WO2024051564A1 - 信息传输方法、ai网络模型训练方法、装置和通信设备 - Google Patents

信息传输方法、ai网络模型训练方法、装置和通信设备 Download PDF

Info

Publication number
WO2024051564A1
WO2024051564A1 PCT/CN2023/116030 CN2023116030W WO2024051564A1 WO 2024051564 A1 WO2024051564 A1 WO 2024051564A1 CN 2023116030 W CN2023116030 W CN 2023116030W WO 2024051564 A1 WO2024051564 A1 WO 2024051564A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
terminal
training data
channel
network model
Prior art date
Application number
PCT/CN2023/116030
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 WO2024051564A1 publication Critical patent/WO2024051564A1/zh

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/10—Scheduling measurement reports ; Arrangements for measurement reports
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/06—Optimizing the usage of the radio link, e.g. header compression, information sizing, discarding information

Definitions

  • This application belongs to the field of communication technology, and specifically relates to an information transmission method, artificial intelligence (Artificial Intelligence, AI) network model training method, device and communication equipment.
  • artificial intelligence Artificial Intelligence, AI
  • the AI network model may include an encoding part (i.e., encoding AI network model) and a decoding part (i.e., decoding AI network model).
  • the encoding AI network model is used to encode channel information into channel feature information
  • the decoding AI network model is used to convert the encoding AI network model into The channel characteristic information output by the network model is restored to channel information.
  • the encoding AI network model and the decoding AI network model are usually jointly trained in the same device, and then jointly trained
  • the obtained encoded AI network model is transmitted to the terminal, and the decoded AI network model obtained through joint training is transmitted to the base station.
  • the base station trains the encoding AI network model and the decoding AI network model based on pre-acquired training data. Since the pre-acquired training data is inconsistent with the actual channel, the trained encoding AI network model and decoding AI network model will be The matching degree with the actual channel is relatively low, thus reducing the accuracy of the encoding AI network model and the decoding AI network model in processing actual channel information.
  • Embodiments of the present application provide an information transmission method, AI network model training method, device and communication equipment, enabling the terminal to report actual estimated channel information to the base station, so that the base station uses the channel information as AI training data to train more accurately.
  • an information transmission method which method includes:
  • the terminal collects artificial intelligence AI training data, wherein the AI training data includes first channel information of the target downlink channel, and the AI training data is used to train the first AI network model and/or the second AI network model, and the third An AI network model is used to process the second channel information into the first channel characteristic information.
  • the second AI network model is used to restore the first channel characteristic information into the second channel information.
  • the second channel The information is the channel information of the target downlink channel;
  • the terminal sends the AI training data to the first device.
  • an information transmission device applied to a terminal, and the device includes:
  • Collection module used to collect artificial intelligence AI training data, wherein the AI training data includes the first channel information of the target downlink channel, and the AI training data is used to train the first AI network model and/or the second AI network model , the first AI network model is used to process the second channel information into the first channel characteristic information, and the second AI network model is used to restore the first channel characteristic information into the second channel information, so
  • the second channel information is the channel information of the target downlink channel;
  • the first sending module is used to send the AI training data to the first device.
  • the third aspect provides an information transmission method, including:
  • the first device receives AI training data from the terminal, where the AI training data includes first channel information of the target downlink channel;
  • the first device trains a first AI network model and/or a second AI network model according to the AI training data, wherein the first AI network model is used to process second channel information into first channel feature information, The second AI network model is used to restore the first channel characteristic information to the second channel information, and the second channel information is the channel information of the target downlink channel.
  • an information transmission device applied to the first device, and the device includes:
  • a first receiving module configured to receive AI training data from the terminal, where the AI training data includes first channel information of the target downlink channel;
  • a training module configured to train a first AI network model and/or a second AI network model according to the AI training data, wherein the first AI network model is used to process the second channel information into the first channel feature information, The second AI network model is used to restore the first channel characteristic information to the second channel information, and the second channel information is the channel information of the target downlink channel.
  • a communication device in a fifth aspect, includes a processor and a memory.
  • the memory stores a program or instructions that can be run on the processor.
  • the program or instructions are implemented when executed by the processor. The steps of the method described in the first aspect or the third aspect.
  • a communication device including a processor and a communication interface, wherein the communication interface is used to collect artificial intelligence AI training data and send the AI training data to the first device, wherein the AI
  • the training data includes the first channel information of the target downlink channel.
  • the AI training data is used to train the first AI network model and/or the second AI network model.
  • the first AI network model is used to process the second channel information into First channel characteristic information
  • the second AI network model is used to restore the first channel characteristic information to the second channel information
  • the second channel information is the channel information of the target downlink channel; or,
  • the communication interface is used to receive AI training data from the terminal, where the AI training data includes first channel information of the target downlink channel; the processor is used to train a first AI network model and/or based on the AI training data. Or a second AI network model, wherein the first AI network model is used to process the second channel information into first channel characteristic information, and the second AI network model is used to restore the first channel characteristic information into The second channel information,
  • the second channel information is the channel information of the target downlink channel.
  • a communication system including: a terminal and a network side device.
  • the terminal can be used to perform the steps of the information transmission method as described in the first aspect.
  • the network side device can be used to perform the steps of the information transmission method as described in the third aspect. The steps of the information transmission method.
  • 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 method described in the first aspect are implemented, or the steps of the method are implemented as described in the first aspect. The steps of the method described in the third aspect.
  • a chip in a ninth 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. , or implement the method as described in the third aspect.
  • a computer program/program product is provided, the computer program/program product is stored in a storage medium, and the computer program/program product is executed by at least one processor to implement the method as described in the first aspect
  • the steps of the information transmission method, or the computer program/program product is executed by at least one processor to implement the steps of the information transmission method as described in the third aspect.
  • the terminal collects artificial intelligence AI training data, wherein the AI training data includes the first channel information of the target downlink channel, and the AI training data is used to train the first AI network model and/or the second AI network model, the first AI network model is used to process the second channel information into the first channel characteristic information, and the second AI network model is used to restore the first channel characteristic information into the second channel information, the second channel information is the channel information of the target downlink channel; the terminal sends the AI training data to the first device.
  • the AI training data reported by the terminal to the first device includes actual estimated channel information.
  • the first device in the process of training the first AI network model and/or the second AI network model based on the AI training data, the first device, The degree of matching between the trained first AI network model and/or the second AI network model and the target downlink channel can be improved.
  • the first step can be improved.
  • the compression coding accuracy of the AI network model and/or the decoding accuracy of the second AI network model can be improved and/or the matching degree between the first AI network model and the second AI network model can be improved, thereby reducing the occupancy of the channel information reporting process. resources, and improve the accuracy of the reporting process.
  • Figure 1 is a schematic structural diagram of a wireless communication system to which embodiments of the present application can be applied;
  • Figure 2 is a flow chart of an information transmission method provided by an embodiment of the present application.
  • Figure 3 is a flow chart of an AI network model training method provided by an embodiment of the present application.
  • Figure 4 is a schematic structural diagram of an information transmission device provided by an embodiment of the present application.
  • Figure 5 is a schematic structural diagram of an AI network model training device provided by an embodiment of the present application.
  • Figure 6 is a schematic structural diagram of a communication device provided by an embodiment of the present application.
  • Figure 7 is a schematic diagram of the hardware structure of a terminal 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 may 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 palmtop 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
  • AR augmented reality
  • VR virtual reality
  • robots wearable devices
  • WUE Vehicle User Equipment
  • PUE Pedestrian User Equipment
  • 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 device 12 may include an access network device or a core network device, where the access network device may also be called a wireless access network device, a radio access network (Radio Access Network, RAN), or a wireless access network device.
  • Access network equipment may include a base station, a Wireless Local Area Network (WLAN) access point or a WiFi node, etc.
  • WLAN Wireless Local Area Network
  • the base station may be called a Node B, an Evolved Node B (eNB), an access point, a base transceiver station ( Base Transceiver Station (BTS), radio base station, radio transceiver, Basic Service Set (BSS), Extended Service Set (ESS), home B-node, home evolved B-node, transmitting and receiving point ( 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 the embodiment of this application, only in the NR system The base station is introduced as an example, and the specific type of base station is not limited.
  • AI network models such as neural networks, decision trees, support vector machines, Bayesian classifiers, etc. This application takes a neural network as an example for explanation, but does not limit the specific type of AI network model.
  • the encoding AI network model is used on the terminal side to compress and encode the channel information, and the compressed and encoded channel characteristic information is reported.
  • the decoding AI network model can be used on the network side to decode the received channel characteristic information. processing to recover channel information.
  • the encoding AI network model needs to match the decoding AI network model.
  • the encoding AI network model and the decoding AI network model are usually jointly trained on the network side to achieve an appropriate matching degree between the two.
  • the AI training data used to train the coding AI network model and the decoding AI network model are usually pre-obtained data, or channel information collected by terminals or network-side devices dedicated to collecting training data, and are trained based on the AI training data.
  • the network side device After obtaining the encoding AI network model and the decoding AI network model, the network side device then delivers the encoding AI network model to the terminal, so that the terminal encodes channel information based on the encoding AI network model delivered by the network side device.
  • the encoding AI network model and decoding AI network model trained by the network side device based on pre-acquired data or channel information collected by a terminal or network side device dedicated to collecting training data may be different from the actual channel information of the terminal.
  • the encoding AI network model has a low degree of compression of the actual channel information of the terminal, causing the compressed channel characteristic information to occupy a large amount of resources, or the encoding AI network model and the decoding AI network model have a negative impact on the terminal.
  • the processing of actual channel information is not accurate enough, which reduces the accuracy of channel information reporting.
  • the terminal will report the estimated channel information of the actual channel as AI training data to the training encoding AI network model (ie, the first AI network model in the embodiment of the present application) and/or the decoding AI network model. (i.e., the second AI network model in the embodiment of the present application).
  • the encoding AI network model and/or the decoding AI network model trained by the first device based on the AI training data can be consistent with the actual channel status of the terminal. matching, thereby reducing the resources occupied by the channel information reporting process and improving the accuracy of the channel information reporting process.
  • an information transmission method provided by an embodiment of the present application is executed by a terminal, such as various types of terminals 11 listed in Figure 1.
  • the information transmission method performed by the terminal may include the following steps:
  • Step 201 The terminal collects artificial intelligence AI training data, where the AI training data includes the target downlink signal.
  • the first channel information of the channel the AI training data is used to train the first AI network model and/or the second AI network model, the first AI network model is used to process the second channel information into the first channel characteristic information , the second AI network model is used to restore the first channel characteristic information to the second channel information, where the second channel information is the channel information of the target downlink channel.
  • the AI training data includes the first channel information of the target downlink channel.
  • the terminal may use the actually estimated channel information of the target downlink channel as the AI training data, or the terminal may determine the actually estimated channel information of the target downlink channel.
  • Preprocessing such as at least one of filtering, quantification, normalization, orthogonalization and other processing methods for channel information
  • the processed channel information is used as AI training data.
  • the first channel information is at least one of a precoding matrix and a channel matrix.
  • the embodiments of this application are summarized by taking the first channel information as a precoding matrix as an example. This does not constitute a detailed description. limited.
  • the AI training data includes at least one of the following:
  • the strongest M-layer precoding matrix among the N-layer precoding matrices, the channel information of the target downlink channel includes the N-layer precoding matrix, N is a positive integer, and M is a positive integer less than or equal to N;
  • the precoding matrix indicated by third indication information where the third indication information comes from the first device
  • the precoding matrix indicated by fourth indication information where the fourth indication information is the indication information sent by the terminal to the first device in advance;
  • a precoding matrix that satisfies a preset condition a precoding matrix that satisfies a preset condition.
  • the strongest M-layer precoding matrix among the N-layer precoding matrices can be that the rank (Rank) N of the target downlink channel is greater than 1.
  • the precoding matrix of the target downlink channel includes N layers, where the strength The larger the layer, the more effectively it can reflect the channel status of the target downlink channel.
  • the strongest one is selected from the N-layer precoding matrix (such as the largest eigenvalue, the largest singular value, or the largest proportion of power, or the signal and
  • SINR Signal to Interference plus Noise Ratio
  • the terminal reports the precoding matrix of the strongest layer.
  • the terminal when the terminal reuses the conventional channel state information (CSI) report as AI training data, it can use the precoding matrix corresponding to layer 0 (layer 0) in the CSI report as AI training data. data.
  • CSI channel state information
  • the terminal can also use the N-layer precoding matrix as AI training data, so that it can accurately reflect the real channel status of the target downlink channel.
  • the terminal can receive the above-mentioned third indication information before sending the AI training data.
  • the third indication information can indicate the precoding matrix that the terminal needs to report. In this way, the terminal only needs to report the precoding matrix of the specified layer according to the instructions of the first device. Encoding matrix is enough.
  • the terminal can choose which layers of precoding matrices to report, and report fourth indication information to indicate which layers of precoding matrices the terminal reports.
  • the terminal may send the fourth instruction information before, after, or at the same time as the AI training data, which is not specifically limited here.
  • the terminal can select the required precoding matrix based on whether each layer in the N-layer precoding matrix meets the preset conditions. Among them, the precoding matrix that meets the preset conditions can more effectively reflect the channel status of the target downlink channel.
  • the preset conditions include at least one of the following:
  • Channel quality indicator Channel quality indicator (CQI) is greater than or equal to the first threshold
  • the signal to interference plus noise ratio SINR is greater than or equal to the second threshold
  • the characteristic value is greater than or equal to the third threshold
  • the singular value is greater than or equal to the fourth threshold.
  • the terminal can filter and report which layer or layer of precoding matrices according to preset conditions. Compared with selecting the entire N-layer precoding matrix, it can reduce the amount of AI training data and minimize the amount of AI training data. It is possible to retain the effective information in the N-layer precoding matrix.
  • the terminal Fourth indication information may also be sent to the first device to indicate which layer or layer of the N-layer precoding matrix the precoding matrix reported by the terminal is.
  • the terminal can retain the true values of the elements in the precoding matrix of the layer that needs to be reported in the N-layer precoding matrix, and use 0 or other values for the elements in the precoding matrix of the layer that does not need to be reported in the N-layer precoding matrix. Instead, in this way, the first device can determine the corresponding physical meaning of each element according to the position of each element in the N-layer precoding matrix.
  • Step 202 The terminal sends the AI training data to the first device.
  • the first device may be any device used to train an AI network model, such as an access network device (such as a base station) or a core network device.
  • an access network device such as a base station
  • a core network device such as a core network device.
  • the first device is a base station as an example. for example.
  • the above-mentioned first AI network model is used to process the second channel information into the first channel characteristic information.
  • the first AI network model may be an encoding AI network side model.
  • the encoding AI network side model is used to compress and encode the channel information into a channel. Feature information.
  • the above-mentioned second AI network model is used to restore the first channel characteristic information to the second channel information.
  • the second AI network model may be a decoding AI that matches the first AI network side model. Network model.
  • the decoding AI network model is used to decode the channel characteristic information processed by the encoding AI network side model to restore the channel information.
  • the above-mentioned first device may train or adjust the first AI network model and/or the second AI network model according to the AI training data.
  • the first AI network model and the second AI network model are neural network models
  • the methods of training the first AI network model and/or the second AI network model include the following three methods:
  • the neural network model includes a joint neural network model composed of the encoder of the terminal and the decoder of the base station, and is jointly trained by the network side device. After the training is completed, the base station sends the AI network model of the encoder to the terminal.
  • the first device includes a base station.
  • Option 2 The terminal trains the encoding AI network model, and the base station trains the decoding AI network model, and matches through the matching process, so that the terminal's encoding AI network model and the base station's decoding AI network model match each other.
  • the matching process may include: the terminal uses the coding AI network model to encode the channel information, obtains the coded data (i.e., channel characteristic information), and sends the original channel information and the terminal's coded data to the base station, and the base station uses decoding The AI network model decodes the terminal's encoded data into corresponding channel information, and adjusts the decoding AI network model.
  • the first device includes a base station.
  • Option 3 The terminal trains the encoding AI network model and the decoding AI network model, and uses the encoding AI network model to encode the original channel information to obtain the encoded data, and then sends the original channel information and the terminal's encoded data to the base station.
  • the base station trains the decoding AI network model based on the channel information and the terminal's encoding data. At this time, the base station does not need to train the encoding AI network model.
  • the first device includes a base station.
  • the time and/or frequency resources for the terminal to report AI training data may be indicated by the network side device or agreed in the agreement.
  • AI training data may be collected and reported periodically, or collected and reported at a designated time. AI training data.
  • the method before the terminal collects AI training data, the method further includes:
  • the terminal receives first instruction information from the first device, where the first instruction information is used to instruct the terminal to collect the AI training data.
  • the first instruction information may instruct the terminal to collect and report the AI training data.
  • the base station instructs the terminal to activate the collection and reporting of AI training data through downlink control information (DCI).
  • DCI downlink control information
  • the terminal receives When the first instruction information is received, the collection and reporting of AI training data can be started immediately.
  • the terminal after receiving the above-mentioned first indication information, the terminal can determine whether it supports collecting and reporting AI training data based on its own status.
  • the information transmission method also includes:
  • the terminal determines that collection of the AI training data is supported based on the status information of the terminal, it sends first information to the first device, and the first information instructs the terminal to collect and report the AI training data;
  • the terminal determines that collection of the AI training data is not supported based on the status information of the terminal, it sends second information to the first device, and the second information indicates that the terminal will not collect or report it.
  • the AI training data When the terminal determines that collection of the AI training data is not supported based on the status information of the terminal, it sends second information to the first device, and the second information indicates that the terminal will not collect or report it.
  • the AI training data When the terminal determines that collection of the AI training data is not supported based on the status information of the terminal, it sends second information to the first device, and the second information indicates that the terminal will not collect or report it.
  • the status information of the terminal may include status information of the terminal that changes over time, such as location, moving speed, remaining power, user settings, etc.
  • the terminal when the terminal receives the first instruction information or the first configuration information, it can determine whether it currently supports collecting and reporting the AI training data based on its actual situation. For example, if the user sets the terminal not to collect and reporting the AI training data, it is determined that the terminal does not support collecting and reporting the AI training data; when the terminal's power is low, it is determined that the terminal does not support collecting and reporting the AI training data; during the movement of the terminal, it is determined The terminal does not support collecting and reporting the AI training data, etc.
  • the first device knows that the terminal can collect and report the AI training data, and thus only needs to wait for the terminal to report the AI training data.
  • the first device after the terminal sends the second information to the first device, the first device knows that the terminal will not report AI training data, so it can re-select another terminal to collect and report AI training data, or use Other ways to obtain AI training data. For example: sending the first instruction information to other terminals that may support the collection of AI training data, etc.
  • the method before the terminal collects AI training data, the method further includes:
  • the terminal receives first configuration information from the first device, where the first configuration information is used to configure target time domain and/or frequency domain resources;
  • the terminal collects AI training data, including:
  • the terminal collects AI training data in the target time domain and/or frequency domain resources.
  • the first configuration information can configure the terminal to collect and report the AI training data at a specified time or according to a specified period.
  • the base station controls it through Radio Resource Control (RRC) signaling or media access control layer.
  • RRC Radio Resource Control
  • the unit configures the time or period for the terminal to collect AI training data, so that the terminal can activate the collection and reporting of AI training data at the corresponding time.
  • MAC CE Medium Access Control Element
