WO2025167989A1 - Communication method and communication apparatus - Google Patents

Abstract

Provided in the present application are a communication method and a communication apparatus. The method comprises: a second device obtaining first indication information from a first device, wherein the first indication information is used for the determination of a time unit in which reset state information of a first artificial intelligence (AI) model takes effect, and the second device can determine, on the basis of the first indication information, the time unit in which the reset state information of the first artificial intelligence (AI) model takes effect. The solution of the embodiments of the present application facilitates an improvement to the robustness of the channel information recovery performance.

Description

Communication method and communication device

This application claims priority to the Chinese patent application filed with the State Intellectual Property Office of China on February 8, 2024, with application number 202410177187.6, and priority to the Chinese patent application entitled “Communication Method and Communication Device”, all contents of which are incorporated by reference into this application.

Technical Field

The embodiments of the present application relate to the field of communications, and more specifically, to a communication method and a communication device.

Background Art

In communications systems, network equipment determines downlink channel configuration information, such as resources, modulation and coding scheme (MCS), and precoding, for scheduling terminal devices' downlink data channels based on downlink channel state information (CSI). Terminal devices can calculate downlink CSI by measuring downlink reference signals and generate a CSI report to feed back to the network device, or send CSI feedback information to the network device.

The introduction of artificial intelligence (AI) into wireless communications has led to the emergence of an AI-based CSI feedback method. AI models have stronger feature extraction capabilities, enabling more effective channel information compression, reducing information loss during the compression process, and ensuring the accuracy of recovered channel information. Taking CSI feedback based on time-domain correlation as an example, the AI model on the terminal device outputs CSI feedback information based on the current channel measurement results and the AI model's status information. The AI model on the network device recovers channel information based on the most recently received CSI feedback information and the AI model's status information. The AI model's status information on the terminal device is determined based on historical channel measurement results. The AI model's status information on the network device is determined based on CSI feedback information received at historical moments. Because the reception of CSI feedback information depends on the channel's transmission conditions, packet loss in CSI feedback can degrade the performance of recovering channel information.

There is an urgent need for a solution to improve the robustness of channel information recovery performance.

Summary of the Invention

Embodiments of the present application provide a communication method and a communication device to improve the robustness of channel information recovery performance.

In a first aspect, a communication method is provided. The method may be executed by a second device or a chip or circuit of the second device.

Optionally, the second device may be a device on the first AI model side. The device on the first AI model side may be replaced by a device on the terminal device side or a device on the network device side. The terminal device side may include at least one of the terminal device and the AI entity on the terminal device side. The AI entity on the terminal device side may be the terminal device itself, or an AI entity serving the terminal device, such as a server, such as an over-the-top (OTT) server or a cloud server. The network device side may include at least one of the network device and the AI entity on the network device side. The AI entity on the network device side may be the network device itself, or an AI entity serving the network device, such as a radio access network (RAN) intelligent controller (RIC), an operation administration and maintenance (OAM), or a server, such as an OTT server or a cloud server.

The method includes: obtaining first indication information from a first device, where the first indication information is used to determine a time unit in which reset status information of a first AI model takes effect.

The effective time unit may also be replaced by the effective moment.

Exemplarily, the time unit may include any of the following: a time slot, a symbol, or a transmission time-interval (TTI), etc.

The first AI model and the second AI model are autoencoding models.

The first device is a device on the second AI model side.

A first AI model, such as the AI model on the terminal device side, can be used to generate CSI feedback information. A second AI model, such as the AI model on the network device side, that matches the first AI model, such as the AI model on the terminal device side, can be used to recover channel information corresponding to the CSI feedback information.

Optionally, the device on the second AI model side may be a device on the network device side or a device on the terminal device side.

The device on the first AI model side obtaining the first indication information from the first device may include the terminal device or a chip used for the terminal device obtaining the first indication information from the first device, or the AI entity on the terminal device side or the chip used for the AI entity on the terminal device side obtaining the first indication information from the first device. Optionally, the AI entity on the terminal device side or the chip used for the AI entity on the terminal device side may obtain the first indication information from the first device through forwarding by the terminal device.

In the embodiments of the present application, resetting the state information of the first AI model helps ensure consistency between the state information of the first AI model and the state information of the second AI model, thereby ensuring the accuracy of the recovered channel information and improving the robustness of the second AI model's channel information recovery performance. For example, in the event of CSI feedback packet loss, resetting the state information of the first AI model and the state information of the second AI model restores consistency between the state information of the first AI model and the state information of the second AI model, thereby improving the accuracy of the recovered channel information.

Furthermore, in the embodiments of the present application, the time unit in which the reset state information of the first AI model takes effect can be determined based on the first indication information. This facilitates aligning the effective time of the reset state information of the first AI model with the reset state information of the second AI model, thereby further improving the performance of the second AI model in recovering channel information.

In combination with the first aspect, in certain implementations of the first aspect, first CSI feedback information is sent to the first device in a first time unit, the first CSI feedback information is related to the reset status information of the first AI model, and the first time unit is no earlier than the time unit in which the reset status information of the first AI model takes effect.

In the solution of the embodiment of the present application, the time unit in which the reset state information of the first AI model takes effect can be determined to be no later than the time when the first CSI feedback information is sent (i.e., the first time unit) based on the first indication information. Before the first CSI feedback information is sent, the reset state information of the first AI model can be applied to generate the first CSI feedback information. In this way, the second AI model side, for example, the network device side, can determine that the first CSI feedback information was generated when the reset state information took effect. Then, the channel information corresponding to the first CSI feedback information can be recovered based on the reset state information of the second AI model, which helps to ensure the accuracy of the recovered channel information, thereby improving the performance of the second AI model in recovering channel information.

Furthermore, the time interval between the first time unit and the time unit in which the reset status information of the first AI model takes effect is greater than or equal to the processing time of the first AI model.

The processing time of the first AI model may also be replaced by the processing time of the terminal device, the CSI processing time, or the CSI calculation time, etc. For example, the processing time of the first AI model may include the time required to generate CSI feedback information through processing of the first AI model.

This helps ensure that the first AI model side, for example, the terminal device side, has sufficient time to generate the first CSI feedback information based on the reset status information of the first AI model.

In combination with the first aspect, in certain implementations of the first aspect, the first indication information is used to indicate that uplink information is sent on a first time resource, and the effective time unit of the reset status information of the first AI model is no later than the start time of the first time resource.

A time resource may include one or more time units. The starting time of a time resource may be understood as the starting time of the starting time unit in the time resource.

In combination with the first aspect, in certain implementations of the first aspect, the method also includes: determining a first time period, the end time of the first time period is no later than the start time of the first time resource, the length of the first time period is predefined, or indicated by the first indication information, or indicated by other information other than the first indication information.

In combination with the first aspect, in certain implementations of the first aspect, the method also includes: obtaining the length of the first time period, the length of the first time period is used to determine the time unit in which the reset status information of the first AI model takes effect, the length of the first time period is predefined, or indicated by the first indication information, or indicated by other information other than the first indication information.

In combination with the first aspect, in certain implementations of the first aspect, the starting time of the first time period is the time when the first indication information is sent, or the time when the first indication information is received, or the time indicated by the first indication information.

In combination with the first aspect, in certain implementations of the first aspect, the method further includes: receiving scheduling information of second CSI feedback information, where the time resources of the second CSI feedback information indicated by the scheduling information overlap with the first time period, ignoring or skipping the sending of the second CSI feedback information, and the second CSI feedback information is related to the first AI model.

The second CSI feedback information may be the output of the first AI model, or may be based on the output of the first AI model. For example, the output of the first AI model is quantized to obtain the second CSI feedback information.

In the solution of the embodiment of the present application, if the time resource of the CSI feedback information (such as the second CSI feedback information) indicated by the scheduling information overlaps with the first time period, the transmission of the CSI feedback information is ignored or skipped. This prevents the second AI model from being unable to determine whether the second CSI feedback information was generated when the reset status information of the first AI model was in effect, thereby preventing the second AI model from recovering channel information based on mismatched status information, which helps ensure the accuracy of the recovered channel information.

