on the first frequency domain resources matched in each cell to obtain the first channel matrix corresponding to each cell;
  • the terminal obtains a channel matrix set based on the first channel matrix corresponding to each cell, and the target channel matrix is the channel matrix set.
  • the terminal performs channel estimation on the matched first frequency domain resources of cell A and cell B respectively, and obtains the first channel matrix A corresponding to cell A and the first channel matrix B corresponding to cell B. Then the target channel matrix also includes The first channel matrix A and the first channel matrix B.
  • the terminal when it performs AI model training based on the target channel matrix, it may traverse all first channel matrices in the channel matrix set.
  • the cell corresponding to the terminal may also be matched with time domain resources, and the time domain resources matched by each cell may be the same or different.
  • the target channel matrix input to the AI model may be preprocessed.
  • the target channel matrix may be oversampled and then input into the AI model to train and/or update the AI model.
  • the preprocessing method may be protocol reservation or network side device configuration, and the preprocessing method may also be matched with an AI model, that is, different AI models may match different or the same preprocessing method.
  • Figure 3 is a flow chart of another AI model processing method provided by an embodiment of the present application. As shown in Figure 3, the method includes the following steps:
  • Step 301 The network side device receives a target channel matrix sent by the terminal, where the target channel matrix is a channel matrix obtained by the terminal performing channel estimation on the first frequency domain resource;
  • Step 302 The network side device trains and/or updates the AI model based on the target channel matrix.
  • the terminal after the terminal determines the first frequency domain resource for AI model training, it performs channel estimation on the first frequency domain resource to obtain a target channel matrix, and can send the target channel matrix To the network side device, the network side device trains and/or updates the AI model based on the received target channel matrix. In addition, the terminal may also train and/or update the AI model on the terminal side based on the target channel matrix.
  • the terminal and/or the network side device can perform channel estimation based on the specific first frequency domain resource to obtain the target channel matrix, that is, the AI model can be trained and/or updated through the specific frequency domain resource. ,avoid This eliminates the need for terminals and/or network-side equipment to occupy excessive frequency domain resources for training and/or updating AI models, thereby avoiding the impact on other communication services of terminals and/or network-side equipment, ensuring that terminals and network-side Communication performance between devices.
  • the method before the network side device receives the target channel matrix sent by the terminal, the method further includes:
  • the network side device sends first indication information to the terminal, where the first indication information is used to indicate the first frequency domain resource for AI model training.
  • the first indication information is also used to indicate a first time domain resource
  • the target channel matrix is obtained by the terminal performing channel estimation on the first frequency domain resource based on the first time domain resource. channel matrix.
  • the first frequency domain resource corresponds to one frequency domain resource block.
  • the method further includes:
  • the network side device sends the first signaling to the terminal, and the terminal is configured to activate the frequency domain resource block configured by the network side device based on the first signaling, and perform channel estimation on the frequency domain resource block, Obtain the target channel matrix.
  • the method also includes:
  • the network side device receives the number of frequency domain resource blocks that can perform channel estimation simultaneously and the size of the frequency domain resource block reported by the terminal.
  • the method also includes:
  • the network side device receives the first request sent by the terminal
  • the network side device sends a first response signal to the terminal based on the first request
  • the network side device receives the trained and/or updated AI model sent by the terminal in response to the first response signal or sends a model among the trained and/or updated AI models that is suitable for the network side device. part;
  • the terminal trains and/or updates the AI model based on the target channel matrix to obtain the trained and/or updated AI model.
  • the network side device receives the trained and/or updated AI model sent by the terminal in response to the first response signal or sends the trained and/or updated AI model that is suitable for the network.
  • the model part of the side device includes:
  • the network side device sends second indication information to the terminal, where the second indication information is used to indicate a second frequency domain resource;
  • the network side device receives the trained and/or updated AI model sent by the terminal through the frequency domain position corresponding to the second frequency domain resource or sends the trained and/or updated AI model that is suitable for the The model part of the network side device.
  • the method also includes:
  • the network side device sends third indication information to the terminal
  • the third indication information is used to indicate a first moment
  • the terminal is used to apply the trained and/or updated AI model at the first moment.
  • the network side device receives the target channel matrix sent by the terminal, including:
  • the network side device receives the codebook sent by the terminal, where the codebook is obtained by mapping the target channel matrix by the terminal.
  • the method also includes:
  • the network side device indicates the third frequency domain resource to the terminal, and the terminal is used to perform channel estimation on the third frequency domain resource, obtain a channel matrix, and perform training or updated AI based on the obtained channel matrix.
  • the model is updated.
  • the first frequency domain resource matches the cell corresponding to the terminal.
  • the method further includes:
  • the network side device receives the CSI-RS sent by the terminal in a code division manner.
  • the AI model processing method provided by the embodiments of the present application is executed by a network-side device and corresponds to the AI model processing method executed by the terminal.
  • the relevant concepts and specific implementation processes involved in the embodiments of the present application can be Referring to the description in the method embodiment described above in Figure 2, to avoid repetition, the details will not be described again here.
  • the execution subject may be an AI model processing device.
  • an AI model processing device executing an AI model processing method is used as an example to illustrate the AI model processing device provided by the embodiment of the present application.
  • the AI model processing device 400 includes:
  • Determination module 401 used to determine the first frequency domain resource used for AI model training
  • Execution module 402 is configured to perform channel estimation on the first frequency domain resource, obtain a target channel matrix, and perform at least one of the following:
  • the target channel matrix is sent to a network side device, and the network side device is used to train and/or update the AI model based on the target channel matrix.
  • the determination module 401 is also used to:
  • the first frequency domain resource is determined based on the first indication information.
  • the first indication information is also used to indicate the first time domain resource
  • the execution module 402 is also used to:
  • the execution module 402 is further configured to perform at least one of the following:
  • the resource quantity of the target frequency domain resource is less than or equal to the resource quantity of the first frequency domain resource, divide the first frequency domain resource into at least one frequency domain in units of the target frequency domain resource. Resource blocks, perform channel estimation on the at least one frequency domain resource block to obtain the target channel matrix;
  • channel estimation is performed on the first frequency domain resources to obtain a first channel matrix, and based on the first channel Matrix determines the second channel A matrix, wherein the target channel matrix includes the first channel matrix and the second channel matrix.
  • the first frequency domain resource corresponds to one frequency domain resource block
  • the execution module 402 is also used to:
  • the frequency domain resource block configured by the network side device is activated based on the first signaling, and channel estimation is performed on the frequency domain resource block to obtain a target channel matrix.
  • the first indication information is also used to indicate the first time domain resource
  • the execution module 402 is also used to:
  • the frequency domain resource block configured by the network side device is activated based on the first signaling, and when the device determines based on the first indication information that the first time domain resource does not need to perform data based on the frequency domain resource block.
  • the device determines that the second time domain resource requires data transmission based on the frequency domain resource block based on the first indication information
  • data transmission is performed based on the frequency domain resource block
  • the second time domain resource is transmitted based on the frequency domain resource block.
  • the resources are time domain resources other than the first time domain resource.
  • the device also includes:
  • a transmission module configured to perform data on the target service in the frequency domain resource block based on the first time domain resource when the device obtains a target service with a priority greater than the priority of the channel estimate. transmission.
  • the device also includes:
  • a reporting module is configured to report to the network side device the number of frequency domain resource blocks capable of simultaneous channel estimation and the size of the frequency domain resource blocks.
  • the device also includes:
  • a sending module used to send the first request to the network side device
  • a receiving module configured to receive a first response signal sent by the network side device in response to the first request
  • the sending module is further configured to: based on the first response signal, send the trained and/or updated AI model to the network side device or send the trained and/or updated AI model. in the model part of the network side device.
  • the receiving module is also used to:
  • the sending module is further configured to: based on the first response signal, send the AI model to the network side device at the frequency domain position corresponding to the second frequency domain resource or send the AI model applicable to The model part of the network side device.
  • the receiving module is also used to:
  • the device further includes an application module, configured to apply the trained and/or updated AI model at the first moment based on the third indication information.
  • execution module 402 is also used to:
  • the device is also used for:
  • the device also includes:
  • An update module configured to perform channel estimation on the third frequency domain resource, obtain a channel matrix, and update the trained or updated AI model based on the obtained channel matrix.
  • the third frequency domain resource is a frequency domain resource indicated by the network side device.
  • the first frequency domain resource and the third frequency domain resource cover the entire carrier bandwidth.
  • the device is also used for:
  • the third frequency domain resource is divided into at least one frequency domain resource block in units of the first frequency domain resource, and the trained and/or updated AI model is used in the at least one frequency domain resource block.
  • the first frequency domain resource matches a cell corresponding to the device.
  • the device further includes:
  • a sending module configured to send the CSI-RS to the network side device in a code division manner.
  • the execution module 402 is further configured to:
  • Channel estimation is performed on the matched first frequency domain resources in each cell to obtain a target channel matrix.
  • execution module 402 is also used to:
  • a channel matrix set is obtained based on the first channel matrix corresponding to each cell, and the target channel matrix is the channel matrix set.
  • the device can perform channel estimation based on specific first frequency domain resources to obtain the target channel matrix, that is, it can perform training and/or updating of the AI model through specific frequency domain resources to avoid
  • the device will occupy excessive frequency domain resources for training and/or updating the AI model, which can avoid affecting other communication services of the device and ensure communication performance between the device and network-side equipment.
  • the AI model processing device 400 in the embodiment of the present application may be an electronic device, such as an electronic device with an operating system, or may be a component in the electronic device, such as an integrated circuit or chip.
  • the electronic device may be a terminal or other devices other than the terminal.
  • terminals may include but are not limited to the types of terminals 11 listed above, and other devices may be servers, network attached storage (Network Attached Storage, NAS), etc., which are not specifically limited in the embodiment of this application.
  • the AI model processing device 400 provided by the embodiment of the present application can implement each process implemented by the terminal in the method embodiment described in Figure 2, and achieve the same technical effect. To avoid duplication, the details will not be described here.
  • FIG. 5 is a structural diagram of another AI model processing device provided by an embodiment of the present application. As shown in Figure 5, the AI model processing device 500 includes:
  • the receiving module 501 is configured to receive a target channel matrix sent by the terminal, where the target channel matrix is a channel matrix obtained by the terminal performing channel estimation on the first frequency domain resource;
  • the processing module 502 is used to train and/or update the AI model based on the target channel matrix.
  • the device also includes:
  • a sending module configured to send first indication information to the terminal, where the first indication information is used to indicate a first frequency domain resource for AI model training.
  • the first indication information is also used to indicate a first time domain resource
  • the target channel matrix is obtained by the terminal performing channel estimation on the first frequency domain resource based on the first time domain resource. channel matrix.