  • This implementation is similar to the above-mentioned way for the terminal to collect AI training data based on the first instruction information from the first device.
  • the terminal collection can be configured through protocol agreement or network side pre-configuration. And/or report time domain and/or frequency domain resources of AI training data.
  • the method before the terminal receives the first indication information or the first configuration information from the first device, the method further includes:
  • the terminal sends target capability information to the first device, where the target capability information indicates the information reporting capability and/or data collection capability supported by the terminal.
  • the target capability information may be the communication capability information of the terminal.
  • the target capability information is used by the first device to determine whether the terminal supports the collection of AI training data, and thereby sends the first indication information or the first instruction to the terminal that supports the collection of AI training data. Configuration information.
  • the target capability information includes at least one of the following:
  • the second indication information is used to indicate whether the terminal supports collecting the AI training data
  • CSI Channel State Information
  • RS Reference Signal
  • the total number of ports of the terminal is the total number of ports of the terminal.
  • the second indication information can clearly indicate whether the terminal supports collecting the AI training data.
  • the terminal determines whether the terminal supports collecting and reporting AI training data based on its own capabilities and/or status information.
  • the maximum number of CSI-RS ports supported by the terminal can be used as a basis for the first device to determine whether the terminal supports collecting the AI training data.
  • the first device can determine the number of CSI-RS ports based on the maximum number of CSI-RS ports supported by the terminal. Whether the terminal can collect AI training data based on the existing CSI-RS port.
  • the total number of ports of the terminal can be used as a basis for the first device to determine whether the terminal supports collecting the AI training data.
  • the first device can use the number of all ports of the terminal to be greater than or equal to the ports required for collecting AI training data. If the number is high, it is judged that the terminal supports collecting AI training data.
  • the first device can obtain the capability information of the terminal in advance, so that it can filter the terminal according to the capability information, and send the first instruction information or the first configuration information to the terminal that supports collecting AI training data. In this way, it can Reduce the probability that the first device instructs or configures a terminal that does not support the collection of AI training data to collect AI training data, thereby saving resources occupied in the process of instructing or configuring a terminal that does not support the collection of AI training data to collect AI training data, and reducing The delay in collecting and reporting AI training data.
  • the first device can also collect other methods to determine which terminal or terminals to send the first instruction information or the first configuration information, for example, obtain a priori information such as the terminal's location, resource configuration, etc., and determine whether the terminal is Support collecting AI training data, thereby sending first instruction information or first configuration information to a terminal that supports collecting AI training data.
  • the terminal collects AI training data, including:
  • the terminal collects AI training data based on the first port corresponding to the first reference signal resource or the CSI-RS port corresponding to the CSI-RS resource, wherein the first port includes a preconfigured port for collecting the AI training data.
  • a port corresponding to the first reference signal resource, and the CSI-RS port includes a port used for CSI measurement.
  • the first port may be a port configured by the base station specifically for collecting AI training data.
  • the terminal can perform channel estimation on the first reference signal resource corresponding to the first port to achieve AI collection. training data.
  • the first reference signal resource may be a CSI-RS resource or may be a resource of other types of reference signals, which is not specifically limited here.
  • the base station can configure an additional first port for the terminal to collect channel estimation of AI training data.
  • the CSI-RS port can represent the port used for existing CSI measurement, that is, the terminal can reuse the CSI-RS port for normal CSI measurement.
  • the AI training data is carried in at least one of the following:
  • the first UCI carries the CSI report of the target downlink channel, and the CSI report is related to the first channel information;
  • a second UCI, the second UCI does not carry the CSI report of the target downlink channel
  • Target physical uplink shared channel Physical Uplink Shared Channel, PUSCH.
  • the AI training data when carried in the first UCI, the AI training data can be used as part of the CSI report and reported through the first UCI.
  • the CSI report and the first channel information The correlation can be understood as: the first channel information in the CSI report may be the original channel information of the CSI report, and the CSI report may be channel information in the form of a codebook or precoded channel information or coding processing based on the first AI network model The subsequent channel characteristic information.
  • the AI training data includes first channel information in the CSI report, where the first channel information in the CSI report is precoding information.
  • the CSI report in the form of the codebook can be reused as AI training data.
  • the terminal can no longer independently report AI training data, and the base station can receive the CSI reports serve as AI training data.
  • the above-mentioned second UCI is a UCI that does not carry the CSI report of the target downlink channel, that is, the AI training data is reported through the UCI independently of the CSI.
  • the above target PUSCH carries AI training data.
  • the AI training data can be used as data content and reported through PUSCH.
  • the terminal can flexibly report AI training data based on any one or more of the above options one to three, which are not specifically limited here.
  • the method further includes:
  • the terminal performs first processing on the first channel information based on the first AI network model to obtain second channel characteristic information
  • the terminal sends the second channel characteristic information to the first device, where the AI training data and the second channel characteristic information are used to train the second AI network model.
  • the terminal is pre-configured or trained to obtain the first AI network model.
  • the terminal in addition to sending the original channel information to the first device, the terminal also sends to the first device second channel characteristic information obtained by compressing and encoding the first channel information using the first AI network model. .
  • the first device can input the second channel characteristic information into the second AI network model, and train or adjust the second AI network model for the purpose of outputting the first channel information from the second AI network model, wherein based on the AI training data and the third Training the second AI network model using the second channel characteristic information can improve the matching degree between the second AI network model and the first AI network model already possessed by the terminal.
  • first channel information there is a one-to-one correspondence between the first channel information and the second channel characteristic information, that is, if a piece of first channel information is input to the first AI network model, then the first AI network model will output the first channel.
  • second channel characteristic information corresponding to the information; if another piece of first channel information is input to the first AI network model, the first AI network model outputs second channel characteristic information corresponding to the other piece of first channel information.
  • the terminal may send the corresponding first channel information and the second channel characteristic information at the same time, or in the mutually related time domain and/or Or the first channel information and the second channel characteristic information corresponding to the first channel information are respectively transmitted on frequency domain resources, or the reported first channel information carries identification information of the corresponding second channel characteristic information, etc.
  • the AI training data carries first identification information of the second channel characteristic information
  • the terminal sends the AI training data to the first device, including:
  • the terminal sends a target CSI report to the first device, where the target CSI report carries the second channel characteristic information and the AI training data; or,
  • the terminal sends the AI training data to the first device on time domain and/or frequency domain resources corresponding to the first identification information of the second channel characteristic information.
  • the terminal sends a target CSI report to the first device.
  • the target CSI report carries the second channel characteristic information and the AI training data.
  • the terminal may be using the first AI network model to perform CSI reporting, and The corresponding first channel information is added to the CSI and reported together with the AI-encoded second channel characteristic information.
  • the base station After receiving the CSI report, the base station can obtain the first channel information and the corresponding AI-encoded second channel characteristic information.
  • the base station can restore the second channel characteristic information to the first channel information by using the second AI network model as Purpose: train the second AI network model.
  • the base station when the base station obtains the first channel information and the second channel characteristic information at the same time, it can also directly use the first channel information as the channel information for subsequent scheduling. In this way, the accuracy of the scheduling process can be improved. Spend.
  • Case 2 When the terminal sends the AI training data to the first device on the time domain and/or frequency domain resources corresponding to the first identification information of the second channel characteristic information, the base station receives the AI training data respectively. data and second channel characteristic information, and can determine which first channel information in the AI training data corresponds to the corresponding relationship between the time domain and/or frequency domain resources of the received AI training data and the first identification information.
  • the second channel characteristic information corresponds to the second AI network model, and the second AI network model can be trained for the purpose of restoring the second channel characteristic information to the first channel information.
  • the first identifier may be an identifier (Identity, ID) of the CSI report configuration (report configuration) or any other identifier that can directly or indirectly identify the second channel characteristic information, which is not specifically limited here.
  • Case 3 When the above-mentioned AI training data carries the first identification information of the second channel characteristic information, the base station can determine the first channel information in the AI training data based on the first identification information carried in the AI training data. The corresponding second channel characteristic information can then be used to train the second AI network model for the purpose of restoring the second channel characteristic information to the first channel information.
  • the second UCI or target PUSCH can be used in conjunction with the AI
  • the CSI report corresponding to the training data is sent on the relevant time domain and/or frequency domain resources.
  • the first device receives the AI training data on the time domain and/or frequency domain resources, it can determine whether the AI training data is consistent with the time domain and/or frequency domain resources.
  • the second channel characteristic information in the CSI report corresponds to domain and/or frequency domain resources.
  • the second UCI or target PUSCH may carry the corresponding The first identification information of the CSI report.
  • the first device receives the second UCI or the target PUSCH, it can obtain the AI training data and the first identification information of the CSI report corresponding to the AI training data, thereby determining the AI training The data corresponds to the second channel characteristic information in the CSI report identified by the first identification information.
  • the first UCI if the AI training data is carried in the first UCI, at this time, the first UCI also carries the second channel characteristic information corresponding to the AI training data. In this way, the first device receives By the time of the first UCI, Corresponding AI training data and second channel feature information can be obtained therefrom.
  • the AI training data includes a precoding matrix determined based on a first codebook, a first parameter of the first codebook is greater than a preset parameter, and the first parameter includes at least one of the following item:
  • the above-mentioned preset parameters may be the number of ports, the number of delay paths, the number of beams, and the non-zero coefficient ratio of the existing R16 codebook.
  • Pv is used to calculate the number of delay paths as shown in Table 1 above, and the R16 codebook supports up to 32 ports.
  • At least one of the number of ports, the number of delay paths, the number of beams, and the non-zero coefficient ratio of the first codebook may be greater than any combination of the number of ports and the number of delay paths in Table 1 above. , the number of beams, and the proportion of non-zero coefficients.
  • the first parameter of the first codebook in this implementation may be a parameter combination as shown in Table 2 below:
  • a codebook with higher accuracy than the R16 codebook can be used to report AI training data.
  • the AI training data can more accurately describe the channel status of the target downlink channel, so that based on the The AI network model trained with AI training data better matches the actual channel status of the target downlink channel.
  • the terminal collects artificial intelligence AI training data, wherein the AI training data includes the first channel information of the target downlink channel, and the AI training data is used to train the first AI network model and/or the second AI network model, the first AI network model is used to process the second channel information into first channel characteristic information, the second The AI network model is used to restore the first channel characteristic information to the second channel information, where the second channel information is the channel information of the target downlink channel; the terminal sends the AI training to the first device data.
  • the AI training data reported by the terminal to the first device includes actual estimated channel information.
  • the first device in the process of training the first AI network model and/or the second AI network model based on the AI training data, the first device, The degree of matching between the trained first AI network model and/or the second AI network model and the target downlink channel can be improved.
  • the first step can be improved.
  • the compression coding accuracy of the AI network model and/or the decoding accuracy of the second AI network model can be improved and/or the matching degree between the first AI network model and the second AI network model can be improved, thereby reducing the occupancy of the channel information reporting process. resources, and improve the accuracy of the reporting process.
  • An embodiment of the present application provides an AI network model training method, the execution subject of which is the first device.
  • the first device may be a network-side device, such as the network-side device 12 listed in the embodiment as shown in Figure 1 or a core network device.
  • the first device is a base station as an example. for example.
  • the AI network model training method executed by the first device may include the following steps:
  • Step 301 The first device receives AI training data from the terminal, where the AI training data includes first channel information of the target downlink channel.
  • AI training data and first channel information have the same meaning as the AI training data and first channel information in the method embodiment shown in Figure 2, and are not specifically limited here.
  • Step 302 The first device trains a first AI network model and/or a second AI network model according to the AI training data, wherein the first AI network model is used to process the second channel information into a first channel. Characteristic information, the second AI network model is used to restore the first channel characteristic information to the second channel information, and the second channel information is the channel information of the target downlink channel.
  • the base station can train the encoding AI network model based on the AI training data, or train the decoding AI network model based on the AI training data, or jointly train the encoding AI network model and the decoding AI network model based on the AI training data.
  • the base station can obtain the channel information obtained by performing channel estimation on the target downlink channel from the terminal to train the encoding and/or decoding AI network model suitable for the target downlink channel, and can improve the first AI network model obtained by training. And/or the degree of matching between the second AI network model and the target downlink channel.
  • the method further includes:
  • the first device sends relevant information of the first AI network model to the terminal.
  • the base station after completing the training of the coding AI network model, delivers the trained coding AI network model to the terminal.
  • the terminal can use the coding AI network model to estimate the Compress and encode the channel information and report the compressed and encoded channel characteristic information.
  • the base station can use the decoding AI network model that matches the encoding AI network model to restore the channel characteristic information to the original channel information.
  • information or the base station can also use a non-AI method to restore the channel characteristic information to the original channel information, for example, using a certain algorithm to restore the channel characteristic information to the original channel information.
  • the method before the first device receives the AI training data from the terminal, the method further includes:
  • the first device sends first instruction information to the terminal, where the first instruction information is used to instruct the terminal to collect the AI training data.
  • the above-mentioned first instruction information has the same meaning and function as the first instruction information in the method embodiment shown in Figure 2, and will not be described again here.
  • the method before the first device receives the AI training data from the terminal, the method further includes:
  • the first device sends first configuration information to the terminal, where the first configuration information is used to configure target time domain and/or frequency domain resources, where the AI training data is in the target time domain and/or frequency domain. /Or collected from frequency domain resources.
  • the above-mentioned first configuration information has the same meaning and function as the first configuration information in the method embodiment shown in Figure 2, and will not be described again here.
  • the method before the first device sends the first indication information or the first configuration information to the terminal, the method further includes:
  • the first device receives target capability information from the terminal, wherein the target capability information indicates the information reporting capability and/or data collection capability supported by the terminal;
  • the first device sends first indication information or first configuration information to the terminal, including:
  • the first device determines that the terminal supports collecting and reporting the AI training data according to the target capability information
  • the first device sends first indication information or first configuration information to the terminal.
  • the target capability information includes at least one of the following:
  • the second indication information is used to indicate whether the terminal supports collecting the AI training data
  • the total number of ports of the terminal is the total number of ports of the terminal.
  • the above target capability information has the same meaning and function as the target capability information in the method embodiment shown in Figure 2, and will not be described again here.
  • the first device obtains the target capability information of the terminal in advance, and accordingly selects a terminal that supports collecting and reporting the AI training data to perform the action of collecting and reporting the AI training data.
  • the method further includes:
  • the first device receives first information from the terminal, wherein the first information instructs the terminal to collect and report the AI training data;
  • the first device receives second information from the terminal, and the second information indicates that the terminal will not collect or report the AI training data.
  • the base station when the base station receives the first information from the terminal, it can receive AI training data from the terminal.
  • the base station when the base station receives the second information from the terminal, it can select other terminals to collect and report AI training data until it receives the first information from the terminal, for example: the base station sends data to other terminals.
  • the terminal sends the first instruction information or the first configuration information, and receives the first information from the other terminal.
  • the base station continues to look for a device that can support collection and reporting of the AI training. Data terminal.
  • the base station when it receives the second information from the terminal, it can obtain AI training data in other ways, such as using pre-stored data to train the encoding and/or decoding AI network model, which will not be described in detail here.
  • the method before the first device receives the AI training data from the terminal, the method further includes:
  • the first device configures a first port corresponding to a first reference signal resource and/or a CSI-RS port corresponding to a CSI-RS resource for the terminal, where the first port is dedicated to collecting the AI training data.
  • the port corresponding to the first reference signal resource, and the CSI-RS port is a port used for CSI measurement and collection of the AI training data.
  • first reference signal resource, first port, and CSI-RS port have the same meaning and function as the first reference signal resource, first port, and CSI-RS port in the method embodiment shown in Figure 2. Here, No longer.
  • the AI training data is carried in at least one of the following:
  • First uplink control information UCI the first UCI carries a CSI report of the target downlink channel, and the CSI report is related to the first channel information;
  • a second UCI, the second UCI does not carry the CSI report of the target downlink channel
  • Target physical uplink shared channel PUSCH Target physical uplink shared channel
  • the AI training data includes first channel information in the CSI report, where the first channel information in the CSI report is precoding information.
  • the base station can receive the CSI report and AI training data from the terminal. At this time, the base station can schedule the target downlink channel based on the more accurate first channel information in the AI training data. In this way, the base station scheduling target downlink channel can be improved. Channel reliability.
  • the method further includes:
  • the first device receives the second channel characteristic information from the terminal, wherein the second channel characteristic information is obtained by first processing the first channel information based on the first AI network model;
  • the first device trains the first AI network model and/or the second AI network model according to the AI training data, including:
  • the first device trains the second AI network model according to the AI training data and the second channel characteristic information.
  • the first device trains the second AI according to the AI training data and the second channel characteristic information.
  • the network model may input the two channel characteristic information into the second AI network model, and train the second AI network model for the purpose of outputting the first channel information corresponding to the second channel characteristic information from the second AI network model.
  • the first AI network model already owned by the terminal may be trained by the terminal based on the estimated actual channel information, or may be an AI network model previously trained and delivered by the base station, which is not specifically limited here.
  • the AI training data carries the first identification information of the second channel characteristic information
  • the first device receives AI training data from the terminal, including:
  • the first device receives a target CSI report from the terminal, where the target CSI report carries the second channel characteristic information and the AI training data; or,
  • the first device receives the AI training data from the terminal on time domain and/or frequency domain resources corresponding to the first identification information of the second channel characteristic information.
  • the AI training data includes a precoding matrix determined based on a first codebook, a first parameter of the first codebook is greater than a preset parameter, and the first parameter includes at least one of the following item:
  • the first channel information is at least one of a precoding matrix and a channel matrix.
  • the AI training data includes at least one of the following:
  • the strongest M-layer precoding matrix among the N-layer precoding matrices, the channel information of the target downlink channel includes the N-layer precoding matrix, N is a positive integer, and M is a positive integer less than or equal to N;
  • the precoding matrix indicated by third indication information where the third indication information comes from the first device
  • the precoding matrix indicated by fourth indication information where the fourth indication information is the indication information sent by the terminal to the first device in advance;
  • a precoding matrix that satisfies a preset condition a precoding matrix that satisfies a preset condition.
  • the preset conditions include at least one of the following:
  • the channel quality indicator CQI is greater than or equal to the first threshold
  • the signal to interference plus noise ratio SINR is greater than or equal to the second threshold
  • the characteristic value is greater than or equal to the third threshold
  • the singular value is greater than or equal to the fourth threshold.
  • the AI network model training method executed by the first device provided by the embodiment of the present application corresponds to the information transmission method executed by the terminal, and can improve the training of the first AI network model and/or the second AI network model and the target downlink channel. degree of matching.
  • the first device is a base station as an example to illustrate the present application.
  • the information transmission method and AI network model training method provided by the embodiment may include the following steps:
  • Step 1 The base station instructs the terminal through DCI to activate the measurement and reporting of AI training data.
  • the base station configures the time or period for measuring AI training data through RRC/MACCE signaling, and the terminal activates the measurement and reporting of AI training data at the corresponding time. .