In combination with the first aspect, in certain implementations of the first aspect, the first indication information indicates a reset of status information of the first AI model.

In combination with the first aspect, in some implementations of the first aspect, the method further includes: obtaining second indication information from the first device, the second indication information indicating a reset of status information of the first AI model.

In combination with the first aspect, in certain implementations of the first aspect, the first indication information is related to a duration required to reset the status information of the first AI model, and the duration required to reset the status information of the first AI model is predefined.

For example, the length of the first time period is indicated by the first indication information, and the length of the first time period is greater than or equal to the time required to reset the status information of the first AI model.

In the solution of the embodiment of the present application, the first indication information can be determined based on the time required to reset the status information of the first AI model, which is conducive to ensuring that the first AI model side, for example, the terminal device side, has sufficient time to complete the reset of the status information of the first AI model, so that the reset status information of the first AI model takes effect.

In combination with the first aspect, in certain implementations of the first aspect, the first indication information is related to the time required to reset the status information of the first AI model, and the method also includes: sending third indication information to the first device or other devices other than the first device, the third indication information indicating the time required to reset the status information of the first AI model.

In the solution of the embodiment of the present application, the first indication information can be determined based on the time required to reset the status information of the first AI model, which is conducive to ensuring that the first AI model side, for example, the terminal device side, has sufficient time to complete the reset of the status information of the first AI model, so that the reset status information of the first AI model takes effect.

In combination with the first aspect, in certain implementations of the first aspect, the third indication information is carried in signaling sent by the terminal device that carries the terminal device capabilities, or the third indication information is carried in signaling that carries the configuration information of the first AI model.

In combination with the first aspect, in some implementations of the first aspect, the first indication information is carried in first downlink control information (DCI) and/or high-layer signaling, and the high-layer signaling includes radio resource control signaling or media access layer control unit signaling.

In combination with the first aspect, in certain implementations of the first aspect, the first DCI is any one of the following: the first DCI is also used to trigger first CSI feedback information, and the first CSI feedback information belongs to a non-periodic CSI report; the first DCI is also used to schedule or configure an uplink shared channel, and the uplink shared channel does not include CSI feedback information; or the first DCI is also used to carry first indication information of other terminal devices.

In conjunction with the first aspect, in certain implementations of the first aspect, the input of the first AI model includes a channel measurement result of a first reference signal, the first CSI feedback information is the output of the first AI model or is based on the output of the first AI model, the output of the first AI model is related to the measurement result of the first reference signal and reset state information of the first AI model, and the reset state information of the first AI model is initial state information, or the reset state information of the first AI model is state information of the first AI model before the first reference signal is sent.

In a second aspect, a communication method is provided. The method can be executed by a first device or a chip or circuit of the first device.

The first device may be a device on the second AI model side. The device on the second AI model side may be replaced by a device on the terminal device side or a device on the network device side. The terminal device side may include at least one of the terminal device and an AI entity on the terminal device side. The AI entity on the terminal device side may be the terminal device itself, or an AI entity serving the terminal device, such as a server, such as an OTT server or a cloud server. The network device side may include at least one of the network device and an AI entity on the network device side. The AI entity on the network device side may be the network device itself, or an AI entity serving the network device, such as a RIC, OAM, or a server, such as an OTT server or a cloud server.

The method includes: sending first indication information to the second device, where the first indication information is used to determine the time unit in which reset status information of the first AI model takes effect.

The first AI model and the second AI model are autoencoding models.

The second device is the device on the first AI model side.

A first AI model, such as the AI model on the terminal device side, can be used to generate CSI feedback information. A second AI model, such as the AI model on the network device side, that matches the first AI model, such as the AI model on the terminal device side, can be used to recover channel information corresponding to the CSI feedback information.

Optionally, the device on the first AI model side may be a device on the terminal device side or a device on the network device side.

The device on the second AI model side sending the first indication information to the second device may include a network device or a chip used for the network device sending the first indication information, or an AI entity on the network device side or a chip used for the AI entity on the network device side sending the first indication information. Optionally, the AI entity on the network device side or the chip used for the AI entity on the network device side may send the first indication information to the second device by forwarding the network device.

In combination with the second aspect, in certain implementations of the second aspect, the method further includes: receiving first CSI feedback information from the second device in a first time unit, the first CSI feedback information is related to the reset status information of the first AI model, and the first time unit is no earlier than the time unit in which the reset status information of the first AI model takes effect.

In combination with the second aspect, in certain implementations of the second aspect, the first indication information is used to indicate that uplink information is sent on a first time resource, and the effective time unit of the reset status information of the first AI model is no later than the start time of the first time resource.

In combination with the second aspect, in some implementations of the second aspect, the first indication information indicates the length of the first time period, or the method also includes: sending fourth indication information to the second device, the fourth indication information indicates the length of the first time period, and the end time of the first time period is no later than the start time of the first time resource.

In combination with the second aspect, in certain implementations of the second aspect, the first indication information indicates the length of the first time period, or the method further includes: sending fourth indication information to the second device, the fourth indication information indicating the length of the first time period, and the length of the first time period is used to determine the time unit in which the reset status information of the first AI model takes effect.

In combination with the second aspect, in certain implementations of the second aspect, the starting time of the first time period is the time when the first indication information is sent, or the time when the first indication information is received, or the time indicated by the first indication information.

In combination with the second aspect, in certain implementations of the second aspect, the method further includes: sending scheduling information of third CSI feedback information, where the time resources of the third CSI feedback information indicated by the scheduling information do not include the first time period; and receiving third CSI feedback information, where the third CSI feedback information is related to the first AI model.

In combination with the second aspect, in certain implementations of the second aspect, the first indication information indicates a reset of status information of the first AI model.

In combination with the second aspect, in some implementations of the second aspect, the method further includes: sending second indication information to the second device, where the second indication information indicates resetting the status information of the first AI model.

In combination with the second aspect, in certain implementations of the second aspect, the first indication information is related to the duration required to reset the status information of the first AI model, and the duration required to reset the status information of the first AI model is predefined.

In combination with the second aspect, in certain implementations of the second aspect, the first indication information is related to the time required to reset the status information of the first AI model, and the method also includes: receiving third indication information from the second device or other devices other than the second device, the third indication information indicating the time required to reset the status information of the first AI model.

In combination with the second aspect, in certain implementations of the second aspect, the third indication information is carried in signaling sent by the terminal device that carries the terminal device capabilities, or the third indication information is carried in signaling that carries the configuration information of the first AI model.

In combination with the second aspect, in certain implementations of the second aspect, the first indication information is carried in first downlink control information DCI and/or higher-layer signaling, where the higher-layer signaling includes radio resource control signaling or media access layer control unit signaling.

In combination with the second aspect, in certain implementations of the second aspect, the first DCI is any one of the following: the first DCI is also used to trigger first CSI feedback information, and the first CSI feedback information belongs to a non-periodic CSI report; the first DCI is also used to schedule or configure an uplink shared channel, and the uplink shared channel does not include CSI feedback information; or the first DCI is also used to carry first indication information of other terminal devices.

In conjunction with the second aspect, in certain implementations of the second aspect, the input of the first AI model includes a channel measurement result of a first reference signal, the first CSI feedback information is the output of the first AI model or is based on the output of the first AI model, the output of the first AI model is related to the measurement result of the first reference signal and reset state information of the first AI model, and the reset state information of the first AI model is initial state information, or the reset state information of the first AI model is state information of the first AI model before the time when the first reference signal was sent.

In a third aspect, a communication device is provided. The communication device may be a terminal device, or a device, module, circuit, or chip configured and provided in the terminal device, or a device capable of being used in conjunction with the terminal device. In one design, the communication device may include a module corresponding to executing the method/operation/step/action described in the first aspect. The module may be a hardware circuit, software, or a combination of hardware circuit and software. In one design, the communication device may include a processing module and a communication module.

Among them, the sending module is used to execute the sending action in the method described in the first aspect above, the processing module is used to execute the processing-related actions in the method described in the first aspect above, and the receiving module is used to execute the receiving-related actions in the method described in the first aspect above.