  • the first frequency domain resource corresponds to one frequency domain resource block
  • the device further includes:
  • a sending module configured to send first signaling to a terminal, where the terminal is configured to activate the frequency domain resource block configured by the device based on the first signaling, and perform channel estimation on the frequency domain resource block, Obtain the target channel matrix.
  • the receiving module 501 is also used to:
  • the receiving module 501 is also used to: receive the first request sent by the terminal;
  • the device further includes a sending module configured to send a first response signal to the terminal based on the first request;
  • the receiving module 501 is also configured to: receive the trained and/or updated AI model sent by the terminal in response to the first response signal or send the trained and/or updated AI model suitable for the network.
  • the terminal trains and/or updates the AI model based on the target channel matrix to obtain the trained and/or updated AI model.
  • the device also includes:
  • a sending module configured to send second indication information to the terminal, where the second indication information is used to indicate a second frequency domain resource
  • the receiving module 501 is also configured to: receive the trained and/or updated AI model sent by the terminal through the frequency domain position corresponding to the second frequency domain resource or send the trained and/or updated AI model. Model part applicable to the network side device.
  • the sending module is also used to:
  • the third indication information is used to indicate a first moment
  • the terminal is used to apply the trained and/or updated AI model at the first moment.
  • the receiving module 501 is also used to:
  • the device also includes:
  • An indication module configured to indicate a third frequency domain resource to the terminal, and the terminal is configured to perform processing on the third frequency domain resource. Perform channel estimation, obtain the channel matrix, and update the trained or updated AI model based on the obtained channel matrix.
  • the first frequency domain resource matches the cell corresponding to the terminal.
  • the receiving module 501 further Used for:
  • the terminal can perform channel estimation based on a specific first frequency domain resource to obtain a target channel matrix, and send the target channel matrix to the device, and then the device performs evaluation of the AI model based on the target channel matrix.
  • Training and/or updating prevents the device from occupying excessive frequency domain resources for training and/or updating the AI model, thereby avoiding impact on other communication services of the device.
  • the AI model processing device 500 provided by the embodiment of this application can implement each process implemented by the network side device in the method embodiment described in Figure 3, and achieve the same technical effect. To avoid duplication, the details will not be described here.
  • this embodiment of the present application also provides a communication device 600, which includes a processor 601 and a memory 602.
  • the memory 602 stores programs or instructions that can be run on the processor 601, such as , when the communication device 600 is a terminal, when the program or instruction is executed by the processor 601, each step of the method embodiment described in Figure 2 is implemented, and the same technical effect can be achieved.
  • the communication device 600 is a network-side device, when the program or instruction is executed by the processor 601, each step of the method embodiment described in FIG. 3 is implemented, and the same technical effect can be achieved. To avoid duplication, the details will not be described here.
  • An embodiment of the present application also provides a terminal, including a processor and a communication interface.
  • the processor is used to determine a first frequency domain resource for AI model training, and to perform channel estimation on the first frequency domain resource to obtain The target channel matrix, and perform at least one of the following: training and/or updating the AI model based on the target channel matrix; sending the target channel matrix to a network side device, and the network side device is configured to use the target channel matrix based on the target channel matrix. Matrices train and/or update AI models.
  • This terminal embodiment corresponds to the above-mentioned terminal-side method embodiment. Each implementation process and implementation manner of the above-mentioned method embodiment can be applied to this terminal embodiment, and can achieve the same technical effect.
  • FIG. 7 is a schematic diagram of the hardware structure of a terminal that implements an embodiment of the present application.
  • the terminal 700 includes but is not limited to: a radio frequency unit 701, a network module 702, an audio output unit 703, an input unit 704, a sensor 705, a display unit 706, a user input unit 707, an interface unit 708, a memory 709, a processor 710, etc. At least some parts.
  • the terminal 700 may also include a power supply (such as a battery) that supplies power to various components.
  • the power supply may be logically connected to the processor 710 through a power management system, thereby managing charging, discharging, and power consumption through the power management system. Management and other functions.
  • the terminal structure shown in FIG. 7 does not constitute a limitation on the terminal.
  • the terminal may include more or fewer components than shown in the figure, or some components may be combined or arranged differently, which will not be described again here.
  • the input unit 704 may include a graphics processing unit (GPU) 7041 and a microphone 7042.
  • the graphics processor 7041 is responsible for the image capture device (GPU) in the video capture mode or the image capture mode. Process the image data of still pictures or videos obtained by cameras (such as cameras).
  • the display unit 706 may include a display panel 7061, which may be configured in the form of a liquid crystal display, an organic light emitting diode, or the like.
  • the user input unit 707 includes a touch panel 7071 and at least one of other input devices 7072 A sort of. Touch panel 7071, also called touch screen.
  • the touch panel 7071 may include two parts: a touch detection device and a touch controller.
  • Other input devices 7072 may include but are not limited to physical keyboards, function keys (such as volume control keys, switch keys, etc.), trackballs, mice, and joysticks, which will not be described again here.
  • the radio frequency unit 701 after receiving downlink data from the network side device, can transmit it to the processor 710 for processing; in addition, the radio frequency unit 701 can send uplink data to the network side device.
  • the radio frequency unit 701 includes, but is not limited to, an antenna, amplifier, transceiver, coupler, low noise amplifier, duplexer, etc.
  • Memory 709 may be used to store software programs or instructions as well as various data.
  • the memory 709 may mainly include a first storage area for storing programs or instructions and a second storage area for storing data, wherein the first storage area may store an operating system, an application program or instructions required for at least one function (such as a sound playback function, Image playback function, etc.) etc.
  • memory 709 may include volatile memory or non-volatile memory, or memory 709 may include both volatile and non-volatile memory.
  • non-volatile memory can be read-only memory (Read-Only Memory, ROM), programmable read-only memory (Programmable ROM, PROM), erasable programmable read-only memory (Erasable PROM, EPROM), electrically removable memory. Erase programmable read-only memory (Electrically EPROM, EEPROM) or flash memory.
  • Volatile memory can be random access memory (Random Access Memory, RAM), static random access memory (Static RAM, SRAM), dynamic random access memory (Dynamic RAM, DRAM), synchronous dynamic random access memory (Synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (Double Data Rate SDRAM, DDRSDRAM), enhanced synchronous dynamic random access memory (Enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (Synch link DRAM) , SLDRAM) and direct memory bus random access memory (Direct Rambus RAM, DRRAM).
  • RAM Random Access Memory
  • SRAM static random access memory
  • DRAM dynamic random access memory
  • DRAM synchronous dynamic random access memory
  • SDRAM double data rate synchronous dynamic random access memory
  • Double Data Rate SDRAM Double Data Rate SDRAM
  • DDRSDRAM double data rate synchronous dynamic random access memory
  • Enhanced SDRAM, ESDRAM enhanced synchronous dynamic random access memory
  • Synch link DRAM synchronous link dynamic random access memory
  • SLDRAM direct memory bus
  • the processor 710 may include one or more processing units; optionally, the processor 710 integrates an application processor and a modem processor, where the application processor mainly handles operations related to the operating system, user interface, application programs, etc., Modem processors mainly process wireless communication signals, such as baseband processors. It can be understood that the above-mentioned modem processor may not be integrated into the processor 710.
  • processor 710 is used for:
  • Perform channel estimation on the first frequency domain resource obtain a target channel matrix, and perform at least one of the following:
  • the target channel matrix is sent to a network side device, and the network side device is used to train and/or update the AI model based on the target channel matrix.
  • the terminal can perform channel estimation based on specific first frequency domain resources to obtain the target channel matrix, that is, it can train and/or update the AI model through specific frequency domain resources to avoid the terminal's influence on the AI.
  • the training and/or updating of the model will occupy too many frequency domain resources, which can avoid affecting other communication services of the terminal and ensure the communication performance between the terminal and the network side equipment.
  • An embodiment of the present application also provides a network side device, including a processor and a communication interface, and the communication interface is used to connect Receive a target channel matrix sent by the terminal, wherein the target channel matrix is a channel matrix obtained by the terminal performing channel estimation on the first frequency domain resource; the processor is configured to train the AI model based on the target channel matrix and/or renew.
  • This network-side device embodiment corresponds to the above-mentioned network-side device method embodiment.
  • Each implementation process and implementation manner of the above-mentioned method embodiment can be applied to this network-side device embodiment, and can achieve the same technical effect.
  • the embodiment of the present application also provides a network side device.
  • the network side device 800 includes: an antenna 81 , a radio frequency device 82 , a baseband device 83 , a processor 84 and a memory 85 .
  • the antenna 81 is connected to the radio frequency device 82 .
  • the radio frequency device 82 receives information through the antenna 81 and sends the received information to the baseband device 83 for processing.
  • the baseband device 83 processes the information to be sent and sends it to the radio frequency device 82.
  • the radio frequency device 82 processes the received information and then sends it out through the antenna 81.
  • the method performed by the network side device in the above embodiment can be implemented in the baseband device 83, which includes a baseband processor.
  • the baseband device 83 may include, for example, at least one baseband board on which multiple chips are disposed, as shown in FIG. Program to perform the network device operations shown in the above method embodiments.
  • the network side device may also include a network interface 86, which is, for example, a common public radio interface (CPRI).
  • a network interface 86 which is, for example, a common public radio interface (CPRI).
  • CPRI common public radio interface
  • the network side device 800 in the embodiment of the present application also includes: instructions or programs stored in the memory 85 and executable on the processor 84.
  • the processor 84 calls the instructions or programs in the memory 85 to execute the various operations shown in Figure 5. The method of module execution and achieving the same technical effect will not be described in detail here to avoid duplication.
  • Embodiments of the present application also provide a readable storage medium.
  • Programs or instructions are stored on the readable storage medium.
  • the program or instructions are executed by a processor, each process of the method embodiment described in Figure 2 is implemented, or Each process of the method embodiment described in Figure 3 above can achieve the same technical effect. To avoid repetition, it will not be described again here.
  • the processor is the processor in the terminal described in the above embodiment.
  • the readable storage medium includes computer readable storage media, such as computer read-only memory ROM, random access memory RAM, magnetic disk or optical disk, etc.
  • An embodiment of the present application further provides a chip.
  • the chip includes a processor and a communication interface.
  • the communication interface is coupled to the processor.
  • the processor is used to run programs or instructions to implement the method described in Figure 2.
  • chips mentioned in the embodiments of this application may also be called system-on-chip, system-on-a-chip, system-on-chip or system-on-chip, etc.
  • Embodiments of the present application further provide a computer program/program product.
  • the computer program/program product is stored in a storage medium.
  • the computer program/program product is executed by at least one processor to implement the method described in Figure 2 above.
  • Each process of the embodiment, or each process of implementing the above method embodiment described in Figure 3, can achieve the same technical effect. To avoid repetition, it will not be described again here.
  • Embodiments of the present application also provide a communication system, including: a terminal and a network side device.
  • the terminal can be used to perform the steps of the method as shown in Figure 2.
  • the network side device can be used to perform the steps of the method as shown in Figure 3 above. Method steps.
  • the methods of the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is better. implementation.
  • the technical solution of the present application can be embodied in the form of a computer software product that is essentially or contributes to related technologies.
  • the computer software product is stored in a storage medium (such as ROM/RAM, disk, CD), including several instructions to cause a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in various embodiments of this application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Power Engineering (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请公开了一种AI模型处理方法、装置、终端及网络侧设备，属于通信技术领域，本申请实施例的AI模型处理方法包括：终端确定用于进行AI模型训练的第一频域资源；所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：基于所述目标信道矩阵对AI模型进行训练和/或更新；向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。