  • the terminal can feedback to the base station the status of whether it supports measuring and reporting AI training data. For example, if the terminal has insufficient power, it will report that it cannot perform measurement and report AI training data. ; If the user does not agree to the report, the report cannot perform measurement and report AI training data; if the terminal is in a high-speed motion state, the channel measurement may be inaccurate at this time, and the report cannot perform measurement and report AI training data.
  • Step 2 The terminal measures the channel information and calculates the corresponding precoding matrix.
  • the terminal uses the R-16TypeII codebook to report CSI, where CSI can include Precoding Matrix Indicator (PMI), Channel Quality Indicator (CQI), Rank Indicator (RI), etc.
  • CSI can include Precoding Matrix Indicator (PMI), Channel Quality Indicator (CQI), Rank Indicator (RI), etc.
  • PMI Precoding Matrix Indicator
  • CQI Channel Quality Indicator
  • RI Rank Indicator
  • the base station may not yet have a decoding AI network model, or the base station's decoding AI network model may not match the terminal's encoding AI network model.
  • the terminal can incorporate the precoding matrix information into the CSI as AI training data and report it together.
  • the base station After the base station obtains the CSI, it can solve the PMI for scheduling and solve the AI training data.
  • the data is used for AI model training, and after training, the encoded AI network model is sent to the terminal.
  • the base station After the base station obtains the CSI, it can decode the AI training data for AI model training. After training, the encoded AI network model is sent to the terminal.
  • the base station can also use the AI training data to replace PMI as the precoding information for scheduling.
  • the terminal can also send AI training data to the base station through PUSCH. After the AI network model is trained, the base station then sends the encoded AI network model to the terminal. Or when the first device is a node other than the base station (such as a core network), the terminal can also send the AI training data to the node through PUSCH. After the AI network model is trained, the node can encode the AI The network model is sent to the terminal, and the decoded AI network model is sent to the base station.
  • the terminal has trained its own coding AI network model, but the base station does not have a decoding AI network model or the base station’s decoding AI network model does not match the terminal’s coding AI network model.
  • the terminal can send the corresponding channel characteristic information encoded by the coded AI network model while sending the AI training data.
  • the specific sending method is similar to the above-mentioned sending method of the AI training data, and will not be described again here.
  • the terminal uses the encoding AI network model to report PMI.
  • the terminal's encoding AI network model has matched the base station's decoding AI network model and can work. However, because there are remaining resources at this time, the terminal can still report AI training data, or because the time is too long and the channel changes, the encoding AI network model and the decoding AI network model need to be rematched. At this time, the terminal collects and reports AI training data data.
  • the terminal since the terminal has used the coded AI network model to report PMI, the terminal will use the corresponding original channel information
  • the information is reported as AI training data, that is, the AI training data and the PMI in the CSI report correspond to each other, and the reported PMI is the encoding result of the AI training data.
  • the terminal can report AI training data in a CSI report or through other UCI.
  • the base station instructs the terminal to report the time-frequency position of the UCI carrying AI training data. This time-frequency position is consistent with the CSI report configuration ID.
  • the base station when the base station receives the coded PMI from the CSI, it can receive the AI training data corresponding to the UCI at the corresponding time-frequency position, thereby determining the correspondence between the AI training data and the PMI; or, the terminal passes PUSCH Report AI training data, and the AI training data also carries CSI report configuration ID.
  • the base station obtains one or more AI training data by receiving PUSCH, and finds the corresponding CSI based on the CSI report configuration ID carried by the AI training data, and then It is determined that the PMI in the CSI is the coded information corresponding to the AI training data.
  • the execution subject may be an information transmission device.
  • an information transmission device performing an information transmission method is used as an example to illustrate the information transmission device provided by the embodiment of the present application.
  • An information transmission device provided by an embodiment of the present application can be a device in a terminal. As shown in Figure 4, the information transmission device 400 can include the following modules:
  • Collection module 401 used to collect artificial intelligence AI training data, wherein the AI training data includes the first channel information of the target downlink channel, and the AI training data is used to train the first AI network model and/or the second AI network Model, the first AI network model is used to process the second channel information into the first channel characteristic information, and the second AI network model is used to restore the first channel characteristic information into the second channel information,
  • the second channel information is the channel information of the target downlink channel;
  • the first sending module 402 is used to send the AI training data to the first device.
  • the information transmission device 400 also includes:
  • the second receiving module is configured to receive first instruction information from the first device, where the first instruction information is used to instruct the terminal to collect the AI training data.
  • the information transmission device 400 also includes:
  • a third receiving module configured to receive first configuration information from the first device, where the first configuration information is used to configure target time domain and/or frequency domain resources;
  • the collection module 401 is specifically used for:
  • the information transmission device 400 also includes:
  • the second sending module is configured to send target capability information to the first device, where the target capability information indicates the information reporting capability and/or data collection capability supported by the terminal.
  • the target capability information includes at least one of the following:
  • the second indication information is used to indicate whether the terminal supports collecting the AI training data
  • the total number of ports of the terminal is the total number of ports of the terminal.
  • the information transmission device 400 also includes:
  • a third sending module configured to send first information to the first device when it is determined that collection of the AI training data is supported based on the status information of the terminal, where the first information instructs the terminal to collect and Report the AI training data;
  • the fourth sending module is used for the terminal to send second information to the first device when it is determined that the collection of the AI training data is not supported based on the status information of the terminal, and the second information indicates that the AI training data is not supported.
  • the terminal will not collect or report the AI training data.
  • the collection module 401 is specifically used to:
  • AI training data is collected based on the first port corresponding to the first reference signal resource or the CSI-RS port corresponding to the CSI-RS resource, where the first port includes a preconfigured first reference for collecting the AI training data.
  • a port corresponding to the signal resource, and the CSI-RS port includes a port used for CSI measurement.
  • the AI training data is carried in at least one of the following:
  • First uplink control information UCI the first UCI carries a CSI report of the target downlink channel, and the CSI report is related to the first channel information;
  • a second UCI, the second UCI does not carry the CSI report of the target downlink channel
  • Target physical uplink shared channel PUSCH Target physical uplink shared channel
  • the AI training data includes first channel information in the CSI report, where the first channel information in the CSI report is precoding information.
  • the information transmission device 400 further includes:
  • a first processing module configured to perform first processing on the first channel information based on the first AI network model to obtain second channel characteristic information
  • a fifth sending module configured to send the second channel characteristic information to the first device, where the AI training data and the second channel characteristic information are used to train the second AI network model.
  • the AI training data carries first identification information of the second channel characteristic information
  • the first sending module 402 is specifically used for:
  • the AI training data is sent to the first device on the time domain and/or frequency domain resources corresponding to the first identification information of the second channel characteristic information.
  • the AI training data includes a precoding matrix determined based on a first codebook, a first parameter of the first codebook is greater than a preset parameter, and the first parameter includes at least one of the following:
  • the first channel information is at least one of a precoding matrix and a channel matrix.
  • the AI training data includes at least one of the following:
  • the strongest M-layer precoding matrix among the N-layer precoding matrices, the channel information of the target downlink channel includes the N-layer precoding matrix, N is a positive integer, and M is a positive integer less than or equal to N;
  • the precoding matrix indicated by third indication information where the third indication information comes from the first device
  • the precoding matrix indicated by fourth indication information where the fourth indication information is the indication information sent by the terminal to the first device in advance;
  • a precoding matrix that satisfies a preset condition a precoding matrix that satisfies a preset condition.
  • the preset conditions include at least one of the following:
  • the channel quality indicator CQI is greater than or equal to the first threshold
  • the signal to interference plus noise ratio SINR is greater than or equal to the second threshold
  • the characteristic value is greater than or equal to the third threshold
  • the singular value is greater than or equal to the fourth threshold.
  • the information transmission device 400 provided by the embodiment of the present application can implement various processes implemented by the terminal in the method embodiment as shown in Figure 2, and can achieve the same beneficial effects. To avoid duplication, details will not be described here.
  • the information transmission device 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.
  • NAS Network Attached Storage
  • the execution subject may be an AI network model training device.
  • the AI network model training device executed by the AI network model training method is used as an example to illustrate the AI network model training device provided by the embodiment of the present application.
  • An AI network model training device provided by an embodiment of the present application can be a device in the first device. As shown in Figure 5, the AI network model training device 500 can include the following modules:
  • the first receiving module 501 is configured to receive AI training data from the terminal, where the AI training data includes first channel information of the target downlink channel;
  • Training module 502 used to train the first AI network model and/or the second AI network model according to the AI training data, wherein the first AI network model is used to process the second channel information into the first channel characteristic information.
  • the second AI network model is used to restore the first channel characteristic information to the second channel information, where the second channel information is the channel information of the target downlink channel.
  • the AI network model training device 500 further includes:
  • a sixth sending module is configured to send relevant information of the first AI network model to the terminal.
  • the AI network model training device 500 also includes:
  • a seventh sending module is configured to send first instruction information to the terminal, where the first instruction information is used to instruct the terminal to collect the AI training data.
  • the AI network model training device 500 also includes:
  • the eighth sending module is configured to send first configuration information to the terminal, where the first configuration information is used to configure target time domain and/or frequency domain resources, where the AI training data is Collected from domain and/or frequency domain resources.
  • the AI network model training device 500 also includes:
  • a fourth receiving module configured to receive target capability information from the terminal, where the target capability information indicates the information reporting capability and/or data collection capability supported by the terminal;
  • the seventh sending module or the eighth sending module is specifically used for:
  • first indication information or first configuration information is sent to the terminal.
  • the target capability information includes at least one of the following:
  • the second indication information is used to indicate whether the terminal supports collecting the AI training data
  • the total number of ports of the terminal is the total number of ports of the terminal.
  • the AI network model training device 500 also includes:
  • a fifth receiving module configured to receive first information from the terminal, where the first information instructs the terminal to collect and report the AI training data
  • a sixth receiving module is configured to receive second information from the terminal, where the second information indicates that the terminal will not collect or report the AI training data.
  • the AI network model training device 500 also includes:
  • a configuration module configured to configure a first port corresponding to the first reference signal resource and/or a CSI-RS port corresponding to the CSI-RS resource for the terminal, wherein the first port is dedicated to collecting the AI training data.
  • the port corresponding to the first reference signal resource, and the CSI-RS port is a port used for CSI measurement and collection of the AI training data.
  • the AI training data is carried in at least one of the following:
  • First uplink control information UCI the first UCI carries a CSI report of the target downlink channel, and the CSI report is related to the first channel information;
  • a second UCI, the second UCI does not carry the CSI report of the target downlink channel
  • Target physical uplink shared channel PUSCH Target physical uplink shared channel
  • the AI training data includes first channel information in the CSI report, where the first channel information in the CSI report is precoding information.
  • the AI network model training device 500 when the terminal has a first AI network model, also includes include:
  • a sixth receiving module configured to receive the second channel characteristic information from the terminal, wherein the second channel characteristic information is obtained by performing the first processing on the first channel information based on the first AI network model.
  • the training module 502 is specifically used for:
  • the second AI network model is trained according to the AI training data and the second channel characteristic information.
  • the AI training data carries the first identification information of the second channel characteristic information, or the first receiving module 501 specifically uses:
  • the AI training data from the terminal is received on time domain and/or frequency domain resources corresponding to the first identification information of the second channel characteristic information.
  • the AI training data includes a precoding matrix determined based on a first codebook, a first parameter of the first codebook is greater than a preset parameter, and the first parameter includes at least one of the following:
  • the first channel information is at least one of a precoding matrix and a channel matrix.
  • the AI training data includes at least one of the following:
  • the strongest M-layer precoding matrix among the N-layer precoding matrices, the channel information of the target downlink channel includes the N-layer precoding matrix, N is a positive integer, and M is a positive integer less than or equal to N;
  • the precoding matrix indicated by third indication information where the third indication information comes from the first device
  • the precoding matrix indicated by fourth indication information where the fourth indication information is the indication information sent by the terminal to the first device in advance;
  • a precoding matrix that satisfies a preset condition a precoding matrix that satisfies a preset condition.
  • the preset conditions include at least one of the following:
  • the channel quality indicator CQI is greater than or equal to the first threshold
  • the signal to interference plus noise ratio SINR is greater than or equal to the second threshold
  • the characteristic value is greater than or equal to the third threshold
  • the singular value is greater than or equal to the fourth threshold.
  • the AI network model training device 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 network side device.
  • the terminal may include but is not limited to the types of network side devices 12 listed above, which are not specifically limited in the embodiments of this application.
  • the AI network model training device 500 provided by the embodiment of the present application can implement the method embodiment shown in Figure 3
  • Each process implemented by the first device can achieve the same beneficial effects, and will not be described again in order to avoid repetition.
  • 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, for example.
  • the communication device 600 is a terminal, when the program or instruction is executed by the processor 601, each step of the method embodiment shown in Figure 2 is implemented, and the same technical effect can be achieved.
  • the communication device 600 is the first device, when the program or instruction is executed by the processor 601, each step of the method embodiment shown in Figure 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 communication interface is used to collect artificial intelligence AI training data and send the AI training data to the first device, where the AI training data includes target downlink
  • the first channel information of the channel the AI training data is used to train the first AI network model and/or the second AI network model
  • the first AI network model is used to process the second channel information into the first channel characteristic information
  • the second AI network model is used to restore the first channel characteristic information to the second channel information, where the second channel information is the channel information of the target downlink channel.
  • 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 (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 .
  • 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.
  • save Memory 709 may include volatile memory or nonvolatile memory, or memory 709 may include both volatile and nonvolatile memory.
  • the 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 modem processor may not be integrated into the processor 710.
  • the radio frequency unit 701 is used to collect artificial intelligence AI training data, wherein the AI training data includes the first channel information of the target downlink channel, and the AI training data is used to train the first AI network model and/or the second AI network model, the first AI network model is used to process the second channel information into the first channel characteristic information, and the second AI network model is used to restore the first channel characteristic information into the second channel Information, the second channel information is the channel information of the target downlink channel;
  • the radio frequency unit 701 is also used to send the AI training data to the first device.
  • the radio frequency unit 701 before performing the collection of AI training data, the radio frequency unit 701 is also used to:
  • the radio frequency unit 701 before performing the collection of AI training data, the radio frequency unit 701 is also used to:
  • the collection of AI training data performed by the radio frequency unit 701 includes:
  • the radio frequency unit 701 collects AI training data in the target time domain and/or frequency domain resources.
  • the radio frequency unit 701 before performing the receiving of the first indication information or the first configuration information from the first device, the radio frequency unit 701 is also configured to:
  • Target capability information indicates the information reporting capability and/or data collection capability supported by the terminal.
  • the target capability information includes at least one of the following:
  • the second indication information is used to indicate whether the terminal supports collecting the AI training data
  • the total number of ports of the terminal is the total number of ports of the terminal.
  • the processor 710 is also configured to control the radio frequency unit 701 to send the first information to the first device when it is determined that collection of the AI training data is supported according to the status information of the terminal.
  • a message instructs the terminal to collect and report the AI training data;
  • the processor 710 is also configured to control the radio frequency unit 701 to send second information to the first device when it is determined according to the status information of the terminal that collection of the AI training data is not supported, and the second information indicates The terminal will not collect or report the AI training data.
  • the collection of AI training data performed by the radio frequency unit 701 includes:
  • the radio frequency unit 701 collects AI training data based on the first port corresponding to the first reference signal resource or the CSI-RS port corresponding to the CSI-RS resource, where the first port includes a preconfigured port for collecting the AI training data.
  • a port corresponding to the first reference signal resource, and the CSI-RS port includes a port used for CSI measurement.
  • the AI training data is carried in at least one of the following:
  • First uplink control information UCI the first UCI carries a CSI report of the target downlink channel, and the CSI report is related to the first channel information;
  • a second UCI, the second UCI does not carry the CSI report of the target downlink channel
  • Target physical uplink shared channel PUSCH Target physical uplink shared channel
  • the AI training data includes first channel information in the CSI report, where the first channel information in the CSI report is precoding information.
  • the terminal has the first AI network model:
  • the processor 710 is also configured to perform first processing on the first channel information based on the first AI network model to obtain second channel characteristic information;
  • the radio frequency unit 701 is also configured to send the second channel characteristic information to the first device, where the AI training data and the second channel characteristic information are used to train the second AI network model.
  • the AI training data carries first identification information of the second channel characteristic information
  • the sending of the AI training data to the first device performed by the radio frequency unit 701 includes:
  • the AI training data is sent to the first device on the time domain and/or frequency domain resources corresponding to the first identification information of the second channel characteristic information.
  • the AI training data includes a precoding matrix determined based on a first codebook, a first parameter of the first codebook is greater than a preset parameter, and the first parameter includes at least one of the following:
  • the first channel information is at least one of a precoding matrix and a channel matrix.
  • the AI training data includes at least one of the following:
  • the strongest M-layer precoding matrix among the N-layer precoding matrices, the channel information of the target downlink channel includes the N-layer precoding matrix, N is a positive integer, and M is a positive integer less than or equal to N;
  • the precoding matrix indicated by third indication information where the third indication information comes from the first device
  • the precoding matrix indicated by fourth indication information where the fourth indication information is the indication information sent by the terminal to the first device in advance;
  • a precoding matrix that satisfies a preset condition a precoding matrix that satisfies a preset condition.
  • the preset conditions include at least one of the following:
  • the channel quality indicator CQI is greater than or equal to the first threshold
  • the signal to interference plus noise ratio SINR is greater than or equal to the second threshold
  • the characteristic value is greater than or equal to the third threshold
  • the singular value is greater than or equal to the fourth threshold.
  • the terminal 700 provided by the embodiment of the present application can implement various processes performed by the information transmission device as shown in Figure 3, and can achieve the same beneficial effects. To avoid duplication, the details will not be described again.
  • Embodiments of the present application also provide a network side device.
  • the network side device may be an access network device or a core network device.
  • the network side device includes a communication interface and a processor, where the communication interface is used to receive AI from the terminal. training data, wherein the AI training data includes first channel information of the target downlink channel; the processor is configured to train a first AI network model and/or a second AI network model according to the AI training data, wherein the The first AI network model is used to process the second channel information into the first channel characteristic information.
  • the second AI network model is used to restore the first channel characteristic information into the second channel information.
  • the second The channel information is the channel information of the target downlink channel.
  • This network side device embodiment corresponds to the method embodiment shown in Figure 3.
  • Each implementation process and implementation manner of the method embodiment shown in Figure 3 can be applied to this network side device embodiment, and can achieve the same technical effect.
  • 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 shown in Figure 2 or Figure 3 is implemented. , and can achieve the same technical effect, so to avoid repetition, they 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.
  • the implementation is as shown in Figure 2 or Figure 3. Each process of the method embodiment is shown, and the same technical effect can be achieved. To avoid repetition, the details will not be described here.
  • 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.
  • the embodiment of the present application further provides a computer program/program product, the computer program/program product is stored in a storage medium, and the computer program/program product is executed by at least one processor to implement Figure 2 or Figure 3
  • a computer program/program product is stored in a storage medium
  • the computer program/program product is executed by at least one processor to implement Figure 2 or Figure 3
  • 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 information transmission method as shown in Figure 2.
  • the network side device can be used to perform the steps of the information transmission method as shown in Figure 3.
  • 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 the existing technology.
  • 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)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请公开了一种信息传输方法、AI网络模型训练方法、装置和通信设备，属于通信技术领域，本申请实施例的信息传输方法包括：终端采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；所述终端向第一设备发送所述AI训练数据。