In a fourth aspect, a communication device is provided. The communication device may be a network device, or a device, module, circuit, or chip configured and provided in the network device, or a device capable of being used in conjunction with the network device. In one design, the communication device may include a module corresponding to executing the method/operation/step/action described in the second aspect. The module may be a hardware circuit, software, or a combination of hardware circuit and software. In one design, the communication device may include a processing module and a communication module.

Among them, the receiving module is used to perform the receiving action in the method described in the second aspect above, the processing module is used to perform the processing-related actions in the method described in the second aspect above, and the sending module is used to perform the sending action in the method described in the second aspect above.

In a fifth aspect, a communication device is provided, comprising one or more processors coupled to one or more storage media, the one or more storage media storing instructions, which, when executed by the one or more processors, enable the method in the first aspect or any possible implementation of the first aspect to be implemented, or enable the method in the second aspect or any possible implementation of the second aspect to be implemented.

In a sixth aspect, a communication device is provided, comprising one or more processors, wherein the one or more processors are used to process data and/or information so that the method in the first aspect or any possible implementation of the first aspect is implemented, or the method in the second aspect or any possible implementation of the second aspect is implemented.

Optionally, the communication device may further include a communication interface, the communication interface being configured to receive data and/or information and transmit the received data and/or information to the processor. Optionally, the communication interface is further configured to output the data and/or information processed by the processor.

In the seventh aspect, a chip is provided, comprising a processor, wherein the processor is used to run a program or instruction so that the method in the first aspect or any possible implementation of the first aspect is implemented, or the method in the second aspect or any possible implementation of the second aspect is implemented.

Optionally, the chip may further include a memory for storing programs or instructions. Optionally, the chip may further include the transceiver.

In an eighth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes instructions, which, when executed by a processor, enable the method in the first aspect or any possible implementation of the first aspect to be implemented, or enable the method in the second aspect or any possible implementation of the second aspect to be implemented.

In a ninth aspect, a computer program product is provided, comprising computer program code or instructions, which, when executed, enables the method according to the first aspect or any possible implementation of the first aspect to be implemented, or enables the method according to the second aspect or any possible implementation of the second aspect to be implemented.

In a tenth aspect, a communication system is provided, comprising a combination of one or more of the following apparatuses: a communication apparatus that performs the method of the first aspect or any possible implementation of the first aspect, and a communication apparatus that performs the method of the second aspect or any possible implementation of the second aspect. For example, the communication system may include the communication apparatus provided in the third aspect and/or the communication apparatus provided in the fourth aspect.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a schematic diagram of a communication system applicable to an embodiment of the present application;

FIG2 is a schematic diagram of another communication system applicable to an embodiment of the present application;

FIG3 is a schematic block diagram of an autoencoder;

FIG4 is a schematic diagram of an AI application framework;

FIG5 is a schematic diagram of a CSI feedback process based on time domain correlation;

FIG6 is a schematic diagram of another CSI feedback process based on time domain correlation;

FIG7 is a schematic flow chart of a communication method according to an embodiment of the present application;

FIG8 is a schematic flow chart of another communication method according to an embodiment of the present application;

FIG9 is a schematic flow chart of another communication method according to an embodiment of the present application;

FIG10 is a schematic flow chart of another communication method according to an embodiment of the present application;

FIG11 is a schematic flow chart of another communication method according to an embodiment of the present application;

FIG12 is a schematic flow chart of another communication method according to an embodiment of the present application;

FIG13 is a schematic flow chart of another communication method according to an embodiment of the present application;

FIG14 is a schematic diagram of a CSI feedback process according to an embodiment of the present application;

FIG15 is a schematic flow chart of another communication method according to an embodiment of the present application;

FIG16 is a schematic flow chart of another communication method according to an embodiment of the present application;

FIG17 is a schematic diagram of another CSI feedback process according to an embodiment of the present application;

FIG18 is a schematic diagram of four types of indication signaling of the first indication information according to an embodiment of the present application;

FIG19 is a schematic block diagram of a communication device provided in an embodiment of the present application;

FIG20 is a schematic block diagram of another communication device provided in an embodiment of the present application;

Figure 21 is a schematic diagram of another CSI feedback information process provided in an embodiment of the present application.

DETAILED DESCRIPTION

The technical solution in this application will be described below with reference to the accompanying drawings.

The technical solutions provided in this application can be applied to various communication systems, such as fifth-generation (5G) or new radio (NR) systems, long-term evolution (LTE) systems, LTE frequency division duplex (FDD) systems, LTE time division duplex (TDD) systems, wireless local area networks (WLAN) systems, satellite communication systems, future communication systems such as future communication networks, or a fusion system of multiple systems. The technical solutions provided in this application can also be applied to device-to-device (D2D) communication, vehicle-to-everything (V2X) communication, machine-to-machine (M2M) communication, machine type communication (MTC), and Internet of Things (IoT) communication systems or other communication systems.

A device in a communication system can send a signal to another device or receive a signal from another device. The signal may include information, signaling, or data, etc. The device can also be replaced by an entity, a network entity, a network element, a communication device, a communication module, a node, a communication node, etc. The present disclosure uses the device as an example for description. For example, a communication system may include at least one terminal device and at least one network device. In the communication system, a network device can send a downlink signal to a terminal device, a terminal device can send an uplink signal to a network device, a network device can send a signal to another network device, and a terminal device can send a sidelink signal to another terminal device. It is understandable that the terminal device in the present disclosure can be replaced by the second device, and the network device can be replaced by the first device, and the two perform the corresponding communication methods in the present disclosure.

In an embodiment of the present application, the terminal device may also be referred to as user equipment (UE), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent or user device.

A terminal device can be a device that provides voice/data, such as a handheld device or vehicle-mounted device with wireless connection function. Currently, some examples of terminals are: mobile phones, tablet computers, laptop computers, PDAs, mobile internet devices (MIDs), wearable devices, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, wireless terminals in self-driving, wireless terminals in telemedicine, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, etc. ty), wireless terminals in smart homes, cellular phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), handheld devices with wireless communication capabilities, computing devices or other processing devices connected to wireless modems, wearable devices, terminal devices in 5G networks or terminal devices in future evolved public land mobile networks (PLMNs), etc., and the embodiments of the present application are not limited to these.

As an example and not a limitation, in the embodiment of the present application, the terminal device may also be a wearable device. Wearable devices may also be called wearable smart devices, which are a general term for wearable devices that are intelligently designed and developed using wearable technology for daily wear, such as glasses, gloves, watches, clothing, and shoes. A wearable device is a portable device that is worn directly on the body or integrated into the user's clothes or accessories. Wearable devices are not only hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are fully functional, large in size, and can achieve complete or partial functions without relying on smartphones, such as smart watches or smart glasses, as well as those that only focus on a certain type of application function and need to be used in conjunction with other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

In the embodiments of the present application, the device for realizing the function of the terminal device can be a terminal device, or a device capable of supporting the terminal device to realize the function, such as a chip system, which can be installed in the terminal device or used in combination with the terminal device. In the embodiments of the present application, the chip system can be composed of a chip, or it can include a chip and other discrete devices. In the embodiments of the present application, only the terminal device is used as an example for description, and the embodiments of the present application are not limited to the solutions of the embodiments of the present application.

The network device in the embodiments of the present application may be a device for communicating with a terminal device, and the network device may also be referred to as an access network device or a wireless access network device. For example, the network device may be a base station. The network device in the embodiments of the present application may refer to a RAN node (or device) that connects a terminal device to a wireless network. The term “base station” can broadly cover the following names or be replaced by the following names, such as: NodeB, evolved NodeB (eNB), next generation NodeB (gNB), relay station, access point, transmitting and receiving point (TRP), transmitting point (TP), master station, auxiliary station, motor slide retainer (MSR) node, home base station, network controller, access node, wireless node, access point (AP), transmission node, transceiver node, baseband unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), central unit (CU), distributed unit (DU), radio unit (RU), positioning node, etc. The base station can be a macro base station, a micro base station, a relay node, a donor node or the like, or a combination thereof. The base station can also refer to a communication module, a modem or a chip used to be set in the aforementioned equipment or device. The base station can also be a mobile switching center and a device that performs the base station function in D2D, V2X, and M2M communications, a network side device in a future communication network, a device that performs the base station function in a future communication system, etc. The base station can support networks with the same or different access technologies. Optionally, the RAN node can also be a server, a wearable device, a vehicle or an on-board device, etc. For example, the access network device in the V2X technology can be a road side unit (RSU). The embodiments of the present application do not limit the specific technology and specific device form adopted by the network equipment.