Description

AI模型处理方法、装置、终端及网络侧设备
相关申请的交叉引用
本申请主张在2022年07月06日在中国提交的中国专利申请No.202210800718.3的优先权，其全部内容通过引用包含于此。
技术领域
本申请属于通信技术领域，具体涉及一种AI模型处理方法、装置、终端及网络侧设备。
背景技术
人工智能(Artificial Intelligence，AI)是研究和开发用于模拟、延伸、扩展人的智能的理论、方法、技术及应用系统的一门新的技术科学，受到人们的广泛关注，针对AI的应用也越来越广泛。目前，AI模型已经能够应用在通信技术领域，由于AI模型是数据驱动，需要大量的数据才能完成特征学习的过程，这也就造成AI模型的训练过程会占据大量的时频资源，导致对其他通信业务造成影响，影响通信设备间的通信性能。
发明内容
本申请实施例提供一种AI模型处理方法、装置、终端及网络侧设备，能够解决相关技术中。
第一方面，提供了一种AI模型处理方法，包括：
终端确定用于进行AI模型训练的第一频域资源；
所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：
基于所述目标信道矩阵对AI模型进行训练和/或更新；
向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
第二方面，提供了一种AI模型处理方法，包括：
网络侧设备接收终端发送的目标信道矩阵，其中，所述目标信道矩阵为终端在第一频域资源进行信道估计获得的信道矩阵；
所述网络侧设备基于所述目标信道矩阵对AI模型进行训练和/或更新。
第三方面，提供了一种AI模型处理装置，包括：
确定模块，用于确定用于进行AI模型训练的第一频域资源；
执行模块，用于在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：
基于所述目标信道矩阵对AI模型进行训练和/或更新；
向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
第四方面，提供了一种AI模型处理装置，包括：
接收模块，用于接收终端发送的目标信道矩阵，其中，所述目标信道矩阵为终端在第一频域资源进行信道估计获得的信道矩阵；
处理模块，用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
第五方面，提供了一种终端，该终端包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如第一方面所述的AI模型处理方法的步骤。
第六方面，提供了一种终端，包括处理器及通信接口，其中，所述处理器用于确定用于进行AI模型训练的第一频域资源，以及用于在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：
基于所述目标信道矩阵对AI模型进行训练和/或更新；
向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
第七方面，提供了一种网络侧设备，该网络侧设备包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如第二方面所述的AI模型处理方法的步骤。
第八方面，提供了一种网络侧设备，包括处理器及通信接口，其中，所述通信接口用于接收终端发送的目标信道矩阵，其中，所述目标信道矩阵为终端在第一频域资源进行信道估计获得的信道矩阵；所述处理器用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
第九方面，提供了一种通信系统，包括：终端及网络侧设备，所述终端可用于执行如第一方面所述的AI模型处理方法的步骤，所述网络侧设备可用于执行如第二方面所述的AI模型处理方法的步骤。
第十方面，提供了一种可读存储介质，所述可读存储介质上存储有程序或指令，所述程序或指令被处理器执行时实现如第一方面所述的AI模型处理方法的步骤，或者实现如第二方面所述的AI模型处理方法的步骤。
第十一方面，提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如第一方面所述的AI模型处理方法，或实现如第二方面所述的AI模型处理方法。
第十二方面，提供了一种计算机程序/程序产品，所述计算机程序/程序产品被存储在 存储介质中，所述计算机程序/程序产品被至少一个处理器执行以实现如第一方面所述的AI模型处理方法，或实现如第二方面所述的AI模型处理方法。
在本申请实施例中，终端能够基于特定的第一频域资源来进行信道估计得到目标信道矩阵，也即能够通过特定的频域资源来进行对AI模型的训练和/或更新，避免终端对于AI模型的训练和/或更新会占用过多的频域资源，也就能够避免对终端的其他通信业务造成影响，确保终端和网络侧设备之间的通信性能。
附图说明
图1是本申请实施例可应用的一种无线通信系统的框图；
图2是本申请实施例提供的一种AI模型处理方法的流程图；
图3是本申请实施例提供的另一种AI模型处理方法的流程图；
图4是本申请实施例提供的一种AI模型处理装置的结构图；
图5是本申请实施例提供的另一种AI模型处理装置的结构图；
图6是本申请实施例提供的一种通信设备的结构图；
图7是本申请实施例提供的一种终端的结构图；
图8是本申请实施例提供的一种网络侧设备的结构图。
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清楚描述，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所获得的所有其他实施例，都属于本申请保护的范围。
本申请的说明书和权利要求书中的术语“第一”、“第二”等是用于区别类似的对象，而不用于描述特定的顺序或先后次序。应该理解这样使用的术语在适当情况下可以互换，以便本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施，且“第一”、“第二”所区别的对象通常为一类，并不限定对象的个数，例如第一对象可以是一个，也可以是多个。此外，说明书以及权利要求中“和/或”表示所连接对象的至少其中之一，字符“/”一般表示前后关联对象是一种“或”的关系。
值得指出的是，本申请实施例所描述的技术不限于长期演进型(Long Term Evolution，LTE)/LTE的演进(LTE-Advanced，LTE-A)系统，还可用于其他无线通信系统，诸如码分多址(Code Division Multiple Access，CDMA)、时分多址(Time Division Multiple Access，TDMA)、频分多址(Frequency Division Multiple Access，FDMA)、正交频分多址(Orthogonal Frequency Division Multiple Access，OFDMA)、单载波频分多址(Single-carrier Frequency Division Multiple Access，SC-FDMA)和其他系统。本申请实施例中的术语“系统”和“网络”常被可互换地使用，所描述的技术既可用于以上提及的系统和无线电技术，也可用于 其他系统和无线电技术。以下描述出于示例目的描述了新空口(New Radio，NR)系统，并且在以下大部分描述中使用NR术语，但是这些技术也可应用于NR系统应用以外的应用，如第6代(6th Generation，6G)通信系统。
图1示出本申请实施例可应用的一种无线通信系统的框图。无线通信系统包括终端11和网络侧设备12。其中，终端11可以是手机、平板电脑(Tablet Personal Computer)、膝上型电脑(Laptop Computer)或称为笔记本电脑、个人数字助理(Personal Digital Assistant，PDA)、掌上电脑、上网本、超级移动个人计算机(ultra-mobile personal computer，UMPC)、移动上网装置(Mobile Internet Device，MID)、增强现实(augmented reality，AR)/虚拟现实(virtual reality，VR)设备、机器人、可穿戴式设备(Wearable Device)、车载设备(Vehicle User Equipment，VUE)、行人终端(Pedestrian User Equipment，PUE)、智能家居(具有无线通信功能的家居设备，如冰箱、电视、洗衣机或者家具等)、游戏机、个人计算机(personal computer，PC)、柜员机或者自助机等终端侧设备，可穿戴式设备包括：智能手表、智能手环、智能耳机、智能眼镜、智能首饰(智能手镯、智能手链、智能戒指、智能项链、智能脚镯、智能脚链等)、智能腕带、智能服装等。需要说明的是，在本申请实施例并不限定终端11的具体类型。网络侧设备12可以包括接入网设备或核心网设备，其中，接入网设备也可以称为无线接入网设备、无线接入网(Radio Access Network,RAN)、无线接入网功能或无线接入网单元。接入网设备可以包括基站、无线局域网(Wireless Local Area Network，WLAN)接入点或无线保真(Wireless Fidelity，WiFi)节点等，基站可被称为节点B、演进节点B(eNB)、接入点、基收发机站(Base Transceiver Station，BTS)、无线电基站、无线电收发机、基本服务集(Basic Service Set，BSS)、扩展服务集(Extended Service Set，ESS)、家用B节点、家用演进型B节点、发送接收点(Transmitting Receiving Point，TRP)或所述领域中其他某个合适的术语，只要达到相同的技术效果，所述基站不限于特定技术词汇，需要说明的是，在本申请实施例中仅以NR系统中的基站为例进行介绍，并不限定基站的具体类型。
为更好地理解本申请实施例提供的技术方案，以下对本申请实施例中可能涉及的相关概念及原理进行解释说明。
由信息论可知，准确的信道状态信息(Channel State Information，CSI)对信道容量的至关重要。尤其是对于多天线系统来讲，发送端可以根据CSI优化信号的发送，使其更加匹配信道的状态。例如：信道质量指示(Channel Quality Indicator，CQI)可以用来选择合适的调制编码方案(Modulation and Coding Scheme，MCS)实现链路自适应；预编码矩阵指示(Precoding Matrix Indicator，PMI)可以用来实现特征波束成形(eigen beamforming)从而最大化接收信号的强度，或者用来抑制干扰(如小区间干扰、多用户之间干扰等)。因此，自从多天线技术(multi-input multi-output，MIMO)被提出以来，CSI获取一直都是研究热点。
通常，基站在在某个时隙(slot)的某些时频资源上发送信道状态信息参考信号 (Channel State Information Reference Signal，CSI-RS)，终端根据CSI-RS进行信道估计，计算这个slot上的信道信息，通过码本将PMI反馈给基站，基站根据终端反馈的码本信息组合出信道信息，在下一次CSI上报之前，基站以此进行数据预编码及多用户调度。
为了进一步减少CSI反馈开销，终端可以将每个子带上报PMI改成按照延迟(delay)上报PMI，由于delay域的信道更集中，用更少的delay的PMI就可以近似表示全部子带的PMI，即将delay域信息压缩之后再上报。
同样，为了减少开销，基站可以事先对CSI-RS进行预编码，将编码后的CSI-RS发送给终端，终端看到的是经过编码之后的CSI-RS对应的信道，终端只需要在网络侧指示的端口中选择若干个强度较大的端口，并上报这些端口对应的系数即可。
进一步地，为了更好地压缩信道信息，终端和网络侧设备可以使用神经网络或机器学习的方法进行信道信息的传递。
具体地，在终端通过AI模型对信道信息进行压缩编码，在基站通过对应的AI模型对压缩后的内容进行解码，从而恢复信道信息。此时基站侧进行解码的AI模型和终端侧进行编码的AI模型需要联合训练，达到合理的匹配度。但是由于AI模型是数据驱动，需要大量的数据才能完成特征学习的过程，这也就造成AI模型的训练过程会占据大量的时频资源，导致对其他通信业务造成影响，影响通信设备间的通信性能。针对这一情况，本申请实施例提出了一种AI模型处理方法。
下面结合附图，通过一些实施例及其应用场景对本申请实施例提供的AI模型处理方法、装置及相关设备进行详细地说明。
请参照图2，图2是本申请实施例提供的一种AI模型处理方法的流程图，如图2所示，所述方法包括以下步骤：
步骤201、终端确定用于进行AI模型训练的第一频域资源。
可选地，所述第一频域资源可以是网络侧设备指示的频域资源，例如网络侧设备指示特定的子带或者特定的物理资源块(Physical Resource Block，PRB)用于进行AI模型训练，则终端可以是基于网络侧设备的指示来确定用于进行AI模型训练的第一频域资源。
或者，所述第一频域资源也可以是协议约定或者是预先约定或者是预先设定的。
又或者，所述第一频域资源也可以是终端自行确定的。例如，终端可以是自行选择某些子带或PRB用于进行AI模型训练。
步骤202、所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：
基于所述目标信道矩阵对AI模型进行训练和/或更新；