Description

信息传输方法、AI网络模型训练方法、装置和通信设备
相关申请的交叉引用
本申请主张在2022年09月07日在中国提交的中国专利申请No.202211091873.9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，具体涉及一种信息传输方法、人工智能(Artificial Intelligence，AI)网络模型训练方法、装置和通信设备。
背景技术
在相关技术中，对借助AI网络模型来传输信道特征信息的方法进行了研究。
该AI网络模型可以包括编码部分(即编码AI网络模型)和解码部分(即解码AI网络模型)，编码AI网络模型用于将信道信息编码成信道特征信息，解码AI网络模型用于将编码AI网络模型输出的信道特征信息恢复成信道信息，这样，为了使编码AI网络模型与解码AI网络模型匹配，该编码AI网络模型和解码AI网络模型通常在同一设备中进行联合训练，然后将联合训练得到的编码AI网络模型传输至终端，将联合训练得到解码AI网络模型传输至基站。
在相关技术中，基站根据预先获取的训练数据来训练编码AI网络模型和解码AI网络模型，由于预先获取的训练数据与实际信道并不一致，会造成训练得到的编码AI网络模型和解码AI网络模型的与实际信道的匹配程度比较低，从而降了编码AI网络模型和解码AI网络模型处理实际的信道信息的准确度。
发明内容
本申请实施例提供一种信息传输方法、AI网络模型训练方法、装置和通信设备，使得终端能够向基站上报实际估计的信道信息，以使基站将该信道信息作为AI训练数据来训练准确性更高的编码AI网络模型和解码AI网络模型。
第一方面，提供了一种信息传输方法，该方法包括：
终端采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；
所述终端向第一设备发送所述AI训练数据。
第二方面，提供了一种信息传输装置，应用于终端，该装置包括：
采集模块，用于采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；
第一发送模块，用于向第一设备发送所述AI训练数据。
第三方面，提供了一种信息传输方法，包括：
第一设备接收来自终端的AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息；
所述第一设备根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型，其中，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息。
第四方面，提供了一种信息传输装置，应用于第一设备，该装置包括：
第一接收模块，用于接收来自终端的AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息；
训练模块，用于根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型，其中，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息。
第五方面，提供了一种通信设备，该通信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一方面或第三方面所述的方法的步骤。
第六方面，提供了一种通信设备，包括处理器及通信接口，其中，所述通信接口用于采集人工智能AI训练数据，以及向第一设备发送所述AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；或者，
所述通信接口用于接收来自终端的AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息；所述处理器用于根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型，其中，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息， 所述第二信道信息为所述目标下行信道的信道信息。
第七方面，提供了一种通信***，包括：终端和网络侧设备，所述终端可用于执行如第一方面所述的信息传输方法的步骤，所述网络侧设备可用于执行如第三方面所述的信息传输方法的步骤。
第八方面，提供了一种可读存储介质，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第一方面所述的方法的步骤，或者实现如第三方面所述的方法的步骤。
第九方面，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方法，或实现如第三方面所述的方法。
第十方面，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信息传输方法的步骤，或者所述计算机程序/程序产品被至少一个处理器执行以实现如第三方面所述的信息传输方法的步骤。
在本申请实施例中，终端采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；所述终端向第一设备发送所述AI训练数据。这样，终端向第一设备上报的AI训练数据包括基于实际估计到的信道信息，这样，第一设备在根据该AI训练数据训练第一AI网络模型和/或第二AI网络模型的过程中，能够提升训练得到的第一AI网络模型和/或第二AI网络模型与目标下行信道的匹配程度。在利用第一AI网络模型将目标下行信道的信道信息压缩成信道特征信息，和/或，利用第二AI网络模型将目标下行信道的信道特征信息恢复成信道信息的过程中，能够提升第一AI网络模型的压缩编码精确度和/或提升第二AI网络模型的解码精确度和/或提升第与AI网络模型和第二AI网络模型之间的匹配程度，进而能够降低信道信息上报过程占用的资源，以及提升该上报过程的精确度。
附图说明
图1是本申请实施例能够应用的一种无线通信***的结构示意图；
图2是本申请实施例提供的一种信息传输方法的流程图；
图3是本申请实施例提供的一种AI网络模型训练方法的流程图；
图4是本申请实施例提供的一种信息传输装置的结构示意图；
图5是本申请实施例提供的一种AI网络模型训练装置的结构示意图；
图6是本申请实施例提供的一种通信设备的结构示意图；
图7是本申请实施例提供的一种终端的硬件结构示意图。
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)***，还可用于其他无线通信***，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***。本申请实施例中的术语“***”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***和无线电技术，也可用于其他***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)***，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***应用以外的应用，如第6代(6th Generation，6G)通信***。
图1示出本申请实施例可应用的一种无线通信***的框图。无线通信***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***中的基站为例进行介绍，并不限定基站的具体类型。
人工智能目前在各个领域获得了广泛的应用。AI网络模型有多种实现方式，例如神经网络、决策树、支持向量机、贝叶斯分类器等。本申请以神经网络为例进行说明，但是并不限定AI网络模型的具体类型。
在相关技术中，在终端侧采用编码AI网络模型来对信道信息进行压缩编码，并上报压缩编码后的信道特征信息，在网络侧则可以采用解码AI网络模型来对接收的信道特征信息进行解码处理，以恢复信道信息。其中，编码AI网络模型需要与解码AI网络模型匹配，在相关技术中，通常是在网络侧联合训练编码AI网络模型和解码AI网络模型，以使两者达到合适的匹配度。其中，训练编码AI网络模型和解码AI网络模型的AI训练数据通常是预先获取的数据，或者是专用于采集训练数据的终端或网络侧设备所采集到的信道信息，基于该AI训练数据所训练得到编码AI网络模型和解码AI网络模型之后，网络侧设备再将编码AI网络模型下发给终端，以使终端基于网络侧设备下发的编码AI网络模型来编码信道信息。但是，网络侧设备基于预先获取的数据或专用于采集训练数据的终端或网络侧设备所采集到的信道信息所训练得到的编码AI网络模型和解码AI网络模型可能存在与终端的实际信道信息不匹配的情况，此时，可能出现编码AI网络模型对终端的实际信道信息的压缩程度较低，使得压缩后的信道特征信息占用资源较大，或者使得编码AI网络模型和解码AI网络模型对终端的实际信道信息的处理不够准确，降低了信道信息上报的精确度。
而本申请实施例中，终端会将估计到的实际信道的信道信息作为AI训练数据上报给训练编码AI网络模型(即本申请实施例中的第一AI网络模型)和/或解码AI网络模型(即本申请实施例中的第二AI网络模型)的第一设备，这样，第一设备基于该AI训练数据训练得到的编码AI网络模型和/或解码AI网络模型能够与终端实际的信道状态相匹配，从而能够降低信道信息上报过程占用的资源，以及提升信道信息上报过程的精确度。
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信息传输方法、信息传输装置及通信设备等进行详细地说明。
请参阅图2，本申请实施例提供的一种信息传输方法，其执行主体是终端，例如：如图1中列举的各种类型的终端11。如图2所示，该终端执行的信息传输方法可以包括以下步骤：
步骤201、终端采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信 道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息。
其中，AI训练数据包括目标下行信道的第一信道信息，可以是终端将实际估计到的目标下行信道的信道信息作为AI训练数据，或者是终端对实际估计到的目标下行信道的信道信息进行一定的预处理(如：对信道信息进行筛选、量化、归一化、正交化等处理方式中的至少一项)，并将处理后的信道信息作为AI训练数据。例如：所述第一信道信息为预编码矩阵和信道矩阵中的至少一项，为了便于说明，本申请实施例汇总以第一信道信息为预编码矩阵为例进行举例说明，在此不构成具体限定。
作为一种可选的实施方式，所述AI训练数据包括以下至少一项：
N层预编码矩阵中最强的M层预编码矩阵，所述目标下行信道的信道信息包括所述N层预编码矩阵，N为正整数，且M为小于或等于N的正整数；
所述N层预编码矩阵；
所述N层预编码矩阵中，第三指示信息指示的预编码矩阵，所述第三指示信息来自所述第一设备；
所述N层预编码矩阵中，第四指示信息指示的预编码矩阵，所述第四指示信息为所述终端预先发送至所述第一设备的指示信息；
所述N层预编码矩阵中的，满足预设条件的预编码矩阵。
选项一，N层预编码矩阵中最强的M层预编码矩阵，可以是目标下行信道的秩(Rank)N大于1，此时，该目标下行信道的预编码矩阵包括N层，其中，强度越大的层越能够有效反映目标下行信道的信道状态，这样，从该N层预编码矩阵中选择最强(如特征值最大，获知奇异值最大，或者是功率占比最大，或者是信号与干扰加噪声比(Signal to Interference plus Noise Ratio，SINR)最大)的M层作为AI训练数据。例如：终端上报最强层(layer)的预编码矩阵。
值得提出的是，终端在复用常规的信道状态信息(Channel state information,CSI)报告作为AI训练数据的情况下，可以将CSI报告中与层0(layer 0)对应的预编码矩阵作为AI训练数据。
其相较于选择整个N层预编码矩阵而言，能够降低AI训练数据的数据量，且尽可能的保留N层预编码矩阵中的有效信息。
选项二，终端也可以将N层预编码矩阵作为AI训练数据，这样，能够精确的反映目标下行信道的真实信道状态。
选项三，终端可以在发送AI训练数据之前，接收上述第三指示信息，该第三指示信息可以指示终端需要上报的预编码矩阵，这样，终端只需要按照第一设备的指示上报指定层的预编码矩阵即可。
选项四，终端可以选择上报哪些层的预编码矩阵，并上报第四指示信息，以指示终端上报的预编码矩阵是哪些层的。其中，终端可以在发送AI训练数据之前或者之后或者同时发送第四指示信息，在此不作具体限定。
选项五，终端可以根据N层预编码矩阵中的各层是否满足预设条件来选择需要上的预编码矩阵。其中，满足预设条件的预编码矩阵能够更有效的反映目标下行信道的信道状态。
可选地，所述预设条件包括以下至少一项：
信道质量指示信道质量指示(Channel quality indicator,CQI)大于或等于第一阈值；
信号与干扰加噪声比SINR大于或等于第二阈值；
特征值大于或等于第三阈值；
奇异值大于或等于第四阈值。
本实施方式中，终端能够根据预设条件来筛选上报哪些层或那一层的预编码矩阵，其相较于选择整个N层预编码矩阵而言，能够降低AI训练数据的数据量，且尽可能的保留N层预编码矩阵中的有效信息。
在一种可能的实现方式中，终端在根据预设条件来筛选上报哪些层或那一层的预编码矩阵，或者上报N层预编码矩阵中最强的M层预编码矩阵的情况下，终端还可以向第一设备发送第四指示信息，以指示终端上报的预编码矩阵是N层预编码矩阵中的哪些层或那一层。
或者，终端可以将N层预编码矩阵中需要上报的层的预编码矩阵保留元素真值，并将N层预编码矩阵中不需要上报的层的预编码矩阵中的元素用0或者其他取值替代，这样，第一设备能够根据N层预编码矩阵中各个元素的位置来确定该元素对应物理意义。
步骤202、所述终端向第一设备发送所述AI训练数据。
其中，第一设备可以是用于训练AI网络模型的任意设备，例如：接入网设备(如基站)或核心网设备，为了便于说明，本申请实施例中以第一设备为基站为例进行举例说明。
上述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，可以是第一AI网络模型为编码AI网络侧模型，该编码AI网络侧模型用于将信道信息压缩编码成信道特征信息。与之相对应的，上述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，可以是第二AI网络模型为与第一AI网络侧模型匹配的解码AI网络模型，该解码AI网络模型用于对编码AI网络侧模型处理后的信道特征信息进行解码处理，以恢复信道信息。
在实施中，上述第一设备可以根据AI训练数据来训练或调整第一AI网络模型和/或第二AI网络模型，例如：假设第一AI网络模型和第二AI网络模型为神经网络模型，则训练第一AI网络模型和/或第二AI网络模型的方式包括以下三种：
方案一：神经网络模型包括由终端的编码器和基站的解码器组成的联合神经网络模型，且由网络侧设备进行联合训练，在训练完成之后，基站将编码器的AI网络模型发送给终端。本方案中，第一设备包括基站。
方案二：终端训练编码AI网络模型，基站训练解码AI网络模型，并通过匹配过程进行匹配，使得终端的编码AI网络模型和基站的解码AI网络模型相互匹配。其中，匹配过程可以包括：终端采用编码AI网络模型对信道信息进行编码，得到编码数据(即信道特征信息)，并把原始的信道信息，以及终端的编码数据都发送给基站，基站以采用解码AI网络模型把终端的编码数据译码成为对应的信道信息为目的，调整解码AI网络模型。本方案中，第一设备包括基站。
方案三：终端训练编码AI网络模型和解码AI网络模型，并采用编码AI网络模型将原始的信道信息进行编码，得到编码数据，然后把原始的信道信息，以及终端的编码数据都发送给基站，基站则根据该信道信息和终端的编码数据训练解码AI网络模型，此时基站不需要训练编码AI网络模型。本方案中，第一设备包括基站。
在实施中，终端上报AI训练数据的时间和/或频率资源可以是网络侧设备指示的或者是协议中约定的，例如：周期性地采集并上报AI训练数据，或者，在指定时间采集并上报AI训练数据。
作为一种可选的实施方式，在所述终端采集AI训练数据之前，所述方法还包括：
所述终端接收来自所述第一设备的第一指示信息，其中，所述第一指示信息用于指示所述终端采集所述AI训练数据。
在实施中，第一指示信息可以指示终端采集并上报所述AI训练数据，例如：基站通过下行控制信息(Downlink Control Information，DCI)指示终端激活AI训练数据的采集和上报，这样，终端在接收到该第一指示信息是，可以立即开始采集和上报AI训练数据。
在一种可能的实现方式中，终端在接收到上述第一指示信息后，可以根据自身状态来判断，是否支持采集和上报AI训练数据。
可选地，所述信息传输方法还包括：
所述终端在根据所述终端的状态信息，确定支持采集所述AI训练数据的情况下，向所述第一设备发送第一信息，所述第一信息指示所述终端采集并上报所述AI训练数据；
和/或，
所述终端在根据所述终端的状态信息，确定不支持采集所述AI训练数据的情况下，向所述第一设备发送第二信息，所述第二信息指示所述终端不会采集或上报所述AI训练数据。
其中，终端的状态信息可以包括终端的随时间变化的状态信息，例如：位置、移动速度、剩余电量、用户设置情况等。
本实施方式中，终端在接收到第一指示信息或第一配置信息的情况下，可以根据自身的实际情况来判断当前是否支持采集和上报所述AI训练数据，例如：在用户设置终端不采集和上报所述AI训练数据时，确定终端不支持采集和上报所述AI训练数据；在终端的电量较低时，确定终端不支持采集和上报所述AI训练数据；在终端移动过程中，确定终端不支持采集和上报所述AI训练数据等等。
在一种实施方式中，终端向第一设备发送第一信息后，第一设备便知道该终端能够采集和上报所述AI训练数据，从而只需要等待该终端上报AI训练数据即可。
在另一种实施方式中，终端向第一设备发送第二信息后，第一设备便知道该终端不会上报AI训练数据，从而可以重新选择其他终端来采集和上报AI训练数据，或者是采用其他方式获取AI训练数据。例如：向其他可能支持采集AI训练数据的终端发送第一指示信息等。
作为一种可选的实施方式，在所述终端采集AI训练数据之前，所述方法还包括：
所述终端接收来自所述第一设备的第一配置信息，其中，所述第一配置信息用于配置目标时域和/或频域资源；
所述终端采集AI训练数据，包括：
所述终端在所述目标时域和/或频域资源采集AI训练数据。
其中，第一配置信息可以配置终端在指定的时刻或按照指定的周期采集并上报所述AI训练数据，例如：基站通过无线资源控制(Radio Resource Control，RRC)信令或媒体接入控制层控制单元(Medium Access Control Control Element，MAC CE)配置终端采集AI训练数据的时刻或周期，这样，终端可以在对应的时刻激活AI训练数据的采集和上报。
本实施方式与上述终端基于来自第一设备的第一指示信息来采集AI训练数据的方式相似，不同之处在于，本实施方式中，可以通过协议约定或网络侧预配置的方式来配置终端采集和/或上报AI训练数据的时域和/或频域资源。
作为一种可选的实施方式，在所述终端接收来自所述第一设备的第一指示信息或第一配置信息之前，所述方法还包括：
所述终端向所述第一设备发送目标能力信息，其中，所述目标能力信息指示所述终端支持的信息上报能力和/或数据采集能力。
其中，目标能力信息可以是终端的通信能力信息，该目标能力信息用于供第一设备判断该终端是否支持采集AI训练数据，从而向支持采集AI训练数据的终端发送第一指示信息或第一配置信息。
可选地，所述目标能力信息包括以下至少一项：
第二指示信息，所述第二指示信息用于指示所述终端是否支持采集所述AI训练数据；
所述终端支持的最大信道状态信息(Channel State Information，CSI)-参考信号(Reference Signal，RS)端口数；
所述终端的全部端口数。
选项一，第二指示信息能够明确的指示终端是否支持采集所述AI训练数据，例如：终端根据自身的能力和/状态信息来确定的该终端是否支持采集和上报AI训练数据。