Base stations can be fixed or mobile. For example, a helicopter or drone can be configured to act as a mobile base station, and one or more cells can move based on the location of the mobile base station. In other examples, a helicopter or drone can be configured to act as a device that communicates with another base station.

In some deployments, the network devices mentioned in the embodiments of the present application may include a CU, a DU, or both a CU and a DU, or a device including a control plane CU node (central unit-control plane (CU-CP)), a user plane CU node (central unit-user plane (CU-UP)), and a DU node. For example, the network devices may include a gNB-CU-CP, a gNB-CU-UP, and a gNB-DU.

In some deployments, multiple RAN nodes collaborate to assist terminals in achieving wireless access, with different RAN nodes implementing portions of the base station's functionality. For example, a RAN node can be a CU, DU, CU-CP, CU-UP, or RU. The CU and DU can be separate or included in the same network element, such as the BBU. The RU can be included in a radio frequency device or radio unit, such as an RRU, AAU, or RRH.

A RAN node can support one or more types of fronthaul interfaces, and different fronthaul interfaces correspond to DUs and RUs with different functions.

If the fronthaul interface between the DU and the RU is the common public radio interface (CPRI), the DU is configured to implement one or more baseband functions, and the RU is configured to implement one or more radio frequency functions.

If the fronthaul interface between the DU and the RU is another interface, relative to the CPRI, part of the downlink and/or uplink baseband functions, for example, for downlink, one or more of precoding, digital beamforming (BF), or inverse fast Fourier transform (IFFT)/cyclic prefix (CP) are moved from the DU to the RU; for uplink, one or more of digital beamforming (BF), or fast Fourier transform (FFT)/cyclic prefix (CP) are moved from the DU to the RU.

In one possible implementation, the interface could be the enhanced common public radio interface (eCPRI). In the eCPRI architecture, the division between the DU and RU is different, corresponding to different eCPRI categories (Categories A, B, C, D, E, and F).

Taking eCPRI Cat A as an example, for downlink transmission, based on layer mapping, the DU is configured to implement layer mapping and one or more functions before it (i.e., one or more of coding, rate matching, scrambling, modulation, and layer mapping), while other functions after layer mapping (for example, one or more of resource element (RE) mapping, digital beamforming (BF), or inverse fast Fourier transform (IFFT)/adding a cyclic prefix (CP)) are moved to the RU for implementation. For uplink transmission, the DU is configured to perform demapping and one or more of the preceding functions (i.e., decoding, rate matching, descrambling, demodulation, inverse discrete Fourier transform (IDFT), channel equalization, and demapping), with demapping being the key division. Other functions after demapping (e.g., one or more of digital BF or fast Fourier transform (FFT)/CP removal) are implemented in the RU. For a functional description of the DU and RU corresponding to various types of eCPRI, please refer to the eCPRI protocol and will not be elaborated on here.

In one possible design, the processing unit used to implement baseband functions in the BBU is called a baseband high (BBH) unit, and the processing unit used to implement baseband functions in the RRU/AAU/RRH is called a baseband low (BBL) unit.

In different systems, CU (or CU-CP and CU-UP), DU, or RU may have different names, but those skilled in the art will understand their meanings. For example, in an open RAN (ORAN) system, CU may also be called O-CU (open CU), DU may also be called O-DU, CU-CP may also be called O-CU-CP, CU-UP may also be called O-CU-UP, and RU may also be called O-RU. Any of the CU (or CU-CP, CU-UP), DU, and RU in this application may be implemented by a software module, a hardware module, or a combination of software and hardware modules.

In the embodiments of the present application, the device for implementing the functions of the network device can be a network device; it can also be a device that can support the network device to implement the functions, such as a chip system, a hardware circuit, a software module, or a hardware circuit and a software module. The device can be installed in the network device or used in conjunction with the network device. In the embodiments of the present application, only the device for implementing the functions of the network device is used as an example to illustrate, and does not constitute a limitation on the solutions of the embodiments of the present application.

The network equipment and/or terminal devices can be deployed on land, including indoors, outdoors, handheld, and/or vehicle-mounted; can also be deployed on water (such as ships); and can also be deployed in the air (such as aircraft, balloons, and/or satellites). The embodiments of this application do not limit the scenarios in which the network equipment and terminal devices are located.

In addition, terminal devices and network devices can be hardware devices, or software functions running on dedicated hardware, software functions running on general-purpose hardware, such as virtualization functions instantiated on a platform (for example, a cloud platform), or entities including dedicated or general-purpose hardware devices and software functions. This application does not limit the specific form of terminal devices and network devices.

In wireless communication networks, such as mobile communication networks, the services supported by the networks are becoming increasingly diverse, and therefore the demands that need to be met are becoming increasingly diverse. For example, the network needs to be able to support ultra-high speeds, ultra-low latency, and/or ultra-large connections. This feature makes network planning, network configuration, and/or resource scheduling increasingly complex. In addition, as network functionality becomes increasingly powerful, such as supporting increasingly high spectrum, supporting advanced multiple input multiple output (MIMO) technology, supporting beamforming, and/or supporting new technologies such as beam management, network energy conservation has become a hot research topic. These new demands, new scenarios, and new features have brought unprecedented challenges to network planning, maintenance, and efficient operation. To meet this challenge, artificial intelligence technology can be introduced into wireless communication networks to achieve network intelligence.

In order to support AI technology in wireless networks, AI nodes (also called AI entities) may be introduced into the network.

Optionally, the AI entity can be deployed in one or more of the following locations in the communication system: access network equipment, terminal equipment, or core network equipment. Alternatively, the AI entity can be deployed separately, for example, in a location other than any of the aforementioned devices, such as a host or cloud server in an OTT system. The AI entity can communicate with other devices in the communication system, such as one or more of the following: network equipment, terminal equipment, or core network elements. Based on the objects served by the AI entity, the AI entity can include an AI entity on the network device side, an AI entity on the terminal device side, or an AI entity on the core network side.

It is understood that this application does not limit the number of AI entities. For example, when there are multiple AI entities, the multiple AI entities can be divided based on function, such as different AI entities are responsible for different functions.

It can also be understood that AI entities can be independent devices, or they can be integrated into the same device to implement different functions, or they can be network elements in hardware devices, or they can be software functions running on dedicated hardware, or they can be virtualized functions instantiated on a platform (for example, a cloud platform). This application does not limit the specific form of the above-mentioned AI entities.

The AI entity can be an AI network element or an AI module. The AI entity is used to implement the corresponding AI function. The AI modules deployed in different network elements can be the same or different. The AI model in the AI entity can implement different functions according to different parameter configurations. The AI model in the AI entity can be configured based on one or more of the following parameters: structural parameters (such as the number of neural network layers, the width of the neural network, the connection relationship between layers, the weight of the neuron, the activation function of the neuron, or at least one of the bias in the activation function), input parameters (such as the type of input parameters and/or the dimension of the input parameters), or output parameters (such as the type of output parameters and/or the dimension of the output parameters). Among them, the bias in the activation function can also be called the bias of the neural network.

An AI entity can have one or more models. The learning, training, or inference processes of different models can be deployed in different entities or devices, or in the same entity or device.

FIG1 is a schematic diagram of a communication system applicable to the communication method of an embodiment of the present application. As shown in FIG1 , the communication system 100 may include at least one network device, such as the network device 110 shown in FIG1 ; the communication system 100 may also include at least one terminal device, such as the terminal device 120 and the terminal device 130 shown in FIG1 . The network device 110 and the terminal device (such as the terminal device 120 and the terminal device 130) can communicate via a wireless link. The communication devices in the communication system, for example, the network device 110 and the terminal device 120, can communicate via multi-antenna technology.