向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
本申请实施例中，终端在确定用于进行AI模型训练的第一频域资源后，则在所述第一频域资源进行信道估计，得到目标信道矩阵，并基于所述目标信道矩阵对AI模型进行 训练和/或更新，和/或，终端可以是将所述目标信道矩阵发送给网络侧设备，网络侧设备基于接收到的所述目标信道矩阵对AI模型进行训练和/或更新。这样，也就使得终端和/或网络侧设备能够基于特定的第一频域资源来进行信道估计得到目标信道矩阵，也即能够通过特定的频域资源来进行对AI模型的训练和/或更新，避免终端和/或网络侧设备对于AI模型的训练和/或更新会占用过多的频域资源，也就能够避免对终端和/或网络侧设备的其他通信业务造成影响，确保终端和网络侧设备之间的通信性能。
需要说明的是，终端在基于第一频域资源进行信道估计，获得目标信道矩阵后，可以是基于所述信道矩阵对终端侧所有的AI模型进行训练和/或更新，或者也可以是针对特定的AI模型进行训练和/或更新。
在一些实施例中，终端在第一频域资源进行信道估计得到的信道矩阵即是目标信道矩阵；或者，在另一些实施例中，终端在第一频域资源进行信道估计得到的信道矩阵为目标信道矩阵的一部分，后续实施例中会对此种情况进行举例说明，此处不做具体赘述。
可选地，终端可以是基于所述目标信道矩阵对匹配的第一AI模型和第二AI模型进行训练和/或更新，其中所述第一AI模型和第二AI模型为配对的两个AI模型，例如第一AI模型适用于终端侧，用于对信道信息进行编码，终端将第一AI模型输出的编码信息发送给网络侧设备，网络侧设备用于通过第二AI模型对编码信息进行解码以恢复信道信息。终端在完成对第一AI模型和第二AI模型的训练和/或更新后，将训练和/或更新后的第二AI模型发送给网络侧设备，进而网络侧设备可以直接使用训练和/或更新后的第二AI模型。
或者，终端也可以是将所述目标信道矩阵发送给网络侧设备，网络侧设备基于所述目标信道矩阵对配对的第一AI模型和第二AI模型进行训练和/或更新，而后将训练和/或更新后的第一AI模型发送给终端，进而终端可以直接使用训练和/或更新后的第一AI模型。
本申请实施例中，所述终端确定用于进行AI模型训练的第一频域资源，包括：
终端接收网络侧设备发送的第一指示信息，所述第一指示信息用于指示进行AI模型训练的第一频域资源；
所述终端基于所述第一指示信息确定所述第一频域资源。
例如，网络侧设备通过第一指示信息指示子带A用于进行AI模型训练，进而终端也就能够基于所述第一指示信息将子带A作为进行AI模型训练的频域资源，终端在子带A进行信道估计来获得信道矩阵，基于所述信道矩阵训练AI模型和/或对已有的AI模型进行更新，或者终端也可以是将在子带A信道估计得到的信道矩阵发送给网络侧设备，网络侧设备基于所述信道矩阵来进行AI模型的训练和/或更新。
这样，通过网络侧设备来指示用于进行AI模型训练的第一频域资源，以避免AI模型的训练和/或更新会占用过多的频域资源，进而以确保终端和网络侧设备之间的通信性能。
可选地，所述第一指示信息还用于指示第一时域资源，所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，包括：
所述终端基于所述第一时域资源在所述第一频域资源进行信道估计，获得目标信道矩 阵。
例如，网络侧设备通过第一指示信息指示了第一频域资源和第一时域资源，如第一频域资源为子带A，第一时域资源为slot0～4，则终端基于所述第一指示信息，在slot0～4在子带A进行信道估计，来获得进行AI模型训练和/或更新的目标信道矩阵。
这样，网络侧设备指示了用于进行AI模型训练的第一时域资源和第一频域资源，进而也就使得终端能够在特定的时域资源和频域资源来进行信道估计，以得到用于AI模型训练的目标信道矩阵，从而也就能够避免AI模型训练占用过多的时域资源和频域资源而影响终端和网络侧设备的其他业务，以确保终端和网络侧设备之间的通信性能。
可选地，所述第一频域资源对应一个频域资源块，所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，包括：
所述终端接收网络侧设备发送的第一信令；
所述终端基于所述第一信令激活网络侧设备配置的所述频域资源块，并在所述频域资源块进行信道估计，获得目标信道矩阵。
需要说明的是，所述频域资源块的数量为至少一个，也即用于AI模型训练的频域资源块可以是多个，或者说第一频域资源可以是多个。
例如，整个带宽上可以是有52个PRB，网络侧设备可以是预先配置0-3号PRB以及28-31号PRB为用于进行AI模型训练的频域资源块，也即一个频域资源块包括4个PRB，网络侧设备预先配置了两个频域资源块(即0-3号PRB为一个频域资源块，28-31号PRB为一个频域资源块)；当网络侧设备向终端发送用于激活所述频域资源块的第一信令，则终端在接收到所述第一信令后激活0-3号PRB以及28-31号PRB并进行信道估计，得到用于进行AI模型训练和/或更新的目标信道矩阵。这样，终端也就只有在接收到激活频域资源块的信令的情况下，才会激活预先配置的频域资源块以进行信道估计，频域资源块在未被激活的情况下则还可以用于进行其他业务的数据传输，从而有效和充分地利用频域资源。
可选地，在网络侧设备发送的第一指示信息还用于指示第一时域资源的情况下，所述终端基于所述第一信令激活网络侧设备配置的所述频域资源块，并在所述频域资源块进行信道估计，获得目标信道矩阵，包括：
所述终端基于所述第一信令激活网络侧设备配置的所述频域资源块，在所述终端基于所述第一指示信息确定在第一时域资源不需要基于所述频域资源块进行数据传输的情况下，所述终端基于所述第一时域资源在所述频域资源块进行信道估计，获得目标信道矩阵；
这种情况下，所述方法还包括：
在所述终端基于所述第一指示信息确定在第二时域资源需要基于所述频域资源块进行数据传输的情况下，所述终端基于所述频域资源块进行数据传输，所述第二时域资源为除所述第一时域资源之外的时域资源。
本申请实施例中，若网络侧设备通过第一指示信息同时指示了第一时域资源和第一频 域资源(也即频域资源块)，也即指示终端需要在第一时域资源进行信道估计，那么也就能够确定在第一时域资源终端无需通过频域资源块来进行数据传输，这种情况下，当终端接收到网络侧设备第一信令后，基于所述第一信令激活网络侧设备预先配置的用于进行AI模型训练的频域资源块，并在所述第一时域资源对应的时刻在所述频域资源块进行信道估计，以获得用于训练和/或更新AI模型的目标信道矩阵。
进一步地，第一指示信息指示了频域资源块在第一时域资源用于进行AI模型训练，进而频域资源块在除第一时域资源以外的其他时域资源(也即第二时域资源)也就无需用来进行AI模型训练，也即可以用于进行数据传输，则终端在第二时域资源对应的时刻通过频域资源块进行数据传输。这样，也就使得频域资源块能够得到充分利用，避免频域资源的浪费。
可选地，所述方法还可以包括：
在所述终端获取到优先级大于所述信道估计的目标业务的情况下，所述终端基于所述第一时域资源在所述频域资源块进行所述目标业务的数据传输。
例如，网络侧设备指示了在时刻A在0-3号PRB进行信道估计，以获得用于进行AI模型训练和/或更新的目标信道矩阵，若终端获取到了优先级更高的目标业务，则终端可以是在时刻A在0-3号PRB进行目标业务的数据传输，从而以保障终端能够优先执行优先级更高的目标业务。
可选地，所述终端还可以包括：
所述终端向网络侧设备上报能够同时进行信道估计的所述频域资源块的数量以及所述频域资源块的大小。
需要说明的是，终端可以是在进行能力上报的时候，同时向网络侧设备上报能够同时进行信道估计的频域资源块的数量以及频域资源块的大小。
例如，若网络侧设备配置用于进行AI模型训练的频域资源块为0-3号PRB以及28-31号PRB，则终端需要上报的频域资源块的数量为2，频域资源块的大小为4个PRB。这样，也就使得网络侧设备能够准确获知终端实际用于进行AI模型训练的频域资源块的数量和大小。
可选地，在所述AI模型与目标频域资源的资源数量对应的情况下，所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，包括如下至少一项：
在所述目标频域资源的资源数量小于或等于所述第一频域资源的资源数量的情况下，所述终端以所述目标频域资源为单位将所述第一频域资源划分为至少一个频域资源块，所述终端在所述至少一个频域资源块进行信道估计，获得目标信道矩阵；
在所述目标频域资源的资源数量大于所述第一频域资源的资源数量的情况下，所述终端在所述第一频域资源进行信道估计，获得第一信道矩阵，并基于所述第一信道矩阵确定第二信道矩阵，其中，所述目标信道矩阵包括所述第一信道矩阵和第二信道矩阵。
本申请实施例中，网络侧设备可以是预先配置AI模型与目标频域资源的资源数量对 应，也即AI模型需要占用目标频域资源的资源数量来进行训练和/或使用。
例如，在一种实施方式中，网络侧设备预先配置AI模型的工作范围为4个PRB，也即AI模型的训练和使用需要占用4个PRB；若网络侧设备指示用于进行AI模型训练的第一频域资源为8个PRB，则终端可以是将这8个PRB的前4个PRB划分为一个频域资源块，后4个PRB划分为另一个频域资源块，然后终端在这两个频域资源块分别进行信道估计，得到用于进行AI模型训练和/或更新的目标信道矩阵。
需要说明的是，终端在以目标频域资源为单位对第一频域资源进行划分时，划分后的频域资源块中的频域资源应该是连续的或者固定间隔的。例如，以上述目标频域资源为4个PRB，第一频域资源为8个PRB为例，终端可以是将8个PRB的前4个PRB(也即1、2、3、4号PRB)划分为一个频域资源块，或者也可以是将2、3、4、5号PRB划分为一个频域资源块，等。
可选地，在另一种实施方式中，网络侧设备预先配置AI模型的工作范围为4个PRB，也即AI模型的训练和使用需要占用4个PRB；若网络侧设备指示用于进行AI模型训练的第一频域资源为2个PRB，也即目标频域资源的资源数量大于第一频域资源的资源数量；这种情况下，终端在网络侧设备指示的2个PRB进行信道估计，得到第一信道矩阵，然后可以是对第一信道矩阵进行复制操作，得到第二信道矩阵，也即第一信道矩阵和第二信道矩阵为相同的信道矩阵，终端将第一信道矩阵和第二信道矩阵组成目标信道矩阵，基于所述目标信道矩阵来进行AI模型的训练和/或更新。
本申请实施例中，对于目标频域资源的资源数量无论是大于还是小于或等于第一频域资源的资源数量，终端都能够灵活地通过相应的处理方式来得到进行AI模型训练的频域资源。
可选地，在所述终端基于所述目标信道矩阵对所述AI模型进行训练和/或更新之后，所述方法还包括：
所述终端向网络侧设备发送第一请求；
所述终端接收所述网络侧设备响应于所述第一请求发送的第一响应信号；
所述终端基于所述第一响应信号，向所述网络侧设备发送所述AI模型或者发送所述AI模型中的适用于所述网络侧设备的模型部分。
本申请实施例中，终端在基于目标信道矩阵完成对AI模型的训练和/或更新之后，终端可以向网络侧设备发送用于申请更新AI模型的第一请求；网络侧设备接收到所述第一请求后，决定是否需要更新AI模型，并向终端发送第一响应信号。
可选地，若网络侧设备确定需要更新AI模型，网络侧设备向终端发送需要更新AI网络模型的第一响应信号，终端基于所述第一响应信号，向网络侧设备发送已经训练和/或更新后的AI模型，或者向网络侧设备发送所述AI模型中的适用于所述网络侧设备的模型部分。这样，也就使得终端需要基于网络侧设备的确认才能向网络侧设备发送AI模型，以避免占用网络侧设备过多的资源。
需要说明地，若终端训练和/或更新的AI模型包括分别适用于终端侧的第一AI模型和适用于网络侧设备的第二AI模型，也即这两个AI模型是在终端侧进行联合训练和/或更新的，则终端向网络侧设备发送的所述AI模型中的适用于所述网络侧设备的模型部分也即第二AI模型。
可选地，所述终端基于所述第一响应信号，向所述网络侧设备发送所述AI模型或者发送所述AI模型中的适用于所述网络侧设备的模型部分，包括：
所述终端接收所述网络侧设备发送的第二指示信息，所述第二指示信息用于指示第二频域资源；
所述终端基于所述第一响应信号，在所述第二频域资源对应的频域位置向所述网络侧设备发送所述AI模型或者发送所述AI模型中的适用于所述网络侧设备的模型部分。
本申请实施例中，当网络侧设备确定需要更新AI模型，网络侧设备向终端发送需要更新AI网络模型的第一响应信号，同时向终端发送用于指示第二频域资源的第二指示信息，进而终端通所述第二频域资源对应的频域位置向网络侧设备发送更新和/或训练后的AI模型或所述AI模型中的适用于所述网络侧设备的模型部分(例如上述第二AI模型)。这样，也就使得终端能够通过网络侧设备指示的特定的频域位置来发送AI模型，进而以避免占用终端其他业务的频域资源。
可选地，所述方法还包括：
所述终端接收所述网络侧设备发送的第三指示信息，所述第三指示信息用于指示第一时刻；
所述终端基于所述第三指示信息，在所述第一时刻应用训练和/或更新后的所述AI模型。
本申请实施例中，终端在基于目标信道矩阵完成对AI模型的训练和/或更新后，终端基于网络侧设备指示的第一时刻，在第一时刻来使用训练和/或更新后的所述AI模型，进而也就通过网络侧设备来指示终端使用AI模型的特定时刻，以避免AI模型的使用会占用终端侧过多的时域资源。
可选地，在终端基于第一频域资源进行信道估计，获得目标信道矩阵后，终端可以向所述网络侧设备发送所述目标信道矩阵，也即发送完整的目标信道矩阵，进而以使得网络侧设备能够基于所述目标信道矩阵来训练和/或更新网络侧设备的AI模型。
或者，所述向所述网络侧设备发送所述目标信道矩阵，包括：
所述终端将所述目标信道矩阵映射为码本，向所述网络侧设备发送所述码本。