选项二，上述终端支持的最大CSI-RS端口数可以作为第一设备判断终端是否支持采集所述AI训练数据的依据，例如：第一设备可以根据终端支持的最大CSI-RS端口数来判断该终端能否基于现有的CSI-RS端口来采集AI训练数据。
选项三，上述终端的全部端口数可以作为第一设备判断终端是否支持采集所述AI训练数据的依据，例如：第一设备可以在终端的全部端口数大于或等于采集AI训练数据所需的端口数的情况下，判断该终端支持采集AI训练数据。
本实施方式中，第一设备可以事先获取终端的能力信息，从而可以根据该能力信息对终端进行筛选，并向支持采集AI训练数据的终端发送第一指示信息或第一配置信息，这样，能够降低第一设备指示或配置不支持采集AI训练数据的终端采集AI训练数据的概率，从而节约了指示或配置不支持采集AI训练数据的终端采集AI训练数据的过程中所占用的资源，以及降低采集和上报AI训练数据的时延。
当然，第一设备也可以采集其他方式确定向哪一些或哪一个终端发送第一指示信息或第一配置信息，例如：获取终端的位置、资源配置情况等先验信息，并据此判断终端是否支持采集AI训练数据，从而向支持采集AI训练数据的终端发送第一指示信息或第一配置信息。
作为一种可选的实施方式，所述终端采集AI训练数据，包括：
所述终端基于第一参考信号资源对应的第一端口或CSI-RS资源对应的CSI-RS端口采集AI训练数据，其中，所述第一端口包括预先配置的用于采集所述AI训练数据的第一参考信号资源对应的端口，所述CSI-RS端口包括用于CSI测量的端口。
在一种可能的实现方式中，第一端口可以是基站配置的专用于采集AI训练数据的端口，这样，终端可以对该第一端口对应的第一参考信号资源进行信道估计，以实现采集AI训练数据。其中，第一参考信号资源可以是CSI-RS的资源，也可能是其他类型的参考信号的资源，在此不作具体限定。本实施方式下，基站可以为终端配置额外的第一端口，用于采集AI训练数据的信道估计。
在另一种可能的实现方式中，CSI-RS端口可以表示现有的CSI测量所使用的端口，即终端可以复用正常的CSI测量的CSI-RS端口来测量
在一种可能的实现方式中，所述AI训练数据携带于以下至少一项：
第一上行控制信息(Uplink Control Information，UCI)，所述第一UCI携带所述目标下行信道的CSI报告，所述CSI报告与所述第一信道信息相关；
第二UCI，所述第二UCI不携带所述目标下行信道的CSI报告；
目标物理上行共享信道(Physical Uplink Shared Channel，PUSCH)。
选项一，在AI训练数据携带于第一UCI的情况下，可以是将AI训练数据可以作为CSI报告的一部分，通过第一UCI进行上报，此时，所述CSI报告与所述第一信道信息相关可以理解为：CSI报告中的第一信道信息可以是该CSI报告的原始信道信息，而CSI报告可以是码本形式的信道信息或预编码的信道信息或者是基于第一AI网络模型编码处理后的信道特征信息。
可选地，所述AI训练数据包括所述CSI报告内的第一信道信息，其中，所述CSI报告内的第一信道信息为预编码信息。
本实施方式下，在终端使用码本进行CSI上报的情况下，可以复用该码本形式的CSI报告作为AI训练数据，此时，终端可以不再独立上报AI训练数据，基站可以将接收到的CSI报告作为AI训练数据。
选项二，上述第二UCI为不携带所述目标下行信道的CSI报告的UCI，即AI训练数据通过UCI独立于CSI上报。
选项三，上述目标PUSCH携带AI训练数据，可以是将AI训练数据作为数据内容，通过PUSCH进行上报。
在实施中，终端可以基于以上选项一至选项三中的任一种或多种方式实现灵活地上报AI训练数据，在此不作具体限定。
作为一种可选的实施方式，在所述终端具有第一AI网络模型的情况下，所述方法还包括：
所述终端基于所述第一AI网络模型对所述第一信道信息进行第一处理，得到第二信道特征信息；
所述终端向所述第一设备发送所述第二信道特征信息，其中，所述AI训练数据和所述第二信道特征信息用于训练所述第二AI网络模型。
其中，终端预先配置或训练得到了第一AI网络模型。
本实施方式中，终端除了向第一设备发送原始的信道信息之外，还向第一设备发送采用第一AI网络模型对所述第一信道信息进行压缩编码处理后得到的第二信道特征信息。这样，第一设备可以将第二信道特征信息输入第二AI网络模型，并以第二AI网络模型输出第一信道信息为目的训练或调整第二AI网络模型，其中，基于AI训练数据和第二信道特征信息对第二AI网络模型进行训练能够提升第二AI网络模型与终端已经具有的第一AI网络模型之间的匹配程度。
需要说明的是，第一信道信息与第二信道特征信息之间为一一对应的关系，即向第一AI网络模型输入一条第一信道信息，则该第一AI网络模型输出该第一信道信息对应的第二信道特征信息；若向第一AI网络模型输入另一条第一信道信息，则该第一AI网络模型输出该另一条第一信道信息对应的第二信道特征信息。为了使第一设备获知第一信道信息与第二信道特征信息之间的对应关系，终端可以将相互对应的第一信道信息与第二信道特征信息同时发送，或者在相互关联的时域和/或频域资源上分别传输第一信道信息和与该第一信道信息对应的第二信道特征信息，或者，在上报的第一信道信息中携带对应的第二信道特征信息的标识信息等。
可选地，所述AI训练数据中携带所述第二信道特征信息的第一标识信息；
或者，
所述终端向第一设备发送所述AI训练数据，包括：
所述终端向第一设备发送目标CSI报告，所述目标CSI报告携带所述第二信道特征信息和所述AI训练数据；或者，
所述终端在与所述第二信道特征信息的第一标识信息对应的时域和/或频域资源上，向第一设备发送所述AI训练数据。
情况一，上述终端向第一设备发送目标CSI报告，所述目标CSI报告携带所述第二信道特征信息和所述AI训练数据，可以是终端正在使用第一AI网络模型来进行CSI上报，且将对应的第一信道信息加入到CSI中与AI编码后的第二信道特征信息中一起上报。
基站在接收CSI报告之后，可以从中获得这个第一信道信息以及对应的AI编码后的第二信道特征信息，基站可以通过利用第二AI网络模型将第二信道特征信息恢复成第一信道信息为目的训练第二AI网络模型。
值得提出的是，在实施中，基站在同时获取第一信道信息和第二信道特征信息的情况下，还可以直接使用第一信道信息作为后续调度的信道信息，这样，可以提升调度过程的精确度。
情况二，上述终端在与所述第二信道特征信息的第一标识信息对应的时域和/或频域资源上，向第一设备发送所述AI训练数据的情况下，基站分别接收AI训练数据和第二信道特征信息，且可以根据接收到AI训练数据的时域和/或频域资源与第一标识信息之间的对应关系来确定该AI训练数据中的第一信道信息与哪一个第二信道特征信息对应，进而可以以第二AI网络模型将第二信道特征信息恢复成第一信道信息为目的训练该第二AI网络模型。
其中，第一标识可以是CSI报告配置(report configuration)的标识(Identity,ID)或者其他任意能够直接或间接标识第二信道特征信息的标识，在此不作具体限定。
情况三，上述AI训练数据中携带所述第二信道特征信息的第一标识信息的情况下，基站能够根据AI训练数据中携带的第一标识信息来确定该AI训练数据中的第一信道信息所对应的第二信道特征信息，进而可以以第二AI网络模型将第二信道特征信息恢复成第一信道信息为目的训练该第二AI网络模型。
在一种可能的实现方式中，若所述AI训练数据携带于第二UCI或目标PUSCH，且终端通过CSI报告上报第二信道特征信息，则该第二UCI或目标PUSCH可以在与所述AI训练数据对应的CSI报告相关的时域和/或频域资源上发送，这样，第一设备在该时域和/或频域资源上接收到AI训练数据时，能够确定该AI训练数据与时域和/或频域资源相关的CSI报告中的第二信道特征信息对应。
在另一种可能的实现方式中，若所述AI训练数据携带于第二UCI或目标PUSCH，且终端通过CSI报告上报第二信道特征信息，则该第二UCI或目标PUSCH中可以携带对应的CSI报告的第一标识信息，这样，第一设备在接收到第二UCI或目标PUSCH时，可以从中获取AI训练数据和该AI训练数据对应的CSI报告的第一标识信息，从而确定该AI训练数据与该第一标识信息所标识的CSI报告中的第二信道特征信息对应。
在一种可能的实现方式中，若所述AI训练数据携带于第一UCI，此时，该第一UCI中还携带该AI训练数据对应的第二信道特征信息，这样，第一设备在接收到第一UCI时， 可以从中获取相互对应的AI训练数据和第二信道特征信息。
作为一种可选的实施方式，所述AI训练数据包括基于第一码本确定的预编码矩阵，所述第一码本的第一参数大于预设参数，所述第一参数包括以下至少一项：
端口数、时延径个数、波束数量、非零系数比例。
其中，上述预设参数可以是现有的R16码本的端口数、时延径个数、波束数量、非零系数比例。
其中，R16码本中的端口数、时延径个数、波束数量、非零系数比例如下表1所示：
表1
其中，如上表1中Pv用于计算时延径个数，且R16码本中最多支持32个端口。
本实施方式中，第一码本的端口数、时延径个数、波束数量和非零系数比例中的至少一项可以大于如上表1中任一种组合的端口数、时延径个数、波束数量、非零系数比例。例如：本实施方式中的第一码本的第一参数可以是如下表2所示的参数组合：
表2
也就是说，本实施方式中，可以采用比R16码本的精确度更高的码本来上报AI训练数据，此时，该AI训练数据能够更加精确的描述目标下行信道的信道状态，从而基于该AI训练数据训练得到的AI网络模型与目标下行信道的实际信道状态更加匹配。
在本申请实施例中，终端采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二 AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；所述终端向第一设备发送所述AI训练数据。这样，终端向第一设备上报的AI训练数据包括基于实际估计到的信道信息，这样，第一设备在根据该AI训练数据训练第一AI网络模型和/或第二AI网络模型的过程中，能够提升训练得到的第一AI网络模型和/或第二AI网络模型与目标下行信道的匹配程度。在利用第一AI网络模型将目标下行信道的信道信息压缩成信道特征信息，和/或，利用第二AI网络模型将目标下行信道的信道特征信息恢复成信道信息的过程中，能够提升第一AI网络模型的压缩编码精确度和/或提升第二AI网络模型的解码精确度和/或提升第与AI网络模型和第二AI网络模型之间的匹配程度，进而能够降低信道信息上报过程占用的资源，以及提升该上报过程的精确度。
请参阅图3，本申请实施例提供的一种AI网络模型训练方法，其执行主体是第一设备。该第一设备可以是网络侧设备，例如：如图1所示实施例中列举的网络侧设备12或者是核心网设备，为了便于说明，本申请实施例中以第一设备是基站为例进行举例说明。
如图3所示，该第一设备执行的AI网络模型训练方法可以包括以下步骤：
步骤301、第一设备接收来自终端的AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息。
上述AI训练数据和第一信道信息的含义与如图2所示方法实施例中的AI训练数据和第一信道信息的含义相同，在此不作具体限定。
步骤302、所述第一设备根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型，其中，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息。
在实施中，基站可以根据AI训练数据训练编码AI网络模型，或者根据AI训练数据训练解码AI网络模型，或者根据AI训练数据联合训练编码AI网络模型和解码AI网络模型。
本申请实施例中，基站能够从终端获取对目标下行信道进行信道估计得到的信道信息来训练适用于该目标下行信道的编码和/或解码AI网络模型，能够提升训练得到的第一AI网络模型和/或第二AI网络模型与目标下行信道的匹配程度。
作为一种可选的实施方式，在所述第一设备根据所述AI训练数据训练第一AI网络模型的情况下，所述方法还包括：
所述第一设备向所述终端发送所述第一AI网络模型的相关信息。
本实施方式中，基站在完成编码AI网络模型的训练之后，将训练得到的编码AI网络模型下发给终端，这样，终端在后续的CSI上报过程中，可以利用该编码AI网络模型对估计到的信道信息进行压缩编码，并上报压缩编码后的信道特征信息，此后，基站可以采用与该编码AI网络模型相匹配的解码AI网络模型将该信道特征信息恢复成原始的信道信 息，或者，基站也可以采用非AI的方式将该信道特征信息恢复成原始的信道信息，例如，在用于某种算法将该信道特征信息恢复成原始的信道信息。
作为一种可选的实施方式，在所述第一设备接收来自终端的AI训练数据之前，所述方法还包括：
所述第一设备向所述终端发送第一指示信息，其中，所述第一指示信息用于指示所述终端采集所述AI训练数据。
其中，上述第一指示信息与如图2所示方法实施例中的第一指示信息具有相同的含义和作用，在此不做赘述。
作为一种可选的实施方式，在所述第一设备接收来自终端的AI训练数据之前，所述方法还包括：
所述第一设备向所述终端发送第一配置信息，其中，所述第一配置信息用于配置目标时域和/或频域资源，其中，所述AI训练数据在所述目标时域和/或频域资源上采集得到。
其中，上述第一配置信息与如图2所示方法实施例中的第一配置信息具有相同的含义和作用，在此不做赘述。
作为一种可选的实施方式，在所述第一设备向所述终端发送第一指示信息或第一配置信息之前，所述方法还包括：
所述第一设备接收来自所述终端的目标能力信息，其中，所述目标能力信息指示所述终端支持的信息上报能力和/或数据采集能力；
所述第一设备向所述终端发送第一指示信息或第一配置信息，包括：
所述第一设备根据所述目标能力信息确定所述终端支持采集和上报所述AI训练数据的情况下，向所述终端发送第一指示信息或第一配置信息。
可选地，所述目标能力信息包括以下至少一项：
第二指示信息，所述第二指示信息用于指示所述终端是否支持采集所述AI训练数据；
所述终端支持的最大信道状态信息参考信号CSI-RS端口数；
所述终端的全部端口数。
其中，上述目标能力信息与如图2所示方法实施例中的目标能力信息具有相同的含义和作用，在此不做赘述。
本实施方式中，第一设备预先获取终端的目标能力信息，并据此选择支持采集和上报所述AI训练数据的终端来执行采集和上报所述AI训练数据的动作。
在第一种可选的实施方中，所述方法还包括：
所述第一设备接收来自所述终端的第一信息，其中，所述第一信息指示所述终端采集并上报所述AI训练数据；
和/或，
所述第一设备接收来自所述终端的第二信息，所述第二信息指示所述终端不会采集或上报所述AI训练数据。
值得提出的是，基站在接收到来自终端的第一信息的情况下，可以接收来自该终端的AI训练数据。
在一种可能的实现方式中，基站在接收到来自终端的第二信息的情况下，可以选择其他终端来采集和上报AI训练数据，直至接收到来自终端的第一信息，例如：基站向其他终端发送第一指示信息或第一配置信息，并接收来自该其他终端的第一信息，当然，如果该其他终端仍然向基站发送第二信息，则基站继续寻找能够支持采集并上报所述AI训练数据的终端。
或者，基站在接收到来自终端的第二信息的情况下，可以采取其他方式来获取AI训练数据，例如：采用预先存储的数据来训练编码和/或解码AI网络模型，在此不做赘述。
作为一种可选的实施方式，在所述第一设备接收来自终端的AI训练数据之前，所述方法还包括：
所述第一设备为所述终端配置第一参考信号资源对应的第一端口和/或CSI-RS资源对应的CSI-RS端口，其中，所述第一端口为专用于采集所述AI训练数据的第一参考信号资源对应的端口，所述CSI-RS端口为用于CSI测量和采集所述AI训练数据的端口。
其中，上述第一参考信号资源、第一端口、CSI-RS端口与如图2所示方法实施例中的第一参考信号资源、第一端口、CSI-RS端口的含义和作用相同，在此不再赘述。
作为一种可选的实施方式，所述AI训练数据携带于以下至少一项：
第一上行控制信息UCI，所述第一UCI携带所述目标下行信道的CSI报告，所述CSI报告与所述第一信道信息相关；
第二UCI，所述第二UCI不携带所述目标下行信道的CSI报告；
目标物理上行共享信道PUSCH。
作为一种可选的实施方式，所述AI训练数据包括所述CSI报告内的第一信道信息，其中，所述CSI报告内的第一信道信息为预编码信息。
在实施中，基站可以接收来自终端的CSI报告和AI训练数据，此时，基站可以基于AI训练数据中精确度更高的第一信道信息来调度目标下行信道，这样，可以提升基站调度目标下行信道的可靠性。
作为一种可选的实施方式，在所述终端具有第一AI网络模型的情况下，所述方法还包括：
所述第一设备接收来自所述终端的所述第二信道特征信息，其中，所述第二信道特征信息基于所述第一AI网络模型对所述第一信道信息进行第一处理得到；
所述第一设备根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型，包括：
所述第一设备根据所述AI训练数据和所述第二信道特征信息训练所述第二AI网络模型。
其中，所述第一设备根据所述AI训练数据和所述第二信道特征信息训练所述第二AI 网络模型，可以是将二信道特征信息输入第二AI网络模型，并以第二AI网络模型输出该第二信道特征信息对应的第一信道信息为目的训练该第二AI网络模型。这样，能够提升训练得到的第二AI网络模型与终端已经具有的第一AI网络模型的匹配程度。其中，终端已经具有的第一AI网络模型可以是终端基于估计到的实际的信道信息训练得到，也可以是基站之前训练并下发的AI网络模型，在此不作具体限定。
作为一种可选的实施方式，所述AI训练数据中携带所述第二信道特征信息的第一标识信息；
或者，
所述第一设备接收来自终端的AI训练数据，包括：
所述第一设备接收来自所述终端的目标CSI报告，所述目标CSI报告携带所述第二信道特征信息和所述AI训练数据；或者，
所述第一设备在与所述第二信道特征信息的第一标识信息对应的时域和/或频域资源上，接收来自所述终端的所述AI训练数据。
作为一种可选的实施方式，所述AI训练数据包括基于第一码本确定的预编码矩阵，所述第一码本的第一参数大于预设参数，所述第一参数包括以下至少一项：
端口数、时延径个数、波束数量、非零系数比例。
作为一种可选的实施方式，所述第一信道信息为预编码矩阵和信道矩阵中的至少一项。
作为一种可选的实施方式，所述AI训练数据包括以下至少一项：
N层预编码矩阵中最强的M层预编码矩阵，所述目标下行信道的信道信息包括所述N层预编码矩阵，N为正整数，且M为小于或等于N的正整数；
所述N层预编码矩阵；
所述N层预编码矩阵中，第三指示信息指示的预编码矩阵，所述第三指示信息来自所述第一设备；
所述N层预编码矩阵中，第四指示信息指示的预编码矩阵，所述第四指示信息为所述终端预先发送至所述第一设备的指示信息；
所述N层预编码矩阵中的，满足预设条件的预编码矩阵。
作为一种可选的实施方式，所述预设条件包括以下至少一项：
信道质量指示CQI大于或等于第一阈值；
信号与干扰加噪声比SINR大于或等于第二阈值；
特征值大于或等于第三阈值；
奇异值大于或等于第四阈值。
本申请实施例提供的第一设备执行的AI网络模型训练方法，与终端执行的信息传输方法相对应，均能够提升训练得到的第一AI网络模型和/或第二AI网络模型与目标下行信道的匹配程度。
为了便于说明本申请实施例提供的信息传输方法，以第一设备是基站为例说明本申请 实施例提供的信息传输方法和AI网络模型训练方法可以包括以下步骤：
步骤1、基站通过DCI指示终端，激活AI训练数据的测量及上报，或者，基站通过RRC/MACCE信令配置测量AI训练数据的时刻或周期，终端在对应的时刻激活AI训练数据的测量和上报。
可选地，终端在接收到DCI指示或RRC/MACCE信令之后，可以向基站反馈是否支持测量和上报AI训练数据的状态，例如：若终端电量不足，则上报无法执行测量和上报AI训练数据；若用户不同意上报，则上报无法执行测量和上报AI训练数据；若终端处于高速运动状态，此时，信道测量可能不准确，则上报无法执行测量和上报AI训练数据。
步骤2、终端测量得到信道信息并计算对应的预编码矩阵。
情景一
终端使用R-16TypeII码本上报CSI，其中，CSI可以包括预编码矩阵指示(Precoding Matrix Iindicator，PMI)，信道质量指示(Channel Quality Indicator，CQI)，秩指示(Rank Indicator，RI)等，此时基站可能还没有解码AI网络模型，或者基站的解码AI网络模型与终端的编码AI网络模型不匹配。