Figure 2 is a schematic diagram of another communication system applicable to the communication method of an embodiment of the present application. Compared to the communication system 100 shown in Figure 1, the communication system 200 shown in Figure 2 also includes an AI network element 140. AI network element 140 is used to perform AI-related operations, such as constructing a training dataset or training an AI model.

In one possible implementation, the network device 110 may send data related to the training of the AI model to the AI network element 140, which constructs a training data set and trains the AI model. For example, the data related to the training of the AI model may include data reported by the terminal device. The AI network element 140 may send the results of the operations related to the AI model to the network device 110, and forward them to the terminal device through the network device 110. For example, the results of the operations related to the AI model may include at least one of the following: an AI model that has completed training, an evaluation result or a test result of the model, etc. Exemplarily, a portion of the trained AI model may be deployed on the network device 110, and another portion may be deployed on the terminal device. Alternatively, the trained AI model may be deployed on the network device 110. Alternatively, the trained AI model may be deployed on the terminal device.

It should be understood that Figure 2 illustrates only the example of a direct connection between AI network element 140 and network device 110. In other scenarios, AI network element 140 may also be connected to a terminal device. Alternatively, AI network element 140 may be connected to both network device 110 and a terminal device simultaneously. Alternatively, AI network element 140 may be connected to network device 110 through a third-party network element. This embodiment of the present application does not limit the connection relationship between the AI network element and other network elements.

AI network element 140 can also be provided as a module in a network device and/or a terminal device, for example, in network device 110 or a terminal device shown in FIG1 . One or more AI modules can be deployed in network device 110. One or more AI modules can be deployed in a terminal device.

It should be noted that Figures 1 and 2 are simplified schematic diagrams for ease of understanding. For example, the communication system may also include other devices, such as wireless relay devices and/or wireless backhaul devices, which are not shown in Figures 1 and 2. In actual applications, the communication system may include multiple network devices and multiple terminal devices. The embodiments of the present application do not limit the number of network devices and terminal devices included in the communication system.

To facilitate understanding of the solutions of the embodiments of the present application, the terms that may be involved in the embodiments of the present application are explained below.

(1) AI model:

An AI model is an algorithm or computer program that can implement AI functions. The AI model represents the mapping relationship between the model's input and output. An AI model can be understood as a function model that maps inputs of a certain dimension to outputs of a certain dimension, and its model parameters are obtained through machine learning training. For example, f(x) = ax ² + b is a quadratic function model, which can be regarded as an AI model. a and b correspond to the parameters of the AI model and can be obtained through machine learning training. An AI model can also be called a model, AI function, or function. An AI function can correspond to one or more AI models.

The type of AI model can be a neural network, linear regression model, decision tree model, support vector machine (SVM), Bayesian network, Q learning model or other machine learning (ML) model.

(2) Two-end model:

The two-end model can also be called a bilateral model, collaborative model, dual model, or two-side model. A two-end model is a model composed of multiple sub-models. The sub-models that make up the model must match each other. These sub-models can be deployed on different nodes.

The embodiments of the present application relate to an encoder for compressing channel information and a decoder for recovering channel information. The encoder and decoder are used in matching manner, and it can be understood that the encoder and decoder are matching AI models. An encoder may include one or more AI models, and the decoder matched with the encoder also includes one or more AI models. The number of AI models included in the matching encoder and decoder is the same and one-to-one corresponding. The encoder may also include a quantization module, which may be used to quantize the output of the AI model in the encoder. The decoder may include an inverse quantization module, which may be used to inverse quantize the feedback information of the received channel information to obtain the input of the AI model in the decoder. Inverse quantization processing may also be called dequantization processing.

In one possible design, a set of matched encoders and decoders can be two parts of the same auto-encoder (AE). The AE model in which the encoder and decoder are deployed on different nodes is a typical bilateral model. The encoder and decoder of the AE model are usually a jointly trained encoder and decoder that are matched and used. An autoencoder is a neural network for unsupervised learning. Its characteristic is that it uses input data as label data, so the autoencoder can also be understood as a neural network for self-supervised learning. An autoencoder can be used for data compression and recovery. For example, the encoder in the autoencoder can compress (encode) data A to obtain data B; the decoder in the autoencoder can decompress (decode) data B to restore data A. Or it can be understood that the decoder is the inverse operation of the encoder.

For example, as shown in FIG3 , the encoder processes the input V to obtain a processed result z, and the decoder can decode the encoder output z into the desired output V’.

The AI model in the embodiments of the present application may include an encoder deployed on the terminal device side and a decoder deployed on the network device side, or an encoder deployed on the terminal device side and a decoder deployed on another terminal device side, or an encoder deployed on the network device side and a decoder deployed on another network device side.

(3) Neural network (NN):

Neural networks are a specific implementation of AI or machine learning. According to the universal approximation theorem, neural networks can theoretically approximate any continuous function, giving them the ability to learn arbitrary mappings.

A neural network can be composed of neural units, which can be a computational unit that takes x _s and an intercept 1 as input. A neural network is formed by connecting many of the aforementioned single neural units together, meaning that the output of one neural unit can be the input of another. The input of each neural unit can be connected to the local receptive field of the previous layer to extract features from that local receptive field, which can be an area consisting of several neural units.

Taking neural networks as an example, the AI models involved in this disclosure may be deep neural networks (DNNs). Depending on how the network is constructed, DNNs can include feedforward neural networks (FNNs), convolutional neural networks (CNNs), and recurrent neural networks (RNNs).

CNN is a neural network specifically designed to process data with a grid-like structure. For example, time series data (discrete sampling along the time axis) and image data (discrete sampling along two dimensions) can both be considered grid-like data. CNNs do not utilize all input information at once for computation. Instead, they use a fixed-size window to intercept a portion of the information for convolution operations, significantly reducing the computational complexity of model parameters. Furthermore, depending on the type of information intercepted by the window (e.g., people and objects in an image represent different types of information), each window can use a different convolution kernel, enabling CNNs to better extract features from the input data.

RNNs are a type of DNN that utilizes feedback time series information. Their input consists of a new input value at the current moment and their own output value at the previous moment. RNNs are suitable for capturing temporally correlated sequence features and are particularly well-suited for applications such as speech recognition and channel coding.

The characteristic of FNN network is that neurons in adjacent layers are fully connected to each other, which makes FNN usually require a large amount of storage space and leads to high computational complexity.

The aforementioned FNNs, CNNs, and RNNs are all constructed based on neurons. As mentioned earlier, each neuron performs a weighted summation operation on its input values, and then applies this weighted summation to a nonlinear function to produce an output. The weights of the weighted summation operation of neurons in a neural network, as well as the nonlinear function, are called the parameters of the neural network. The parameters of all neurons in a neural network constitute the parameters of the neural network.

(4) Dataset:

A dataset refers to the data used for model training, verification, and testing in machine learning. The quantity and quality of the data will affect the effectiveness of machine learning.

In the field of machine learning, ground truth usually refers to data that is believed to be accurate or real.

A training dataset is used to train an AI model. It may include the input to the AI model, or the input and target output of the AI model. A training dataset includes one or more training data. Training data may include training samples input to the AI model, or the target output of the AI model. The target output may also be referred to as a label, sample label, or labeled sample. A label is the true value.

In the communications field, training datasets can include simulated data collected through simulation platforms, experimental data collected in experimental scenarios, or measured data collected in actual communication networks. Because the geographical environments and channel conditions in which data are generated vary, such as indoor and outdoor locations, mobile speeds, frequency bands, or antenna configurations, the collected data can be categorized during acquisition. For example, data with the same channel propagation environment and antenna configuration can be grouped together.