也就是说，终端通过码本的方式来向网络侧设备上报所述目标信道矩阵。
本申请实施例中，在所述终端基于所述目标信道矩阵对所述AI模型进行训练和/或更新之后，所述方法还包括：
所述终端在第三频域资源使用训练和/或更新后的所述AI模型。
例如，所述第一频域资源为子带A，第三频域资源为子带B，则终端在子带A进行信 道估计，获得子带A对应的信道矩阵后，终端基于子带A获得的信道矩阵进行AI模型的训练和/或更新，然后在子带B应用该训练和/或更新后的AI模型。这样，终端也就能够通过不同的频域资源来分别进行模型的训练和使用。
可选地，所述方法还可以包括：
所述终端在所述第三频域资源进行信道估计，获得信道矩阵，并基于获得的信道矩阵对训练或更新后的所述AI模型进行更新。
例如，所述第一频域资源为子带A，第三频域资源为子带B，终端在子带A进行信道估计，并基于获得的信道矩阵完成对AI模型的训练和/或更新后，终端还可以在子带B进行信道估计，获得对应的信道矩阵，并基于子带B获得的信道矩阵对训练和/或更新后的所述AI模型进行更新。这样，终端也就能够基于不同的频域资源来进一步对AI模型进行更新，以进一步修正AI模型，使得终端对于AI模型的训练和更新更灵活。
可选地，所述第三频域资源为网络侧设备指示的频域资源，例如网络侧设备通过指示信息来指示第三频域资源。进一步地，网络侧设备还可以是同时指示第三频域资源和第三时域资源，也即终端能够在第三时域资源对应的时刻基于对第三频域资源进行信道估计获得的信道矩阵，对训练和/或更新后的所述AI模型进行更新。
可选地，所述第三频域资源可以有多个，例如第三频域资源可以是多个频域资源块。
可选地，所述第一频域资源和所述第三频域资源覆盖整个载波带宽。
可选地，所述终端在第三频域资源使用训练和/或更新后的所述AI模型，包括：
所述终端以所述第一频域资源为单位将所述第三频域资源划分为至少一个频域资源块，所述终端在所述至少一个频域资源块使用训练和/或更新后的所述AI模型。
例如，第一频域资源为4个PRB，网络侧设备指示的第三频域资源为8个PRB，则终端可以是将这8个PRB划分为两个频域资源块，其中前4个PRB为一个频域资源块，后4个PRB为一个频域资源块，终端在这两个频域资源块来使用训练和/或更新后的所述AI模型，且终端在这两个频域资源块使用的AI模型相同，例如该AI模型是基于第一频域资源训练得到的AI模型。这样，也就使得AI模型训练用的频域资源和使用时的频域资源是匹配的，以保障AI模型的使用。
本申请实施例中，所述第一频域资源与所述终端对应的小区匹配。也就是说，终端对应的每个小区都有与其匹配的第一频域资源，每个小区匹配的第一频域资源可以是相同的，也可以是不同的。小区匹配的第一频域资源可以是网络侧设备指示或配置。其中，所述终端对应的小区，可以是指终端能够进行数据收发的小区。
可选地，在所述终端对应的至少两个小区匹配的所述第一频域资源相同的情况下，所述方法还包括：
所述终端通过码分的方式向所述网络侧设备发送CSI-RS。
例如，终端对应的每个小区所匹配的第一频域资源都相同，这种情况下，终端可以通过码分的方式向网络侧设备发送CSI-RS。
可选地，在所述终端对应的每个小区匹配的所述第一频域资源不同的情况下，所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，包括：
所述终端在所述每个小区匹配的所述第一频域资源分别进行信道估计，获得目标信道矩阵。
该实施方式中，终端对应的每个小区所匹配的第一频域资源不同，则终端可以是在每个小区匹配的第一频域资源分别进行信道估计，以获得目标信道矩阵，该目标信道矩阵可以是每个小区匹配的第一频域资源分别进行信道估计得到的信道矩阵的集合。
可选地，所述终端在所述每个小区匹配的所述第一频域资源分别进行信道估计，获得目标信道矩阵，包括：
所述终端在所述每个小区匹配的所述第一频域资源分别进行信道估计，获得每个小区对应的第一信道矩阵；
所述终端基于所述每个小区对应的第一信道矩阵得到信道矩阵集合，所述目标信道矩阵为所述信道矩阵集合。
例如，终端在小区A和小区B各自匹配的第一频域资源分别进行信道估计，得到小区A对应的第一信道矩阵A和小区B对应的第一信道矩阵B，则目标信道矩阵也即包括第一信道矩阵A和第一信道矩阵B。
可选地，终端在基于目标信道矩阵进行AI模型的训练时，可以是遍历所述信道矩阵集合中所有的第一信道矩阵。
需要说明的是，终端对应的小区还可以是与时域资源匹配，每个小区匹配的时域资源可以是相同或者不同。
本申请实施例中，输入到AI模型的目标信道矩阵可以是经过预处理的，例如可以是将目标信道矩阵经过过采样处理后再输入AI模型，以对AI模型进行训练和/或更新。可选地，所述预处理方式可以是协议预定或网络侧设备配置，所述预处理方式还可以是与AI模型匹配，也即不同的AI模型可以是匹配不同或相同的预处理方式。
请参照图3，图3是本申请实施例提供的另一种AI模型处理方法的流程图，如图3所示，所述方法包括以下步骤：
步骤301、网络侧设备接收终端发送的目标信道矩阵，其中，所述目标信道矩阵为终端在第一频域资源进行信道估计获得的信道矩阵；
步骤302、所述网络侧设备基于所述目标信道矩阵对AI模型进行训练和/或更新。
本申请实施例中，终端在确定用于进行AI模型训练的第一频域资源后，则在所述第一频域资源进行信道估计，得到目标信道矩阵，并能够将所述目标信道矩阵发送给网络侧设备，网络侧设备基于接收到的所述目标信道矩阵对AI模型进行训练和/或更新。另外，终端也可以是基于所述目标信道矩阵来对终端侧的AI模型进行训练和/或更新。
这样，也就使得终端和/或网络侧设备能够基于特定的第一频域资源来进行信道估计得到目标信道矩阵，也即能够通过特定的频域资源来进行对AI模型的训练和/或更新，避 免终端和/或网络侧设备对于AI模型的训练和/或更新会占用过多的频域资源，也就能够避免对终端和/或网络侧设备的其他通信业务造成影响，确保终端和网络侧设备之间的通信性能。
可选地，所述网络侧设备接收终端发送的目标信道矩阵之前，所述方法还包括：
所述网络侧设备向终端发送第一指示信息，所述第一指示信息用于指示进行AI模型训练的第一频域资源。
可选地，所述第一指示信息还用于指示第一时域资源，所述目标信道矩阵为所述终端基于所述第一时域资源在所述第一频域资源进行信道估计获得的信道矩阵。
可选地，所述第一频域资源对应一个频域资源块，所述网络侧设备接收终端发送的目标信道矩阵之前，所述方法还包括：
所述网络侧设备向终端发送第一信令，所述终端用于基于所述第一信令激活网络侧设备配置的所述频域资源块，并在所述频域资源块进行信道估计，获得所述目标信道矩阵。
可选地，所述方法还包括：
所述网络侧设备接收终端上报的能够同时进行信道估计的所述频域资源块的数量以及所述频域资源块的大小。
可选地，所述方法还包括：
所述网络侧设备接收终端发送的第一请求；
所述网络侧设备基于所述第一请求向所述终端发送第一响应信号；
所述网络侧设备接收所述终端响应于所述第一响应信号发送的训练和/或更新后的AI模型或者发送训练和/或更新后的AI模型中的适用于所述网络侧设备的模型部分；
其中，终端基于所述目标信道矩阵对AI模型进行训练和/或更新，以获得所述训练和/或更新后的AI模型。
可选地，所述网络侧设备接收所述终端响应于所述第一响应信号发送的训练和/或更新后的AI模型或者发送训练和/或更新后的AI模型中的适用于所述网络侧设备的模型部分，包括：
所述网络侧设备向所述终端发送第二指示信息，所述第二指示信息用于指示第二频域资源；
所述网络侧设备接收所述终端通过所述第二频域资源对应的频域位置发送的训练和/或更新后的AI模型或者发送训练和/或更新后的AI模型中的适用于所述网络侧设备的模型部分。
可选地，所述方法还包括：
所述网络侧设备向所述终端发送第三指示信息；
其中，所述第三指示信息用于指示第一时刻，所述终端用于在所述第一时刻应用所述训练和/或更新后的AI模型。
可选地，所述网络侧设备接收终端发送的目标信道矩阵，包括：
所述网络侧设备接收终端发送的码本，所述码本为所述终端将所述目标信道矩阵进行映射得到。
可选地，所述方法还包括：
所述网络侧设备向所述终端指示第三频域资源，所述终端用于在所述第三频域资源进行信道估计，获得信道矩阵，并基于获得的信道矩阵对训练或更新后的AI模型进行更新。
可选地，所述第一频域资源与所述终端对应的小区匹配，在所述终端对应的至少两个小区匹配的所述第一频域资源相同的情况下，所述方法还包括：
所述网络侧设备接收所述终端通过码分的方式发送的CSI-RS。
需要说明地，本申请实施例所提供的AI模型处理方法，执行主体为网络侧设备，与上述终端执行的AI模型处理方法相对应，本申请实施例所涉及的相关概念及具体实现过程可以是参照上述图2所述方法实施例中的描述，为避免重复，此处不再赘述。
本申请实施例提供的AI模型处理方法，执行主体可以为AI模型处理装置。本申请实施例中以AI模型处理装置执行AI模型处理方法为例，说明本申请实施例提供的AI模型处理装置。
请参照图4，图4是本申请实施例提供的一种AI模型处理装置的结构图，如图4所示，所述AI模型处理装置400包括：
确定模块401，用于确定用于进行AI模型训练的第一频域资源；
执行模块402，用于在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：
基于所述目标信道矩阵对AI模型进行训练和/或更新；
向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
可选地，所述确定模块401还用于：
接收网络侧设备发送的第一指示信息，所述第一指示信息用于指示进行AI模型训练的第一频域资源；
基于所述第一指示信息确定所述第一频域资源。
可选地，所述第一指示信息还用于指示第一时域资源，所述执行模块402还用于：
基于所述第一时域资源在所述第一频域资源进行信道估计，获得目标信道矩阵。
可选地，在所述AI模型与目标频域资源的资源数量对应的情况下，所述执行模块402还用于执行如下至少一项：
在所述目标频域资源的资源数量小于或等于所述第一频域资源的资源数量的情况下，以所述目标频域资源为单位将所述第一频域资源划分为至少一个频域资源块，在所述至少一个频域资源块进行信道估计，获得目标信道矩阵；
在所述目标频域资源的资源数量大于所述第一频域资源的资源数量的情况下，在所述第一频域资源进行信道估计，获得第一信道矩阵，并基于所述第一信道矩阵确定第二信道 矩阵，其中，所述目标信道矩阵包括所述第一信道矩阵和第二信道矩阵。
可选地，所述第一频域资源对应一个频域资源块，所述执行模块402还用于：
接收网络侧设备发送的第一信令；
基于所述第一信令激活网络侧设备配置的所述频域资源块，并在所述频域资源块进行信道估计，获得目标信道矩阵。
可选地，所述第一指示信息还用于指示第一时域资源，所述执行模块402还用于：
基于所述第一信令激活网络侧设备配置的所述频域资源块，并在所述装置基于所述第一指示信息确定在第一时域资源不需要基于所述频域资源块进行数据传输的情况下，基于所述第一时域资源在所述频域资源块进行信道估计，获得目标信道矩阵；
在所述装置基于所述第一指示信息确定在第二时域资源需要基于所述频域资源块进行数据传输的情况下，基于所述频域资源块进行数据传输，所述第二时域资源为除所述第一时域资源之外的时域资源。
可选地，所述装置还包括：
传输模块，用于在所述装置获取到优先级大于所述信道估计的优先级的目标业务的情况下，基于所述第一时域资源在所述频域资源块进行所述目标业务的数据传输。
可选地，所述装置还包括：
上报模块，用于向网络侧设备上报能够同时进行信道估计的所述频域资源块的数量以及所述频域资源块的大小。
可选地，所述装置还包括：
发送模块，用于向网络侧设备发送第一请求；
接收模块，用于接收所述网络侧设备响应于所述第一请求发送的第一响应信号；
所述发送模块还用于：基于所述第一响应信号，向所述网络侧设备发送训练和/或更新后的所述AI模型或者发送训练和/或更新后的所述AI模型中的适用于所述网络侧设备的模型部分。
可选地，所述接收模块还用于：
接收所述网络侧设备发送的第二指示信息，所述第二指示信息用于指示第二频域资源；
所述发送模块还用于：基于所述第一响应信号，在所述第二频域资源对应的频域位置向所述网络侧设备发送所述AI模型或者发送所述AI模型中的适用于所述网络侧设备的模型部分。
可选地，所述接收模块还用于：
接收所述网络侧设备发送的第三指示信息，所述第三指示信息用于指示第一时刻；
所述装置还包括应用模块，用于基于所述第三指示信息，在所述第一时刻应用训练和/或更新后的所述AI模型。
可选地，所述执行模块402还用于：
将所述目标信道矩阵映射为码本，向所述网络侧设备发送所述码本。
可选地，所述装置还用于：
在第三频域资源使用训练和/或更新后的所述AI模型。
可选地，所述装置还包括：
更新模块，用于在所述第三频域资源进行信道估计，获得信道矩阵，并基于获得的信道矩阵对训练或更新后的所述AI模型进行更新。
可选地，所述第三频域资源为网络侧设备指示的频域资源。
可选地，所述第一频域资源和所述第三频域资源覆盖整个载波带宽。
可选地，所述装置还用于：
以所述第一频域资源为单位将所述第三频域资源划分为至少一个频域资源块，在所述至少一个频域资源块使用训练和/或更新后的所述AI模型。
可选地，所述第一频域资源与所述装置对应的小区匹配。
可选地，在所述装置对应的至少两个小区匹配的所述第一频域资源相同的情况下，所述装置还包括：
发送模块，用于通过码分的方式向所述网络侧设备发送CSI-RS。
可选地，在所述装置对应的每个小区匹配的所述第一频域资源不同的情况下，所述执行模块402还用于：
在所述每个小区匹配的所述第一频域资源分别进行信道估计，获得目标信道矩阵。