第一方面，对于终端没有编码AI网络模型的情况，终端可以将预编码矩阵信息作为AI训练数据并入CSI中一起上报，基站得到CSI之后，可以解出PMI用于调度，并解出AI训练数据用于AI模型训练，训练好之后将编码AI网络模型发送给终端。或者，基站得到CSI之后，可以解出AI训练数据用于AI模型训练，训练好之后将编码AI网络模型发送给终端，且基站还可以使用AI训练数据取代PMI作为调度的预编码信息。
值得提出的是，终端还可以将AI训练数据通过PUSCH发送给基站，在AI网络模型训练好后，基站再将编码AI网络模型发送给终端。或者在第一设备是除了基站之外的其他节点(如核心网)的情况下，终端还可以将AI训练数据通过PUSCH发送给该节点，在AI网络模型训练好后该节点可以在将编码AI网络模型发送给终端，以及将解码AI网络模型发送给基站。
第二方面，对于终端有编码AI网络模型的情况，此时终端已经训练了自己的编码AI网络模型，但是基站没有解码AI网络模型或者基站的解码AI网络模型与终端的编码AI网络模型不匹配。此时，终端在发送AI训练数据的同时可以发送对应的编码AI网络模型编码好的信道特征信息，具体发送方式与上述AI训练数据的发送方式类似，在此不再赘述。
情景二
终端使用编码AI网络模型上报PMI，此时终端的编码AI网络模型已经和基站的解码AI网络模型匹配，可以工作。但是，此时由于有资源剩余，终端还可以上报AI训练数据，或者是由于时间过长，信道发生变化，需要重新匹配编码AI网络模型和解码AI网络模型，此时，终端采集并上报AI训练数据。
其中，由于终端已经使用编码AI网络模型进行PMI上报，终端将对应的原始信道信 息作为AI训练数据进行上报，即AI训练数据和CSI报告中的PMI相互对应，上报的PMI即为AI训练数据的编码结果。
在实施中，终端上报AI训练数据的方式可以是在CSI报告中上报或者通过其他UCI上报，例如：基站指示终端上报携带AI训练数据的UCI的时频位置，这个时频位置与CSI report configuration ID对应，即基站接收到CSI获得编码后的PMI时，可以在对应的时频位置上接收该UCI对应的AI训练数据，从而确定AI训练数据与该PMI之间的对应关系；或者，终端通过PUSCH上报AI训练数据，同时该AI训练数据中还携带CSI report configuration ID，基站通过接收PUSCH获得一个或多个AI训练数据，并根据该AI训练数据携带的CSI report configuration ID，找到对应的CSI，然后确定该CSI中的PMI即为与AI训练数据对应的编码信息。
本申请实施例提供的信息传输方法，执行主体可以为信息传输装置。本申请实施例中以信息传输装置执行信息传输方法为例，说明本申请实施例提供的信息传输装置。
请参阅图4，本申请实施例提供的一种信息传输装置，可以是终端内的装置，如图4所示，该信息传输装置400可以包括以下模块：
采集模块401，用于采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；
第一发送模块402，用于向第一设备发送所述AI训练数据。
可选地，信息传输装置400还包括：
第二接收模块，用于接收来自所述第一设备的第一指示信息，其中，所述第一指示信息用于指示所述终端采集所述AI训练数据。
可选地，信息传输装置400还包括：
第三接收模块，用于接收来自所述第一设备的第一配置信息，其中，所述第一配置信息用于配置目标时域和/或频域资源；
采集模块401具体用于：
在所述目标时域和/或频域资源采集AI训练数据。
可选地，信息传输装置400还包括：
第二发送模块，用于向所述第一设备发送目标能力信息，其中，所述目标能力信息指示所述终端支持的信息上报能力和/或数据采集能力。
可选地，所述目标能力信息包括以下至少一项：
第二指示信息，所述第二指示信息用于指示所述终端是否支持采集所述AI训练数据；
所述终端支持的最大信道状态信息参考信号CSI-RS端口数；
所述终端的全部端口数。
可选地，信息传输装置400还包括：
第三发送模块，用于在根据所述终端的状态信息，确定支持采集所述AI训练数据的情况下，向所述第一设备发送第一信息，所述第一信息指示所述终端采集并上报所述AI训练数据；
和/或，
第四发送模块，用于所述终端在根据所述终端的状态信息，确定不支持采集所述AI训练数据的情况下，向所述第一设备发送第二信息，所述第二信息指示所述终端不会采集或上报所述AI训练数据。
可选地，采集模块401具体用于：
基于第一参考信号资源对应的第一端口或CSI-RS资源对应的CSI-RS端口采集AI训练数据，其中，所述第一端口包括预先配置的用于采集所述AI训练数据的第一参考信号资源对应的端口，所述CSI-RS端口包括用于CSI测量的端口。
可选地，所述AI训练数据携带于以下至少一项：
第一上行控制信息UCI，所述第一UCI携带所述目标下行信道的CSI报告，所述CSI报告与所述第一信道信息相关；
第二UCI，所述第二UCI不携带所述目标下行信道的CSI报告；
目标物理上行共享信道PUSCH。
可选地，所述AI训练数据包括所述CSI报告内的第一信道信息，其中，所述CSI报告内的第一信道信息为预编码信息。
可选地，在所述终端具有第一AI网络模型的情况下，信息传输装置400还包括：
第一处理模块，用于基于所述第一AI网络模型对所述第一信道信息进行第一处理，得到第二信道特征信息；
第五发送模块，用于向所述第一设备发送所述第二信道特征信息，其中，所述AI训练数据和所述第二信道特征信息用于训练所述第二AI网络模型。
可选地，所述AI训练数据中携带所述第二信道特征信息的第一标识信息；
或者，
第一发送模块402具体用于：
向第一设备发送目标CSI报告，所述目标CSI报告携带所述第二信道特征信息和所述AI训练数据；或者，
在与所述第二信道特征信息的第一标识信息对应的时域和/或频域资源上，向第一设备发送所述AI训练数据。
可选地，所述AI训练数据包括基于第一码本确定的预编码矩阵，所述第一码本的第一参数大于预设参数，所述第一参数包括以下至少一项：
端口数、时延径个数、波束数量、非零系数比例。
可选地，所述第一信道信息为预编码矩阵和信道矩阵中的至少一项。
可选地，所述AI训练数据包括以下至少一项：
N层预编码矩阵中最强的M层预编码矩阵，所述目标下行信道的信道信息包括所述N层预编码矩阵，N为正整数，且M为小于或等于N的正整数；
所述N层预编码矩阵；
所述N层预编码矩阵中，第三指示信息指示的预编码矩阵，所述第三指示信息来自所述第一设备；
所述N层预编码矩阵中，第四指示信息指示的预编码矩阵，所述第四指示信息为所述终端预先发送至所述第一设备的指示信息；
所述N层预编码矩阵中的，满足预设条件的预编码矩阵。
可选地，所述预设条件包括以下至少一项：
信道质量指示CQI大于或等于第一阈值；
信号与干扰加噪声比SINR大于或等于第二阈值；
特征值大于或等于第三阈值；
奇异值大于或等于第四阈值。
本申请实施例提供的信息传输装置400，能够实现如图2所示方法实施例中终端实现的各个过程，且能够取得相同的有益效果，为避免重复，在此不再赘述。
本申请实施例中的信息传输装置可以是电子设备，例如具有操作***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网络模型训练方法，执行主体可以为AI网络模型训练装置。本申请实施例中以AI网络模型训练装置执行AI网络模型训练方法为例，说明本申请实施例提供的AI网络模型训练装置。
请参阅图5，本申请实施例提供的一种AI网络模型训练装置，可以是第一设备内的装置，如图5所示，该AI网络模型训练装置500可以包括以下模块：
第一接收模块501，用于接收来自终端的AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息；
训练模块502，用于根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型，其中，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息。
可选地，在所述第一设备根据所述AI训练数据训练第一AI网络模型的情况下，AI网络模型训练装置500还包括：
第六发送模块，用于向所述终端发送所述第一AI网络模型的相关信息。
可选地，AI网络模型训练装置500还包括：
第七发送模块，用于向所述终端发送第一指示信息，其中，所述第一指示信息用于指示所述终端采集所述AI训练数据。
可选地，AI网络模型训练装置500还包括：
第八发送模块，用于向所述终端发送第一配置信息，其中，所述第一配置信息用于配置目标时域和/或频域资源，其中，所述AI训练数据在所述目标时域和/或频域资源上采集得到。
可选地，AI网络模型训练装置500还包括：
第四接收模块，用于接收来自所述终端的目标能力信息，其中，所述目标能力信息指示所述终端支持的信息上报能力和/或数据采集能力；
第七发送模块或第八发送模块具体用于：
在根据所述目标能力信息确定所述终端支持采集和上报所述AI训练数据的情况下，向所述终端发送第一指示信息或第一配置信息。
可选地，所述目标能力信息包括以下至少一项：
第二指示信息，所述第二指示信息用于指示所述终端是否支持采集所述AI训练数据；
所述终端支持的最大信道状态信息参考信号CSI-RS端口数；
所述终端的全部端口数。
可选地，AI网络模型训练装置500还包括：
第五接收模块，用于接收来自所述终端的第一信息，其中，所述第一信息指示所述终端采集并上报所述AI训练数据；
和/或，
第六接收模块，用于接收来自所述终端的第二信息，所述第二信息指示所述终端不会采集或上报所述AI训练数据。
可选地，AI网络模型训练装置500还包括：
配置模块，用于为所述终端配置第一参考信号资源对应的第一端口和/或CSI-RS资源对应的CSI-RS端口，其中，所述第一端口为专用于采集所述AI训练数据的第一参考信号资源对应的端口，所述CSI-RS端口为用于CSI测量和采集所述AI训练数据的端口。
可选地，所述AI训练数据携带于以下至少一项：
第一上行控制信息UCI，所述第一UCI携带所述目标下行信道的CSI报告，所述CSI报告与所述第一信道信息相关；
第二UCI，所述第二UCI不携带所述目标下行信道的CSI报告；
目标物理上行共享信道PUSCH。
可选地，所述AI训练数据包括所述CSI报告内的第一信道信息，其中，所述CSI报告内的第一信道信息为预编码信息。
可选地，在所述终端具有第一AI网络模型的情况下，AI网络模型训练装置500还包 括：
第六接收模块，用于接收来自所述终端的所述第二信道特征信息，其中，所述第二信道特征信息基于所述第一AI网络模型对所述第一信道信息进行第一处理得到；
训练模块502具体用于：
根据所述AI训练数据和所述第二信道特征信息训练所述第二AI网络模型。
可选地，所述AI训练数据中携带所述第二信道特征信息的第一标识信息，或者，第一接收模块501具体用：
接收来自所述终端的目标CSI报告，所述目标CSI报告携带所述第二信道特征信息和所述AI训练数据；
或者，
在与所述第二信道特征信息的第一标识信息对应的时域和/或频域资源上，接收来自所述终端的所述AI训练数据。
可选地，所述AI训练数据包括基于第一码本确定的预编码矩阵，所述第一码本的第一参数大于预设参数，所述第一参数包括以下至少一项：
端口数、时延径个数、波束数量、非零系数比例。
可选地，所述第一信道信息为预编码矩阵和信道矩阵中的至少一项。
可选地，所述AI训练数据包括以下至少一项：
N层预编码矩阵中最强的M层预编码矩阵，所述目标下行信道的信道信息包括所述N层预编码矩阵，N为正整数，且M为小于或等于N的正整数；
所述N层预编码矩阵；
所述N层预编码矩阵中，第三指示信息指示的预编码矩阵，所述第三指示信息来自所述第一设备；
所述N层预编码矩阵中，第四指示信息指示的预编码矩阵，所述第四指示信息为所述终端预先发送至所述第一设备的指示信息；
所述N层预编码矩阵中的，满足预设条件的预编码矩阵。
可选地，所述预设条件包括以下至少一项：
信道质量指示CQI大于或等于第一阈值；
信号与干扰加噪声比SINR大于或等于第二阈值；
特征值大于或等于第三阈值；
奇异值大于或等于第四阈值。
本申请实施例中的AI网络模型训练装置可以是电子设备，例如具有操作***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。该电子设备可以是网络侧设备。示例性的，终端可以包括但不限于上述所列举的网络侧设备12的类型，本申请实施例不作具体限定。
本申请实施例提供的AI网络模型训练装置500，能够实现如图3所示方法实施例中 第一设备实现的各个过程，且能够取得相同的有益效果，为避免重复，在此不再赘述。
可选的，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执行时实现如图2所示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为第一设备时，该程序或指令被处理器601执行时实现如图3所示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，通信接口用于采集人工智能AI训练数据，以及向第一设备发送所述AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息。
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***与处理器710逻辑相连，从而通过电源管理***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***、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***、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，射频单元701，用于采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；
射频单元701，还用于向第一设备发送所述AI训练数据。
可选地，射频单元701在执行所述采集AI训练数据之前，还用于：
接收来自所述第一设备的第一指示信息，其中，所述第一指示信息用于指示所述终端采集所述AI训练数据。
可选地，射频单元701在执行所述采集AI训练数据之前，还用于：
接收来自所述第一设备的第一配置信息，其中，所述第一配置信息用于配置目标时域和/或频域资源；
射频单元701执行的所述采集AI训练数据，包括：
射频单元701在所述目标时域和/或频域资源采集AI训练数据。
可选地，射频单元701在执行所述接收来自所述第一设备的第一指示信息或第一配置信息之前，还用于：
向所述第一设备发送目标能力信息，其中，所述目标能力信息指示所述终端支持的信息上报能力和/或数据采集能力。
可选地，所述目标能力信息包括以下至少一项：
第二指示信息，所述第二指示信息用于指示所述终端是否支持采集所述AI训练数据；
所述终端支持的最大信道状态信息参考信号CSI-RS端口数；
所述终端的全部端口数。
可选地，处理器710，还用于在根据所述终端的状态信息，确定支持采集所述AI训练数据的情况下，控制射频单元701向所述第一设备发送第一信息，所述第一信息指示所述终端采集并上报所述AI训练数据；
和/或，
处理器710，还用于在根据所述终端的状态信息，确定不支持采集所述AI训练数据的情况下，控制射频单元701向所述第一设备发送第二信息，所述第二信息指示所述终端不会采集或上报所述AI训练数据。
可选地，射频单元701执行的所述采集AI训练数据，包括：
射频单元701基于第一参考信号资源对应的第一端口或CSI-RS资源对应的CSI-RS端口采集AI训练数据，其中，所述第一端口包括预先配置的用于采集所述AI训练数据的第一参考信号资源对应的端口，所述CSI-RS端口包括用于CSI测量的端口。
可选地，所述AI训练数据携带于以下至少一项：
第一上行控制信息UCI，所述第一UCI携带所述目标下行信道的CSI报告，所述CSI报告与所述第一信道信息相关；
第二UCI，所述第二UCI不携带所述目标下行信道的CSI报告；
目标物理上行共享信道PUSCH。
可选地，所述AI训练数据包括所述CSI报告内的第一信道信息，其中，所述CSI报告内的第一信道信息为预编码信息。
可选地，在所述终端具有第一AI网络模型的情况下：
处理器710，还用于基于所述第一AI网络模型对所述第一信道信息进行第一处理，得到第二信道特征信息；
射频单元701，还用于向所述第一设备发送所述第二信道特征信息，其中，所述AI训练数据和所述第二信道特征信息用于训练所述第二AI网络模型。
可选地，所述AI训练数据中携带所述第二信道特征信息的第一标识信息；
或者，
射频单元701执行的所述向第一设备发送所述AI训练数据，包括：
向第一设备发送目标CSI报告，所述目标CSI报告携带所述第二信道特征信息和所述AI训练数据；或者，
在与所述第二信道特征信息的第一标识信息对应的时域和/或频域资源上，向第一设备发送所述AI训练数据。
可选地，所述AI训练数据包括基于第一码本确定的预编码矩阵，所述第一码本的第一参数大于预设参数，所述第一参数包括以下至少一项：
端口数、时延径个数、波束数量、非零系数比例。
可选地，所述第一信道信息为预编码矩阵和信道矩阵中的至少一项。
可选地，所述AI训练数据包括以下至少一项：
N层预编码矩阵中最强的M层预编码矩阵，所述目标下行信道的信道信息包括所述N层预编码矩阵，N为正整数，且M为小于或等于N的正整数；
所述N层预编码矩阵；
所述N层预编码矩阵中，第三指示信息指示的预编码矩阵，所述第三指示信息来自所述第一设备；
所述N层预编码矩阵中，第四指示信息指示的预编码矩阵，所述第四指示信息为所述终端预先发送至所述第一设备的指示信息；
所述N层预编码矩阵中的，满足预设条件的预编码矩阵。
可选地，所述预设条件包括以下至少一项：
信道质量指示CQI大于或等于第一阈值；
信号与干扰加噪声比SINR大于或等于第二阈值；
特征值大于或等于第三阈值；
奇异值大于或等于第四阈值。
本申请实施例提供的终端700能够实现如图3所示信息传输装置执行的各个过程，且能够取得相同的有益效果，为避免重复，在此不再赘述。
本申请实施例还提供一种网络侧设备，在该网络侧设备可以是接入网设备或核心网设备，该网络侧设备包括通信接口和处理器，其中，通信接口用于接收来自终端的AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息；所述处理器用于根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型，其中，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息。
该网络侧设备实施例与图3所示方法实施例对应，图3所示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。
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如图2或图3所示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，实现如图2或图3所示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***级芯片，***芯片，芯片***或片上***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如图2或图3所示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***，包括：终端和网络侧设备，所述终端可用于执行如图2所示的信息传输方法的步骤，所述网络侧设备可用于执行如图3所示的AI网络模型训练方法的步骤。
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现有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 (35)