Model training essentially involves learning certain characteristics from training data. When training an AI model (such as a neural network), the goal is to ensure that the model's output is as close as possible to the desired predicted value. This is done by comparing the network's predictions with the desired target values. The weight vectors of each layer of the AI model are then updated based on the difference between the two. (Of course, this initialization process typically precedes the first update, where parameters are preconfigured for each layer of the AI model.) For example, if the network's prediction is too high, the weight vectors are adjusted to predict a lower value. This adjustment is repeated until the AI model predicts the desired target value, or a value very close to it. Therefore, it's necessary to predefine how to compare the difference between the predicted and target values. This is known as the loss function, or objective function. These are crucial equations used to measure the difference between the predicted and target values. For example, a higher loss function indicates a greater discrepancy. Therefore, training the AI model becomes a process of minimizing this loss, keeping the loss function below a threshold or ensuring that the loss function meets the desired target. For example, the AI model is a neural network, and adjusting the model parameters of the neural network includes adjusting at least one of the following parameters: the number of layers, width, weights of neurons, or parameters in the activation function of neurons of the neural network.

Inference data can be used as input to a trained AI model for inference. During the inference process, the inference data is input into the AI model, and the corresponding output is the inference result.

(5) AI model design:

The design of an AI model primarily involves data collection (e.g., collecting training data and/or inference data), model training, and model inference. Furthermore, it can also include the application of inference results.

FIG4 shows an AI application framework.

In the aforementioned data collection phase, the data source provides training datasets and inference data. In the model training phase, an AI model is generated by analyzing or training the training data provided by the data source. The AI model represents the mapping relationship between the model's inputs and outputs. Learning the AI model through the model training node is equivalent to learning the mapping relationship between the model's inputs and outputs using the training data. In the model inference phase, the AI model, trained in the model training phase, performs inference based on the inference data provided by the data source, generating an inference result. This phase can also be understood as inputting inference data into the AI model and generating an output, which is the inference result. The inference result can indicate the configuration parameters used (executed) by the execution object and/or the operations performed by the execution object. In the inference result application phase, the inference result is published. For example, the inference result can be centrally planned by an actor, for example, the actor can send the inference result to one or more actors (e.g., network devices or terminal devices) for execution. Furthermore, the actor can provide feedback on model performance to the data source to facilitate subsequent model updates and training.

It is understandable that a communication system may include network elements with AI functions. The above-mentioned AI model design-related links can be performed by one or more network elements with artificial intelligence functions. In one possible design, AI functions (such as AI modules or AI entities) can be configured in existing network elements in the communication system to implement AI-related operations, such as AI model training and/or reasoning. For example, the existing network element can be a network device or a terminal device. Or in another possible design, an independent network element can also be introduced into the communication system to perform AI-related operations, such as training an AI model. The independent network element can be called an AI network element or an AI node, etc., and the embodiments of the present application do not limit these names. For example, the AI network element can be directly connected to the network equipment in the communication system, or it can be indirectly connected to the network equipment through a third-party network element. Among them, the third-party network element can be a core network element such as an authentication management function (AMF) network element, a user plane function (UPF) network element, an OAM, a cloud server, or other network elements, without limitation. Exemplarily, the independent network element can be deployed on one or more of the following: the network device side, the terminal device side, or the core network side. Optionally, it can be deployed on a cloud server, OTT, or OAM. Exemplarily, the communication system shown in FIG2 introduces an AI network element 140.

The training process of different models can be deployed in different devices or nodes, or in the same device or node. The inference process of different models can be deployed in different devices or nodes, or in the same device or node. Taking the completion of the model training phase of a terminal device as an example, the terminal device can train the matching encoder and decoder, and then send the model parameters of the decoder to the network device. Taking the completion of the model training phase of a network device as an example, after the network device trains the matching encoder and decoder, it can indicate the model parameters of the encoder to the terminal device. Taking the completion of the model training phase of an independent AI network element as an example, the AI network element can train the matching encoder and decoder, and then send the model parameters of the encoder to the terminal device and the model parameters of the decoder to the network device. Then, the model inference phase corresponding to the encoder is performed in the terminal device, and the model inference phase corresponding to the decoder is performed in the network device.

Among them, the model parameters may include one or more of the following structural parameters of the model (such as the number of layers and/or weights of the model, etc.), the input parameters of the model (such as input dimension, number of input ports), or the output parameters of the model (such as output dimension, number of output ports). It can be understood that the input dimension may refer to the size of an input data. For example, when the input data is a sequence, the input dimension corresponding to the sequence may indicate the length of the sequence. The number of input ports may refer to the number of input data. Similarly, the output dimension may refer to the size of an output data. For example, when the output data is a sequence, the output dimension corresponding to the sequence may indicate the length of the sequence. The number of output ports may refer to the number of output data.

(6) Channel information:

In communication systems (e.g., LTE or NR), network equipment determines one or more of the following configurations, including resources, MCS, and precoding, for scheduling downlink data channels of terminal devices based on channel information. Channel information, also known as channel state information (CSI) or channel environment information, reflects channel characteristics and quality.

Channel information measurement refers to the receiver determining channel information based on a reference signal sent by the transmitter, i.e., estimating the channel information using a channel estimation method. For example, the reference signal may include one or more of a channel state information reference signal (CSI-RS), a synchronization signal/physical broadcast channel block (SSB), a sounding reference signal (SRS), or a demodulation reference signal (DMRS). One or more of CSI-RS, SSB, and DMRS can be used to measure downlink channel information. SRS and/or DMRS can be used to measure uplink channel information.

The channel information may be determined based on a channel measurement result of a reference signal. Alternatively, the channel information may be a channel measurement result of a reference signal. In an embodiment of the present application, the channel measurement result of a reference signal may also be replaced by the channel information.

Taking the FDD communication scenario as an example, in the FDD communication scenario, since the uplink and downlink channels are not reciprocal or the reciprocity of the uplink and downlink channels cannot be guaranteed, the network device needs to obtain the downlink CSI through the uplink feedback of the terminal device. The network device usually sends a downlink reference signal to the terminal device, and the terminal device receives the downlink reference signal. Since the terminal device knows the sending information of the downlink reference signal, the terminal device can perform channel measurement and interference measurement based on the received downlink reference signal to estimate (measure) the downlink channel experienced by the downlink reference signal. The terminal device generates the downlink CSI based on the downlink channel matrix obtained by the measurement. The terminal device generates a CSI report according to the method predefined by the protocol or the method configured by the network device, and feeds it back to the network device so that it can obtain the downlink CSI.

In the present application, the meaning of CSI is broader than that of CSI in traditional schemes, and is not limited to channel quality indication (CQI), precoding matrix indicator (PMI), rank indicator (RI), or CSI-RS resource indicator (CRI). It can also be one or more of channel response information (such as channel response matrix, frequency domain channel response information, time domain channel response information), weight information corresponding to channel response, reference signal receiving power (RSRP) or signal to interference plus noise ratio (SINR), etc.

The RI indicates the recommended number of downlink transmission layers for the reference signal receiver, such as a terminal device. The CQI indicates the modulation and coding scheme supported by the current channel conditions determined by the reference signal receiver, such as a terminal device. The PMI indicates the recommended precoding layer for the reference signal receiver, such as a terminal device. The number of precoding layers indicated by the PMI corresponds to the RI.

As previously described, channel information can be obtained by measuring the reference signal. Feedback information can be obtained by compressing and/or quantizing the channel information. The feedback information can be reported via a channel information report. Channel information can be recovered by decompressing and/or dequantizing the feedback information.

Feedback information may also be referred to as feedback information of channel information, feedback information of CSI, CSI feedback information, compressed information, compressed information of channel information, compressed information of CSI, compressed channel information, or compressed CSI, etc.

The recovered channel information may also be referred to as CSI recovery information.

As MIMO system antenna arrays continue to grow in size, the number of supported antenna ports increases, and the corresponding channel and precoding matrix dimensions grow. Approximating large channel and precoding matrices with a limited number of predefined codewords increases errors. One approach to improving channel recovery accuracy is to increase the number of codewords in the codebook. However, this also increases the overhead of CSI feedback (including the corresponding codeword number and one or more weighting coefficients), thereby reducing the resources available for data transmission and resulting in system capacity loss.