可选地，所述执行模块402还用于：
在所述每个小区匹配的所述第一频域资源分别进行信道估计，获得每个小区对应的第一信道矩阵；
基于所述每个小区对应的第一信道矩阵得到信道矩阵集合，所述目标信道矩阵为所述信道矩阵集合。
本申请实施例中，所述装置能够基于特定的第一频域资源来进行信道估计得到目标信道矩阵，也即能够通过特定的频域资源来进行对AI模型的训练和/或更新，避免所述装置对于AI模型的训练和/或更新会占用过多的频域资源，也就能够避免对所述装置的其他通信业务造成影响，确保所述装置和网络侧设备之间的通信性能。
本申请实施例中的AI模型处理装置400可以是电子设备，例如具有操作系统的电子设备，也可以是电子设备中的部件，例如集成电路或芯片。该电子设备可以是终端，也可以为除终端之外的其他设备。示例性的，终端可以包括但不限于上述所列举的终端11的类型，其他设备可以为服务器、网络附属存储器(Network Attached Storage，NAS)等，本申请实施例不作具体限定。
本申请实施例提供的AI模型处理装置400能够实现图2所述方法实施例中终端实现的各个过程，并达到相同的技术效果，为避免重复，这里不再赘述。
请参照图5，图5是本申请实施例提供的另一种AI模型处理装置的结构图，如图5所示，所述AI模型处理装置500包括：
接收模块501，用于接收终端发送的目标信道矩阵，其中，所述目标信道矩阵为终端在第一频域资源进行信道估计获得的信道矩阵；
处理模块502，用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
可选地，所述装置还包括：
发送模块，用于向终端发送第一指示信息，所述第一指示信息用于指示进行AI模型训练的第一频域资源。
可选地，所述第一指示信息还用于指示第一时域资源，所述目标信道矩阵为所述终端基于所述第一时域资源在所述第一频域资源进行信道估计获得的信道矩阵。
可选地，所述第一频域资源对应一个频域资源块，所述装置还包括：
发送模块，用于向终端发送第一信令，所述终端用于基于所述第一信令激活所述装置配置的所述频域资源块，并在所述频域资源块进行信道估计，获得所述目标信道矩阵。
可选地，所述接收模块501还用于：
接收终端上报的能够同时进行信道估计的所述频域资源块的数量以及所述频域资源块的大小。
可选地，所述接收模块501还用于：接收终端发送的第一请求；
所述装置还包括发送模块，用于基于所述第一请求向所述终端发送第一响应信号；
所述接收模块501还用于：接收所述终端响应于所述第一响应信号发送的训练和/或更新后的AI模型或者发送训练和/或更新后的AI模型中的适用于所述网络侧设备的模型部分；
其中，终端基于所述目标信道矩阵对AI模型进行训练和/或更新，以获得所述训练和/或更新后的AI模型。
可选地，所述装置还包括：
发送模块，用于向所述终端发送第二指示信息，所述第二指示信息用于指示第二频域资源；
所述接收模块501还用于：接收所述终端通过所述第二频域资源对应的频域位置发送的训练和/或更新后的AI模型或者发送训练和/或更新后的AI模型中的适用于所述网络侧设备的模型部分。
可选地，发送模块还用于：
向所述终端发送第三指示信息；
其中，所述第三指示信息用于指示第一时刻，所述终端用于在所述第一时刻应用所述训练和/或更新后的AI模型。
可选地，所述接收模块501还用于：
接收终端发送的码本，所述码本为所述终端将所述目标信道矩阵进行映射得到。
可选地，所述装置还包括：
指示模块，用于向所述终端指示第三频域资源，所述终端用于在所述第三频域资源进 行信道估计，获得信道矩阵，并基于获得的信道矩阵对训练或更新后的AI模型进行更新。
可选地，所述第一频域资源与所述终端对应的小区匹配，在所述终端对应的至少两个小区匹配的所述第一频域资源相同的情况下，所述接收模块501还用于：
接收所述终端通过码分的方式发送的CSI-RS。
本申请实施例中，终端能够基于特定的第一频域资源来进行信道估计得到目标信道矩阵，并将目标信道矩阵发送给所述装置，进而所述装置基于目标信道矩阵来进行对AI模型的训练和/或更新，避免所述装置对于AI模型的训练和/或更新会占用过多的频域资源，也就能够避免对所述装置的其他通信业务造成影响。
本申请实施例提供的AI模型处理装置500能够实现图3所述方法实施例中网络侧设备实现的各个过程，并达到相同的技术效果，为避免重复，这里不再赘述。
可选地，如图6所示，本申请实施例还提供一种通信设备600，包括处理器601和存储器602，存储器602上存储有可在所述处理器601上运行的程序或指令，例如，该通信设备600为终端时，该程序或指令被处理器601执行时实现上述图2所述方法实施例的各个步骤，且能达到相同的技术效果。该通信设备600为网络侧设备时，该程序或指令被处理器601执行时实现上述图3所述方法实施例的各个步骤，且能达到相同的技术效果，为避免重复，这里不再赘述。
本申请实施例还提供一种终端，包括处理器和通信接口，处理器用于确定用于进行AI模型训练的第一频域资源，以及用于在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：基于所述目标信道矩阵对AI模型进行训练和/或更新；向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。该终端实施例与上述终端侧方法实施例对应，上述方法实施例的各个实施过程和实现方式均可适用于该终端实施例中，且能达到相同的技术效果。具体地，图7为实现本申请实施例的一种终端的硬件结构示意图。
该终端700包括但不限于：射频单元701、网络模块702、音频输出单元703、输入单元704、传感器705、显示单元706、用户输入单元707、接口单元708、存储器709以及处理器710等中的至少部分部件。
本领域技术人员可以理解，终端700还可以包括给各个部件供电的电源(比如电池)，电源可以通过电源管理系统与处理器710逻辑相连，从而通过电源管理系统实现管理充电、放电、以及功耗管理等功能。图7中示出的终端结构并不构成对终端的限定，终端可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置，在此不再赘述。
应理解的是，本申请实施例中，输入单元704可以包括图形处理单元(Graphics Processing Unit，GPU)7041和麦克风7042，图形处理器7041对在视频捕获模式或图像捕获模式中由图像捕获装置(如摄像头)获得的静态图片或视频的图像数据进行处理。显示单元706可包括显示面板7061，可以采用液晶显示器、有机发光二极管等形式来配置显示面板7061。用户输入单元707包括触控面板7071以及其他输入设备7072中的至少 一种。触控面板7071，也称为触摸屏。触控面板7071可包括触摸检测装置和触摸控制器两个部分。其他输入设备7072可以包括但不限于物理键盘、功能键(比如音量控制按键、开关按键等)、轨迹球、鼠标、操作杆，在此不再赘述。
本申请实施例中，射频单元701接收来自网络侧设备的下行数据后，可以传输给处理器710进行处理；另外，射频单元701可以向网络侧设备发送上行数据。通常，射频单元701包括但不限于天线、放大器、收发信机、耦合器、低噪声放大器、双工器等。
存储器709可用于存储软件程序或指令以及各种数据。存储器709可主要包括存储程序或指令的第一存储区和存储数据的第二存储区，其中，第一存储区可存储操作系统、至少一个功能所需的应用程序或指令(比如声音播放功能、图像播放功能等)等。此外，存储器709可以包括易失性存储器或非易失性存储器，或者，存储器709可以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)，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，DDRSDRAM)、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)和直接内存总线随机存取存储器(Direct Rambus RAM，DRRAM)。本申请实施例中的存储器709包括但不限于这些和任意其它适合类型的存储器。
处理器710可包括一个或多个处理单元；可选地，处理器710集成应用处理器和调制解调处理器，其中，应用处理器主要处理涉及操作系统、用户界面和应用程序等的操作，调制解调处理器主要处理无线通信信号，如基带处理器。可以理解的是，上述调制解调处理器也可以不集成到处理器710中。
其中，处理器710用于：
确定用于进行AI模型训练的第一频域资源；
在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：
基于所述目标信道矩阵对AI模型进行训练和/或更新；
向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
本申请实施例中，终端能够基于特定的第一频域资源来进行信道估计得到目标信道矩阵，也即能够通过特定的频域资源来进行对AI模型的训练和/或更新，避免终端对于AI模型的训练和/或更新会占用过多的频域资源，也就能够避免对终端的其他通信业务造成影响，确保终端和网络侧设备之间的通信性能。
本申请实施例还提供一种网络侧设备，包括处理器和通信接口，所述通信接口用于接 收终端发送的目标信道矩阵，其中，所述目标信道矩阵为终端在第一频域资源进行信道估计获得的信道矩阵；所述处理器用于基于所述目标信道矩阵对AI模型进行训练和/或更新。该网络侧设备实施例与上述网络侧设备方法实施例对应，上述方法实施例的各个实施过程和实现方式均可适用于该网络侧设备实施例中，且能达到相同的技术效果。
具体地，本申请实施例还提供了一种网络侧设备。如图8所示，该网络侧设备800包括：天线81、射频装置82、基带装置83、处理器84和存储器85。天线81与射频装置82连接。在上行方向上，射频装置82通过天线81接收信息，将接收的信息发送给基带装置83进行处理。在下行方向上，基带装置83对要发送的信息进行处理，并发送给射频装置82，射频装置82对收到的信息进行处理后经过天线81发送出去。
以上实施例中网络侧设备执行的方法可以在基带装置83中实现，该基带装置83包括基带处理器。
基带装置83例如可以包括至少一个基带板，该基带板上设置有多个芯片，如图8所示，其中一个芯片例如为基带处理器，通过总线接口与存储器85连接，以调用存储器85中的程序，执行以上方法实施例中所示的网络设备操作。
该网络侧设备还可以包括网络接口86，该接口例如为通用公共无线接口(common public radio interface，CPRI)。
具体地，本申请实施例的网络侧设备800还包括：存储在存储器85上并可在处理器84上运行的指令或程序，处理器84调用存储器85中的指令或程序执行图5所示各模块执行的方法，并达到相同的技术效果，为避免重复，故不在此赘述。
本申请实施例还提供一种可读存储介质，所述可读存储介质上存储有程序或指令，该程序或指令被处理器执行时实现上述图2所述方法实施例的各个过程，或者实现上述图3所述方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
其中，所述处理器为上述实施例中所述的终端中的处理器。所述可读存储介质，包括计算机可读存储介质，如计算机只读存储器ROM、随机存取存储器RAM、磁碟或者光盘等。
本申请实施例另提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现上述图2所述方法实施例的各个过程，或者实现上述图3所述方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
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。
本申请实施例另提供了一种计算机程序/程序产品，所述计算机程序/程序产品被存储在存储介质中，所述计算机程序/程序产品被至少一个处理器执行以实现上述图2所述方法实施例的各个过程，或者实现上述图3所述方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
本申请实施例还提供了一种通信系统，包括：终端及网络侧设备，所述终端可用于执行如图2所述的方法的步骤，所述网络侧设备可用于执行如上图3所述的方法的步骤。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。此外，需要指出的是，本申请实施方式中的方法和装置的范围不限按示出或讨论的顺序来执行功能，还可包括根据所涉及的功能按基本同时的方式或按相反的顺序来执行功能，例如，可以按不同于所描述的次序来执行所描述的方法，并且还可以添加、省去、或组合各种步骤。另外，参照某些示例所描述的特征可在其他示例中被组合。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对相关技术做出贡献的部分可以以计算机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(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端(可以是手机，计算机，服务器，空调器，或者网络设备等)执行本申请各个实施例所述的方法。
上面结合附图对本申请的实施例进行了描述，但是本申请并不局限于上述的具体实施方式，上述的具体实施方式仅仅是示意性的，而不是限制性的，本领域的普通技术人员在本申请的启示下，在不脱离本申请宗旨和权利要求所保护的范围情况下，还可做出很多形式，均属于本申请的保护之内。