  1. 一种信息传输方法，包括：
    终端采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；
    所述终端向第一设备发送所述AI训练数据。
  2. 根据权利要求1所述的方法，其中，在所述终端采集AI训练数据之前，所述方法还包括：
    所述终端接收来自所述第一设备的第一指示信息，其中，所述第一指示信息用于指示所述终端采集所述AI训练数据。
  3. 根据权利要求1所述的方法，其中，在所述终端采集AI训练数据之前，所述方法还包括：
    所述终端接收来自所述第一设备的第一配置信息，其中，所述第一配置信息用于配置目标时域和/或频域资源；
    所述终端采集AI训练数据，包括：
    所述终端在所述目标时域和/或频域资源采集AI训练数据。
  4. 根据权利要求2或3所述的方法，其中，在所述终端接收来自所述第一设备的第一指示信息或第一配置信息之前，所述方法还包括：
    所述终端向所述第一设备发送目标能力信息，其中，所述目标能力信息指示所述终端支持的信息上报能力和/或数据采集能力。
  5. 根据权利要求4所述的方法，其中，所述目标能力信息包括以下至少一项：
    第二指示信息，所述第二指示信息用于指示所述终端是否支持采集所述AI训练数据；
    所述终端支持的最大信道状态信息参考信号CSI-RS端口数；
    所述终端的全部端口数。
  6. 根据权利要求2或3所述的方法，其中，所述方法还包括：
    所述终端在根据所述终端的状态信息，确定支持采集所述AI训练数据的情况下，向所述第一设备发送第一信息，所述第一信息指示所述终端采集并上报所述AI训练数据；
    和/或，
    所述终端在根据所述终端的状态信息，确定不支持采集所述AI训练数据的情况下，向所述第一设备发送第二信息，所述第二信息指示所述终端不会采集或上报所述AI训练数据。
  7. 根据权利要求1所述的方法，其中，所述终端采集AI训练数据，包括：
    所述终端基于第一参考信号资源对应的第一端口或CSI-RS资源对应的CSI-RS端口采集AI训练数据，其中，所述第一端口包括预先配置的用于采集所述AI训练数据的第一参考信号资源对应的端口，所述CSI-RS端口包括用于CSI测量的端口。
  8. 根据权利要求1至3中任一项所述的方法，其中，所述AI训练数据携带于以下至少一项：
    第一上行控制信息UCI，所述第一UCI携带所述目标下行信道的CSI报告，所述CSI报告与所述第一信道信息相关；
    第二UCI，所述第二UCI不携带所述目标下行信道的CSI报告；
    目标物理上行共享信道PUSCH。
  9. 根据权利要求8所述的方法，其中，所述AI训练数据包括所述CSI报告内的第一信道信息，其中，所述CSI报告内的第一信道信息为预编码信息。
  10. 根据权利要求1至3中任一项所述的方法，其中，在所述终端具有第一AI网络模型的情况下，所述方法还包括：
    所述终端基于所述第一AI网络模型对所述第一信道信息进行第一处理，得到第二信道特征信息；
    所述终端向所述第一设备发送所述第二信道特征信息，其中，所述AI训练数据和所述第二信道特征信息用于训练所述第二AI网络模型。
  11. 根据权利要求10所述的方法，其中，所述AI训练数据中携带所述第二信道特征信息的第一标识信息；
    或者，
    所述终端向第一设备发送所述AI训练数据，包括：
    所述终端向第一设备发送目标CSI报告，所述目标CSI报告携带所述第二信道特征信息和所述AI训练数据；或者，
    所述终端在与所述第二信道特征信息的第一标识信息对应的时域和/或频域资源上，向第一设备发送所述AI训练数据。
  12. 根据权利要求1至3中任一项所述的方法，其中，所述AI训练数据包括基于第一码本确定的预编码矩阵，所述第一码本的第一参数大于预设参数，所述第一参数包括以下至少一项：
    端口数、时延径个数、波束数量、非零系数比例。
  13. 根据权利要求1至3中任一项所述的方法，其中，所述第一信道信息为预编码矩阵和信道矩阵中的至少一项。
  14. 根据权利要求13所述的方法，其中，所述AI训练数据包括以下至少一项：
    N层预编码矩阵中最强的M层预编码矩阵，所述目标下行信道的信道信息包括所述N层预编码矩阵，N为正整数，且M为小于或等于N的正整数；
    所述N层预编码矩阵；
    所述N层预编码矩阵中，第三指示信息指示的预编码矩阵，所述第三指示信息来自所述第一设备；
    所述N层预编码矩阵中，第四指示信息指示的预编码矩阵，所述第四指示信息为所述终端预先发送至所述第一设备的指示信息；
    所述N层预编码矩阵中的，满足预设条件的预编码矩阵。
  15. 根据权利要求14所述的方法，其中，所述预设条件包括以下至少一项：
    信道质量指示CQI大于或等于第一阈值；
    信号与干扰加噪声比SINR大于或等于第二阈值；
    特征值大于或等于第三阈值；
    奇异值大于或等于第四阈值。
  16. 一种人工智能AI网络模型训练方法，包括：
    第一设备接收来自终端的AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息；
    所述第一设备根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型，其中，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息。
  17. 根据权利要求16所述的方法，其中，在所述第一设备根据所述AI训练数据训练第一AI网络模型的情况下，所述方法还包括：
    所述第一设备向所述终端发送所述第一AI网络模型的相关信息。
  18. 根据权利要求16所述的方法，其中，在所述第一设备接收来自终端的AI训练数据之前，所述方法还包括：
    所述第一设备向所述终端发送第一指示信息，其中，所述第一指示信息用于指示所述终端采集所述AI训练数据。
  19. 根据权利要求16所述的方法，其中，在所述第一设备接收来自终端的AI训练数据之前，所述方法还包括：
    所述第一设备向所述终端发送第一配置信息，其中，所述第一配置信息用于配置目标时域和/或频域资源，其中，所述AI训练数据在所述目标时域和/或频域资源上采集得到。
  20. 根据权利要求18或19所述的方法，其中，在所述第一设备向所述终端发送第一指示信息或第一配置信息之前，所述方法还包括：
    所述第一设备接收来自所述终端的目标能力信息，其中，所述目标能力信息指示所述终端支持的信息上报能力和/或数据采集能力；
    所述第一设备向所述终端发送第一指示信息或第一配置信息，包括：
    所述第一设备根据所述目标能力信息确定所述终端支持采集和上报所述AI训练数据的情况下，向所述终端发送第一指示信息或第一配置信息。
  21. 根据权利要求20所述的方法，其中，所述目标能力信息包括以下至少一项：
    第二指示信息，所述第二指示信息用于指示所述终端是否支持采集所述AI训练数据；
    所述终端支持的最大信道状态信息参考信号CSI-RS端口数；
    所述终端的全部端口数。
  22. 根据权利要求18或19所述的方法，其中，所述方法还包括：
    所述第一设备接收来自所述终端的第一信息，其中，所述第一信息指示所述终端采集并上报所述AI训练数据；
    和/或，
    所述第一设备接收来自所述终端的第二信息，所述第二信息指示所述终端不会采集或上报所述AI训练数据。
  23. 根据权利要求16所述的方法，其中，在所述第一设备接收来自终端的AI训练数据之前，所述方法还包括：
    所述第一设备为所述终端配置第一参考信号资源对应的第一端口和/或CSI-RS资源对应的CSI-RS端口，其中，所述第一端口为专用于采集所述AI训练数据的第一参考信号资源对应的端口，所述CSI-RS端口为用于CSI测量和采集所述AI训练数据的端口。
  24. 根据权利要求16至19中任一项所述的方法，其中，所述AI训练数据携带于以下至少一项：
    第一上行控制信息UCI，所述第一UCI携带所述目标下行信道的CSI报告，所述CSI报告与所述第一信道信息相关；
    第二UCI，所述第二UCI不携带所述目标下行信道的CSI报告；
    目标物理上行共享信道PUSCH。
  25. 根据权利要求24所述的方法，其中，所述AI训练数据包括所述CSI报告内的第一信道信息，其中，所述CSI报告内的第一信道信息为预编码信息。
  26. 根据权利要求16至19中任一项所述的方法，其中，在所述终端具有第一AI网络模型的情况下，所述方法还包括：
    所述第一设备接收来自所述终端的所述第二信道特征信息，其中，所述第二信道特征信息基于所述第一AI网络模型对所述第一信道信息进行第一处理得到；
    所述第一设备根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型，包括：
    所述第一设备根据所述AI训练数据和所述第二信道特征信息训练所述第二AI网络模型。
  27. 根据权利要求26所述的方法，其中，所述AI训练数据中携带所述第二信道特征信息的第一标识信息；
    或者，
    所述第一设备接收来自终端的AI训练数据，包括：
    所述第一设备接收来自所述终端的目标CSI报告，所述目标CSI报告携带所述第二信道特征信息和所述AI训练数据；或者，
    所述第一设备在与所述第二信道特征信息的第一标识信息对应的时域和/或频域资源上，接收来自所述终端的所述AI训练数据。
  28. 根据权利要求16至19中任一项所述的方法，其中，所述AI训练数据包括基于第一码本确定的预编码矩阵，所述第一码本的第一参数大于预设参数，所述第一参数包括以下至少一项：
    端口数、时延径个数、波束数量、非零系数比例。
  29. 根据权利要求16至19中任一项所述的方法，其中，所述第一信道信息为预编码矩阵和信道矩阵中的至少一项。
  30. 根据权利要求29所述的方法，其中，所述AI训练数据包括以下至少一项：
    N层预编码矩阵中最强的M层预编码矩阵，所述目标下行信道的信道信息包括所述N层预编码矩阵，N为正整数，且M为小于或等于N的正整数；
    所述N层预编码矩阵；
    所述N层预编码矩阵中，第三指示信息指示的预编码矩阵，所述第三指示信息来自所述第一设备；
    所述N层预编码矩阵中，第四指示信息指示的预编码矩阵，所述第四指示信息为所述终端预先发送至所述第一设备的指示信息；
    所述N层预编码矩阵中的，满足预设条件的预编码矩阵。
  31. 根据权利要求30所述的方法，其中，所述预设条件包括以下至少一项：
    信道质量指示CQI大于或等于第一阈值；
    信号与干扰加噪声比SINR大于或等于第二阈值；
    特征值大于或等于第三阈值；
    奇异值大于或等于第四阈值。
  32. 一种信息传输装置，应用于终端，所述装置包括：
    采集模块，用于采集人工智能AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息，所述AI训练数据用于训练第一AI网络模型和/或第二AI网络模型，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息；
    第一发送模块，用于向第一设备发送所述AI训练数据。
  33. 一种人工智能AI网络模型训练装置，应用于第一设备，所述装置包括：
    第一接收模块，用于接收来自终端的AI训练数据，其中，所述AI训练数据包括目标下行信道的第一信道信息；
    训练模块，用于根据所述AI训练数据训练第一AI网络模型和/或第二AI网络模型， 其中，所述第一AI网络模型用于将第二信道信息处理成第一信道特征信息，所述第二AI网络模型用于将所述第一信道特征信息恢复成所述第二信道信息，所述第二信道信息为所述目标下行信道的信道信息。
  34. 一种通信设备，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1至15中任一项所述的信息传输方法的步骤，或者实现如权利要求16至31中任一项所述的人工智能AI网络模型训练方法的步骤。
  35. 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至15中任一项所述的信息传输方法的步骤，或者实现如权利要求16至31中任一项所述的人工智能AI网络模型训练方法的步骤。
PCT/CN2023/116030 2022-09-07 2023-08-31 信息传输方法、ai网络模型训练方法、装置和通信设备 WO2024051564A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211091873.9 2022-09-07
CN202211091873.9A CN117715106A (zh) 2022-09-07 2022-09-07 信息传输方法、ai网络模型训练方法、装置和通信设备