The introduction of AI technology into wireless communication networks has resulted in an AI-based CSI feedback method, known as AI-CSI feedback. Terminal devices use AI models to compress and feedback CSI, while network devices use AI models to decompress and recover the compressed CSI. AI-based CSI feedback transmits a sequence (such as a bit sequence), resulting in lower overhead than traditional CSI feedback. Furthermore, AI models have stronger nonlinear feature extraction capabilities, enabling more efficient compression and representation of channel information and more effective channel recovery based on feedback information compared to traditional solutions.

CSI feedback can be implemented based on an AI model for automated event processing (AE). For example, in Figure 3, the encoder can be a CSI generator, and the decoder can be a CSI reconstructor. For example, the encoder can be deployed in a terminal device, and the decoder can be deployed in a network device. Channel information V is passed through the encoder to generate CSI feedback information z. The decoder reconstructs the channel information, resulting in recovered channel information V'.

Channel information V can be obtained through channel information measurement. For example, channel information V can include the eigenvector matrix (a matrix composed of eigenvectors) of the downlink channel. The encoder processes the eigenvector matrix of the downlink channel to obtain CSI feedback information z. In other words, the compression and/or quantization operations of the eigenvector matrix based on the codebook in related schemes are replaced by operations in which the encoder processes the eigenvector matrix to obtain CSI feedback information z. The decoder processes the CSI feedback information z to obtain recovered channel information V'.

The following further illustrates the training process and reasoning process of the AI model in the embodiments of the present application.

The training data used to train AI models includes training samples and sample labels. For example, the training samples are channel information measured by the terminal device, and the sample labels are actual channel information, such as ground-truth CSI. If the encoder and decoder belong to the same autoencoder, the training data can only include the training samples, or in other words, the training samples are the sample labels.

In the field of wireless communications, the true CSI may be high-precision CSI.

The specific training process is as follows: the model training node uses the encoder to process the channel information, that is, the training samples, to obtain CSI feedback information, and uses the decoder to process the feedback information to obtain the recovered channel information, that is, the CSI recovery information. Then, the difference between the CSI recovery information and the corresponding sample label is calculated, that is, the value of the loss function, and the parameters of the encoder and decoder are updated according to the value of the loss function, so that the difference between the recovered channel information and the corresponding sample label is minimized, that is, the loss function is minimized. Exemplarily, the loss function can be the minimum mean square error (MSE) or cosine similarity. Repeating the above operations can obtain an encoder and decoder that meet the target requirements. The above model training node can be a terminal device, a network device, or other network element with AI function in the communication system. The loss of CSI feedback information will affect the feedback performance, that is, it will affect the accuracy of channel information recovery on the network device side. The following is an example of CSI feedback based on time domain correlation.

In wireless communication links, the channels of medium- and low-speed users vary continuously over time. This can improve feedback performance by exploiting channel time-domain correlation. For example, channel information compression can be achieved by exploiting the time-domain correlation between historical and current channel measurement results. This reduces the overhead of feedback channel information while minimizing information loss during the compression process. Terminal devices and network equipment can each leverage time-domain correlation for compressed feedback and recovery of channel information.

Figure 5 shows a schematic diagram of a CSI feedback process based on time domain correlation. In a CSI feedback scenario based on time domain correlation, when CSI feedback information is generated on the terminal device side, the CSI feedback information is not only related to the current channel measurement result, but also to the channel measurement result at the historical moment. When the channel information is restored on the network device side, the restored channel information is not only related to the current CSI feedback information, but also to the CSI feedback information at the historical moment. It can be understood that this application is described using the communication between a terminal device and a network device as an example, but the solution of this application can also be applied to other sending ends and receiving ends, such as wireless communications between terminal devices and terminal devices, or network devices and network devices, and AI entities are respectively deployed on the sending end side and the receiving end side for compression and recovery of CSI.

A dual-end model based on time-domain correlation is deployed on both the terminal device and the network equipment to implement CSI feedback based on time-domain correlation. This CSI feedback process is also known as AI-CSI dual-end compressed feedback. The feedback process shown in Figure 5 is explained below using the dual-end model shown in Figure 3 as an example.

On the terminal device side, the terminal device performs channel measurement on the reference signal R _t to obtain the current channel measurement result. As shown in Figure 5, the current channel measurement result H _t is input into the encoder, which outputs CSI feedback information c _t . The CSI feedback information c _t is related to the current channel measurement result H _t and the encoder state information e _t-1 . The encoder state information e _t-1 is determined based on the encoder's historical input. As shown in Figure 5, the encoder's historical input may include channel measurement results at historical moments. Therefore, the time domain correlation of channel information can be mined based on H _t and e _t-1 . As shown in Figure 5, t = 1, 2, 3, 4. On the network device side, the current CSI feedback information c _t is input into the decoder, which outputs CSI recovery information H' _t , which is the channel information corresponding to the reference signal R t. The CSI recovery information H' _t is related to the current CSI feedback information c _t and the decoder state information d _t-1 . The decoder's state information dt _-1 is determined based on the decoder's historical inputs. As shown in Figure 5, this historical input can include CSI feedback information from historical moments. Therefore, the time-domain correlation of channel information can be exploited based on _ct and dt _-1 . To ensure feedback performance, the state information on the network device and the terminal device must be updated synchronously. As shown in Figure 5, before the first execution, the encoder and decoder models can be initialized. After initialization, the encoder and decoder state information are both initial, denoted as _e0 and _d0 , respectively. Furthermore, in addition to outputting CSI feedback information _ct , the encoder also updates the terminal device's state information. In addition to outputting CSI recovery information _H't , the decoder also updates the network device's state information. This ensures synchronous updating of the network device and terminal device's state information.

The following example illustrates this. A network device sends reference signal R ₃ to a terminal device. The terminal device performs channel measurement on reference signal R ₃ to obtain channel measurement result H _3. This channel measurement result H ₃ is input into the terminal device's encoder. The encoder obtains CSI feedback information c ₃ based on the channel measurement result H ₃ and the encoder's state information e ₂ , and updates the encoder's state information to obtain state information e _3. Reference signal R ₃ is the reference signal corresponding to CSI feedback information c _3. The network device receives CSI feedback information c ₃ and inputs it into its decoder. The decoder obtains CSI recovery information H' ₃ based on the CSI feedback information c ₃ and the decoder's state information _{d 2} , and updates the decoder's state information to obtain state information d _3. This CSI recovery information H' ₃ is the channel information corresponding to CSI feedback information c ₃ , or in other words, this CSI recovery information H' ₃ is the channel information corresponding to reference signal R _3. The network device sends reference signal R ₄ to the terminal device. The terminal device performs channel measurement on reference signal R ₄ to obtain a channel measurement result H ₄ . This channel measurement result H ₄ is input into the terminal device's encoder. The encoder obtains CSI feedback information c ₄ based on the channel measurement result H ₄ and state information e ₃ , and updates the state information to obtain state information e ₄ . Reference signal R ₄ is the reference signal corresponding to CSI feedback information c ₄ . The network device receives CSI feedback information c ₄ and inputs _it into its decoder. The decoder obtains CSI recovery information H' ₄ based on CSI feedback information c ₄ and decoder state information d ₃ , and updates the decoder state information to obtain state information d ₄ . This CSI recovery information H' ₄ is the channel information corresponding to CSI feedback information c ₄ , or in other words, this CSI recovery information H' ₄ is the channel information corresponding to reference signal R ₄ .

Taking reference signal R ₃ as a reference, the reference signal received by the terminal device before receiving reference signal R ₃ is the historical reference signal. Taking reference signal R ₄ as a reference, the reference signal received by the terminal device before receiving reference signal R ₄ is the historical reference signal, for example, R _3. The encoder's state information is determined based on the encoder's historical inputs, which may include channel measurement results of historical reference signals. For example, as shown in Figure 5, state information e ₄ is updated based on channel measurement result H _3. Taking CSI feedback information c ₃ as a reference, the CSI feedback information received by the network device before receiving CSI feedback information c ₃ is the historical CSI feedback information. Taking CSI feedback information c ₄ as a reference, the CSI feedback information received by the network device before receiving CSI feedback information c ₄ is the historical CSI feedback information, for example, c _3. The decoder's state information is determined based on the decoder's historical inputs, which may include historical CSI feedback information. For example, as shown in Figure 5, state information d ₄ is updated based on CSI feedback information c ₃ .