Claims (37)

  1. 一种人工智能AI模型处理方法，包括：
    终端确定用于进行AI模型训练的第一频域资源；
    所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：
    基于所述目标信道矩阵对AI模型进行训练和/或更新；
    向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
  2. 根据权利要求1所述的方法，其中，所述终端确定用于进行AI模型训练的第一频域资源，包括：
    终端接收网络侧设备发送的第一指示信息，所述第一指示信息用于指示进行AI模型训练的第一频域资源；
    所述终端基于所述第一指示信息确定所述第一频域资源。
  3. 根据权利要求2所述的方法，其中，所述第一指示信息还用于指示第一时域资源，所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，包括：
    所述终端基于所述第一时域资源在所述第一频域资源进行信道估计，获得目标信道矩阵。
  4. 根据权利要求1所述的方法，其中，在所述AI模型与目标频域资源的资源数量对应的情况下，所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，包括如下至少一项：
    在所述目标频域资源的资源数量小于或等于所述第一频域资源的资源数量的情况下，所述终端以所述目标频域资源为单位将所述第一频域资源划分为至少一个频域资源块，所述终端在所述至少一个频域资源块进行信道估计，获得目标信道矩阵；
    在所述目标频域资源的资源数量大于所述第一频域资源的资源数量的情况下，所述终端在所述第一频域资源进行信道估计，获得第一信道矩阵，并基于所述第一信道矩阵确定第二信道矩阵，其中，所述目标信道矩阵包括所述第一信道矩阵和第二信道矩阵。
  5. 根据权利要求2所述的方法，其中，所述第一频域资源对应一个频域资源块，所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，包括：
    所述终端接收网络侧设备发送的第一信令；
    所述终端基于所述第一信令激活网络侧设备配置的所述频域资源块，并在所述频域资源块进行信道估计，获得目标信道矩阵。
  6. 根据权利要求5所述的方法，其中，所述第一指示信息还用于指示第一时域资源，所述终端基于所述第一信令激活网络侧设备配置的所述频域资源块，并在所述频域资源块进行信道估计，获得目标信道矩阵，包括：
    所述终端基于所述第一信令激活网络侧设备配置的所述频域资源块，在所述终端基于所述第一指示信息确定在第一时域资源不需要基于所述频域资源块进行数据传输的情况下，所述终端基于所述第一时域资源在所述频域资源块进行信道估计，获得目标信道矩阵；
    所述方法还包括：
    在所述终端基于所述第一指示信息确定在第二时域资源需要基于所述频域资源块进行数据传输的情况下，所述终端基于所述频域资源块进行数据传输，所述第二时域资源为除所述第一时域资源之外的时域资源。
  7. 根据权利要求6所述的方法，还包括：
    在所述终端获取到优先级大于所述信道估计的优先级的目标业务的情况下，所述终端基于所述第一时域资源在所述频域资源块进行所述目标业务的数据传输。
  8. 根据权利要求5所述的方法，还包括：
    所述终端向网络侧设备上报能够同时进行信道估计的所述频域资源块的数量以及所述频域资源块的大小。
  9. 根据权利要求1所述的方法，其中，在所述终端基于所述目标信道矩阵对所述AI模型进行训练和/或更新之后，所述方法还包括：
    所述终端向网络侧设备发送第一请求；
    所述终端接收所述网络侧设备响应于所述第一请求发送的第一响应信号；
    所述终端基于所述第一响应信号，向所述网络侧设备发送训练和/或更新后的所述AI模型或者发送训练和/或更新后的所述AI模型中的适用于所述网络侧设备的模型部分。
  10. 根据权利要求9所述的方法，其中，所述终端基于所述第一响应信号，向所述网络侧设备发送所述AI模型或者发送所述AI模型中的适用于所述网络侧设备的模型部分，包括：
    所述终端接收所述网络侧设备发送的第二指示信息，所述第二指示信息用于指示第二频域资源；
    所述终端基于所述第一响应信号，在所述第二频域资源对应的频域位置向所述网络侧设备发送所述AI模型或者发送所述AI模型中的适用于所述网络侧设备的模型部分。
  11. 根据权利要求9所述的方法，还包括：
    所述终端接收所述网络侧设备发送的第三指示信息，所述第三指示信息用于指示第一时刻；
    所述终端基于所述第三指示信息，在所述第一时刻应用训练和/或更新后的所述AI模型。
  12. 根据权利要求1所述的方法，其中，所述向所述网络侧设备发送所述目标信道矩阵，包括：
    所述终端将所述目标信道矩阵映射为码本，向所述网络侧设备发送所述码本。
  13. 根据权利要求1所述的方法，其中，在所述终端基于所述目标信道矩阵对所述AI 模型进行训练和/或更新之后，所述方法还包括：
    所述终端在第三频域资源使用训练和/或更新后的所述AI模型。
  14. 根据权利要求13所述的方法，还包括：
    所述终端在所述第三频域资源进行信道估计，获得信道矩阵，并基于获得的信道矩阵对训练或更新后的所述AI模型进行更新。
  15. 根据权利要求14所述的方法，其中，所述第三频域资源为网络侧设备指示的频域资源。
  16. 根据权利要求13所述的方法，其中，所述第一频域资源和所述第三频域资源覆盖整个载波带宽。
  17. 根据权利要求13所述的方法，其中，所述终端在第三频域资源使用训练和/或更新后的所述AI模型，包括：
    所述终端以所述第一频域资源为单位将所述第三频域资源划分为至少一个频域资源块，所述终端在所述至少一个频域资源块使用训练和/或更新后的所述AI模型。
  18. 根据权利要求1所述的方法，其中，所述第一频域资源与所述终端对应的小区匹配。
  19. 根据权利要求18所述的方法，其中，在所述终端对应的至少两个小区匹配的所述第一频域资源相同的情况下，所述方法还包括：
    所述终端通过码分的方式向所述网络侧设备发送信道状态信息参考信号CSI-RS。
  20. 根据权利要求18所述的方法，其中，在所述终端对应的每个小区匹配的所述第一频域资源不同的情况下，所述终端在所述第一频域资源进行信道估计，获得目标信道矩阵，包括：
    所述终端在每个所述小区匹配的所述第一频域资源分别进行信道估计，获得目标信道矩阵。
  21. 根据权利要求18所述的方法，其中，所述终端在所述每个小区匹配的所述第一频域资源分别进行信道估计，获得目标信道矩阵，包括：
    所述终端在每个所述小区匹配的所述第一频域资源分别进行信道估计，获得每个小区对应的第一信道矩阵；
    所述终端基于所述每个小区对应的第一信道矩阵得到信道矩阵集合，所述目标信道矩阵为所述信道矩阵集合。
  22. 一种AI模型处理方法，包括：
    网络侧设备接收终端发送的目标信道矩阵，其中，所述目标信道矩阵为终端在第一频域资源进行信道估计获得的信道矩阵；
    所述网络侧设备基于所述目标信道矩阵对AI模型进行训练和/或更新。
  23. 根据权利要求22所述的方法，其中，所述网络侧设备接收终端发送的目标信道矩阵之前，所述方法还包括：
    所述网络侧设备向终端发送第一指示信息，所述第一指示信息用于指示进行AI模型训练的第一频域资源。
  24. 根据权利要求23所述的方法，其中，所述第一指示信息还用于指示第一时域资源，所述目标信道矩阵为所述终端基于所述第一时域资源在所述第一频域资源进行信道估计获得的信道矩阵。
  25. 根据权利要求22所述的方法，其中，所述第一频域资源对应一个频域资源块，所述网络侧设备接收终端发送的目标信道矩阵之前，所述方法还包括：
    所述网络侧设备向终端发送第一信令，所述终端用于基于所述第一信令激活网络侧设备配置的所述频域资源块，并在所述频域资源块进行信道估计，获得所述目标信道矩阵。
  26. 根据权利要求25所述的方法，还包括：
    所述网络侧设备接收终端上报的能够同时进行信道估计的所述频域资源块的数量以及所述频域资源块的大小。
  27. 根据权利要求22所述的方法，还包括：
    所述网络侧设备接收终端发送的第一请求；
    所述网络侧设备基于所述第一请求向所述终端发送第一响应信号；
    所述网络侧设备接收所述终端响应于所述第一响应信号发送的训练和/或更新后的AI模型或者发送训练和/或更新后的AI模型中的适用于所述网络侧设备的模型部分；
    其中，终端基于所述目标信道矩阵对AI模型进行训练和/或更新，以获得所述训练和/或更新后的AI模型。
  28. 根据权利要求27所述的方法，其中，所述网络侧设备接收所述终端响应于所述第一响应信号发送的训练和/或更新后的AI模型或者发送训练和/或更新后的AI模型中的适用于所述网络侧设备的模型部分，包括：
    所述网络侧设备向所述终端发送第二指示信息，所述第二指示信息用于指示第二频域资源；
    所述网络侧设备接收所述终端通过所述第二频域资源对应的频域位置发送的训练和/或更新后的AI模型或者发送训练和/或更新后的AI模型中的适用于所述网络侧设备的模型部分。
  29. 根据权利要求27所述的方法，还包括：
    所述网络侧设备向所述终端发送第三指示信息；
    其中，所述第三指示信息用于指示第一时刻，所述终端用于在所述第一时刻应用所述训练和/或更新后的AI模型。
  30. 根据权利要求22所述的方法，其中，所述网络侧设备接收终端发送的目标信道矩阵，包括：
    所述网络侧设备接收终端发送的码本，所述码本为所述终端将所述目标信道矩阵进行映射得到。
  31. 根据权利要求22所述的方法，还包括：
    所述网络侧设备向所述终端指示第三频域资源，所述终端用于在所述第三频域资源进行信道估计，获得信道矩阵，并基于获得的信道矩阵对训练或更新后的AI模型进行更新。
  32. 根据权利要求22所述的方法，其中，所述第一频域资源与所述终端对应的小区匹配，在所述终端对应的至少两个小区匹配的所述第一频域资源相同的情况下，所述方法还包括：
    所述网络侧设备接收所述终端通过码分的方式发送的CSI-RS。
  33. 一种AI模型处理装置，包括：
    确定模块，用于确定用于进行AI模型训练的第一频域资源；
    执行模块，用于在所述第一频域资源进行信道估计，获得目标信道矩阵，并执行如下至少一项：
    基于所述目标信道矩阵对AI模型进行训练和/或更新；
    向网络侧设备发送所述目标信道矩阵，所述网络侧设备用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
  34. 一种AI模型处理装置，包括：
    接收模块，用于接收终端发送的目标信道矩阵，其中，所述目标信道矩阵为终端在第一频域资源进行信道估计获得的信道矩阵；
    处理模块，用于基于所述目标信道矩阵对AI模型进行训练和/或更新。
  35. 一种终端，包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如权利要求1-21中任一项所述的AI模型处理方法的步骤。
  36. 一种网络侧设备，包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如权利要求22-32中任一项所述的AI模型处理方法的步骤。
  37. 一种可读存储介质，所述可读存储介质上存储程序或指令，所述程序或指令被处理器执行时实现如权利要求1-21中任一项所述的AI模型处理方法的步骤，或者实现如权利要求22-32中任一项所述的AI模型处理方法的步骤。
PCT/CN2023/103909 2022-07-06 2023-06-29 Ai模型处理方法、装置、终端及网络侧设备 WO2024007949A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210800718.3A CN117411746A (zh) 2022-07-06 2022-07-06 Ai模型处理方法、装置、终端及网络侧设备
CN202210800718.3 2022-07-06