Publications (1)

Publication Number Publication Date
WO2024051564A1 true WO2024051564A1 (zh) 2024-03-14

Family

ID=90155746

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/116030 WO2024051564A1 (zh) 2022-09-07 2023-08-31 信息传输方法、ai网络模型训练方法、装置和通信设备

Country Status (2)

Country Link
CN (1) CN117715106A (zh)
WO (1) WO2024051564A1 (zh)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200413316A1 (en) * 2018-03-08 2020-12-31 Telefonaktiebolaget Lm Ericsson (Publ) Managing communication in a wireless communications network
CN113765830A (zh) * 2020-06-03 2021-12-07 华为技术有限公司 获取信道信息的方法及通信装置
WO2022141397A1 (zh) * 2020-12-31 2022-07-07 华为技术有限公司 一种神经网络的训练方法以及相关装置
CN114726413A (zh) * 2021-01-04 2022-07-08 ***通信有限公司研究院 信道信息获取方法、装置及相关设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200413316A1 (en) * 2018-03-08 2020-12-31 Telefonaktiebolaget Lm Ericsson (Publ) Managing communication in a wireless communications network
CN113765830A (zh) * 2020-06-03 2021-12-07 华为技术有限公司 获取信道信息的方法及通信装置
WO2022141397A1 (zh) * 2020-12-31 2022-07-07 华为技术有限公司 一种神经网络的训练方法以及相关装置
CN114726413A (zh) * 2021-01-04 2022-07-08 ***通信有限公司研究院 信道信息获取方法、装置及相关设备

Also Published As

Publication number Publication date
CN117715106A (zh) 2024-03-15

Similar Documents

Publication Publication Date Title
WO2021142836A1 (zh) 通信方法及装置
WO2023246618A1 (zh) 信道矩阵处理方法、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
WO2024051564A1 (zh) 信息传输方法、ai网络模型训练方法、装置和通信设备
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
WO2024149156A1 (zh) 信息传输方法、装置、终端及网络侧设备
WO2024007949A1 (zh) Ai模型处理方法、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
WO2024051594A1 (zh) 信息传输方法、ai网络模型训练方法、装置和通信设备
WO2024055993A1 (zh) Cqi传输方法、装置、终端及网络侧设备
WO2024041420A1 (zh) 测量反馈处理方法、装置、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
WO2023198062A1 (zh) Csi测量和上报方法、装置、设备、***及存储介质
WO2023185995A1 (zh) 信道特征信息传输方法、装置、终端及网络侧设备
WO2023109828A1 (zh) 数据收集方法及装置、第一设备、第二设备
WO2024041421A1 (zh) 测量反馈处理方法、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
WO2024061287A1 (zh) 人工智能ai模型传输方法、装置、终端及介质
WO2024037380A1 (zh) 信道信息处理方法、装置、通信设备及存储介质
WO2023179474A1 (zh) 信道特征信息辅助上报及恢复方法、终端和网络侧设备
WO2023197953A1 (zh) 预编码矩阵的反馈方法、终端及网络侧设备
WO2024027682A1 (zh) 特征信息传输方法、转换信息确定方法、装置和通信设备
WO2022206908A1 (zh) 上行传输方法、装置、终端、网络侧设备和存储介质
WO2023179651A1 (zh) 波束处理方法、装置及设备

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: 23862258

Country of ref document: EP

Kind code of ref document: A1