Publications (1)

Publication Number Publication Date
WO2024007949A1 true WO2024007949A1 (zh) 2024-01-11

Family

ID=89454257

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/103909 WO2024007949A1 (zh) 2022-07-06 2023-06-29 Ai模型处理方法、装置、终端及网络侧设备

Country Status (2)

Country Link
CN (1) CN117411746A (zh)
WO (1) WO2024007949A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2014039056A1 (en) * 2012-09-10 2014-03-13 Nokia Siemens Networks Oy Codebook construction using csi feedback for evolving deployment scenarios
CN103685093A (zh) * 2013-11-18 2014-03-26 北京邮电大学 一种显式反馈方法及设备
CN112737985A (zh) * 2020-12-25 2021-04-30 东南大学 基于深度学习的大规模mimo信道联合估计和反馈方法
US20210243632A1 (en) * 2020-01-31 2021-08-05 Qualcomm Incorporated Measurements on a first band applicable to procedures on a second band
CN114079599A (zh) * 2020-08-20 2022-02-22 展讯通信（上海）有限公司 Pdsch的信道估计方法、系统及ue

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2014039056A1 (en) * 2012-09-10 2014-03-13 Nokia Siemens Networks Oy Codebook construction using csi feedback for evolving deployment scenarios
CN103685093A (zh) * 2013-11-18 2014-03-26 北京邮电大学 一种显式反馈方法及设备
US20210243632A1 (en) * 2020-01-31 2021-08-05 Qualcomm Incorporated Measurements on a first band applicable to procedures on a second band
CN114079599A (zh) * 2020-08-20 2022-02-22 展讯通信（上海）有限公司 Pdsch的信道估计方法、系统及ue
CN112737985A (zh) * 2020-12-25 2021-04-30 东南大学 基于深度学习的大规模mimo信道联合估计和反馈方法

Also Published As

Publication number Publication date
CN117411746A (zh) 2024-01-16

Similar Documents

Publication Publication Date Title
WO2019223634A1 (zh) 信息处理方法、装置、终端及通信设备
WO2023246618A1 (zh) 信道矩阵处理方法、装置、终端及网络侧设备
JP2024512358A (ja) 情報報告方法、装置、第１機器及び第２機器
WO2023198058A1 (zh) 信息传输方法、装置、终端及网络侧设备
WO2023185978A1 (zh) 信道特征信息上报及恢复方法、终端和网络侧设备
WO2024007949A1 (zh) Ai模型处理方法、装置、终端及网络侧设备
WO2024055974A1 (zh) Cqi传输方法、装置、终端及网络侧设备
WO2023179460A1 (zh) 信道特征信息传输方法、装置、终端及网络侧设备
WO2023179570A1 (zh) 信道特征信息传输方法、装置、终端及网络侧设备
WO2024164961A1 (zh) 信息处理方法、装置、终端及网络侧设备
WO2024055993A1 (zh) Cqi传输方法、装置、终端及网络侧设备
WO2024149157A1 (zh) Csi传输方法、装置、终端及网络侧设备
WO2023185980A1 (zh) 信道特征信息传输方法、装置、终端及网络侧设备
WO2024149156A1 (zh) 信息传输方法、装置、终端及网络侧设备
WO2023185995A1 (zh) 信道特征信息传输方法、装置、终端及网络侧设备
WO2024140578A1 (zh) 基于ai模型的csi反馈方法、终端及网络侧设备
WO2024093999A1 (zh) 信道信息的上报和接收方法、终端及网络侧设备
WO2024140422A1 (zh) Ai单元的性能监测方法、终端及网络侧设备
WO2024217495A1 (zh) 信息处理方法、信息处理装置、终端及网络侧设备
WO2024088161A1 (zh) 信息传输方法、信息处理方法、装置和通信设备
WO2024032606A1 (zh) 信息传输方法、装置、设备、系统及存储介质
WO2023179476A1 (zh) 信道特征信息上报及恢复方法、终端和网络侧设备
WO2024001981A1 (zh) 预编码矩阵的指示方法、终端及网络侧设备
WO2023207920A1 (zh) 信道信息反馈方法、终端及网络侧设备
WO2024051564A1 (zh) 信息传输方法、ai网络模型训练方法、装置和通信设备

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: 23834716

Country of ref document: EP

Kind code of ref document: A1