When CSI feedback information fails to transmit due to poor channel transmission conditions or other reasons, the network device is unable to synchronously update its state information, resulting in a discrepancy between the terminal device's and network device's state information, which in turn reduces the encoder and decoder's matching. Alternatively, when feedback transmission fails, the terminal device is able to normally update its state information, but the network device, unable to obtain CSI feedback information, is unable to properly update its state information. This results in a discrepancy between the network device's and terminal device's state information updates, further leading to a discrepancy between the terminal device's and network device's state information, which in turn reduces the encoder and decoder's matching.

The discrepancy between the state information on the terminal device side and the network device side can prevent the decoder from recovering accurate channel information, resulting in performance loss. Furthermore, the discrepancy between the state information on the terminal device side and the network device side increases with the number of packet losses, potentially leading to a mismatch in subsequent dual-end processing and a continuous degradation of feedback performance.

When CSI feedback transmission failures, such as CSI feedback packet loss, occur frequently, the network device may trigger a reset of the state information of the AI model (ie, encoder) on the terminal device side, for example, triggering the state information of the AI model on the terminal device side to be reset to the initial state information e ₀ .

Figure 6 shows a schematic diagram of another CSI feedback process based on time-domain correlation. As shown in Figure 6, when feedback transmission fails, the decoder can process based on the last received CSI feedback information. For example, when CSI feedback information _c3 fails to be reported, the network device inputs the last received CSI feedback information _c2 into the decoder. The decoder outputs CSI recovery information _H'3 based on CSI feedback information _c2 and state information _d2 . Because CSI feedback information _c3 was not received, the network device cannot normally update the state information. The decoder state information _d3 and subsequent decoder state information in Figure 6 are different from the state information _d3 and subsequent decoder state information obtained after the update based on CSI feedback information _c3 in Figure 5. The decoded CSI recovery information _H'3 and subsequent CSI recovery information in Figure 6 are different from the CSI recovery information _H'3 and subsequent CSI recovery information in Figure 5. The accuracy of the decoded CSI recovery information _H'3 and subsequent CSI recovery information in FIG6 is lower than that of the CSI recovery information _H'3 and subsequent CSI recovery information in FIG5.

When CSI feedback transmission failures occur frequently, the network device triggers the terminal device to reset the encoder state information to the encoder's initial state information e ₀ . Correspondingly, the decoder state information is also reset to the decoder's initial state information d ₀ .

The network device sends reference signal R ₄ to the terminal device. The terminal device measures reference signal R ₄ to obtain a channel measurement result H ₄ . The channel measurement result H ₄ is input into the encoder. At this point, the encoder's state information has been reset. The encoder outputs CSI feedback information c ₄ based on the channel measurement result H ₄ and state information e ₀ , and updates state information e ₀ to obtain state information e ₁ '. This state information e ₁ ' can be used in the next CSI feedback process. The terminal device sends CSI feedback information c ₄ to the network device. The network device inputs CSI feedback information c ₄ into the decoder. At this point, the decoder's state information has been reset. The decoder outputs CSI recovery information H' ₄ based on CSI feedback information c ₄ and state information d ₀ , and updates state information d ₀ to obtain state information d ₁ '. This state information d ₁ ' can be used in the next CSI feedback process.

However, if the terminal device has unfinished CSI feedback tasks before resetting the model's state information, conflicts may occur between different tasks, affecting feedback performance. For example, if the network device sends an indication to trigger the reset of the model's state information on the terminal device side, but CSI feedback information corresponding to one or more reference signals has not yet been received, when the network device subsequently receives the one or more CSI feedback information, it cannot determine whether the one or more CSI feedback information was generated before or after the model's state information was reset, thus affecting feedback performance.

In view of this, the present application provides a communication method and communication device that are conducive to improving the robustness of the AI model or decoder on the channel information recovery side, such as the AI model on the network device side, for the recovery performance of channel information. The communication method can be applied to the above-mentioned communication system, such as an FDD communication scenario. In addition, the communication method can also be optionally used in a TDD communication scenario, which is not limited by the present disclosure.

It should be understood that, in this application, indication includes direct indication (also known as explicit indication) and implicit indication. Direct indication of information A refers to including information A; implicit indication of information A refers to indicating information A through the correspondence between information A and information B and the direct indication of information B. The correspondence between information A and information B can be predefined, pre-stored, pre-burned, or pre-configured.

It should be understood that, in this application, information C is used to determine information D, which includes both information D being determined solely based on information C and information D being determined based on information C and other information. Furthermore, information C can also be used to determine information D indirectly, for example, where information D is determined based on information E, and information E is determined based on information C.

In addition, in each embodiment of the present application, "network element A sends information A to network element B" can be understood as the destination end of the information A or the intermediate network element in the transmission path between the destination end and the network element B, which may include directly or indirectly sending information to network element B. "Network element B receives information A from network element A" can be understood as the source end of the information A or the intermediate network element in the transmission path between the source end and the network element A, which may include directly or indirectly receiving information from network element A. The information may be processed as necessary between the source end and the destination end of the information transmission, such as format changes, but the destination end can understand the valid information from the source end. Similar expressions in this application can be understood similarly and will not be elaborated here.

FIG7 is a schematic flow chart of a communication method provided by the present application.

The method 700 shown in FIG. 7 may be applied to a CSI feedback scenario based on time domain correlation.

In a feedback scenario based on time-domain correlation, the CSI feedback information corresponding to a reference signal is determined based on a channel measurement result relative to the reference signal and channel measurement results of historical reference signals relative to the reference signal. Historical reference signals relative to a reference signal include reference signals received before the terminal device received the reference signal. CSI recovery information corresponding to the CSI feedback information is determined based on the CSI feedback information and historical CSI feedback information relative to the CSI feedback information. Historical CSI feedback information relative to a CSI feedback information includes CSI feedback information received before the network device received the CSI feedback information.

In other words, the generated CSI feedback information for a reference signal is not only related to the current channel measurement result (i.e., the measurement result of the reference signal), but also to channel measurement results from past times. The recovered channel information for a reference signal is not only related to the currently received CSI feedback information (i.e., the CSI feedback information corresponding to the reference signal), but also to CSI feedback information received from past times.

CSI feedback based on time domain correlation can be achieved through AI-CSI dual-end compressed feedback.

Specifically, a dual-end model based on time domain correlation is deployed on two devices to implement CSI feedback. The dual-end model based on time domain correlation includes a first AI model and a second AI model. For example, the first AI model can be the AI model in the encoder, and the second AI model can be the AI model in the decoder. The first AI model can also be replaced by an encoder, and the second AI model can also be replaced by a decoder. That is, the model inference link corresponding to the encoder is performed in the second device, and the model inference link corresponding to the decoder is performed in the third device. The first AI model and the second AI model are matched.

For example, the architecture design of the two-end model can adopt any of the following: Transformer, RNN, CNN, or long short-term memory (LSTM) network. Alternatively, the two-end model can also be other self-built AI models.

In order to better describe the solution of the embodiment of the present application, the CSI feedback process based on time domain correlation is explained below using the first AI model and the second AI model as examples.

The output of the first AI model is related to the input of the first AI model and the state information of the first AI model. The output of the second AI model is related to the input of the second AI model and the state information of the second AI model.

The output of the first AI model includes CSI feedback information corresponding to the reference signal, or the CSI feedback information is based on the output of the first AI model. For example, the output of the first AI model is quantized to obtain CSI feedback information corresponding to the reference signal. For ease of description, this embodiment of the present application is primarily described using the example of the output of the first AI model including CSI feedback information, and does not limit the embodiments of this application.

The reference signal may be one or more of CSI-RS, SSB or DMRS.

Exemplarily, the reference signal may be a reference signal sent periodically, or may be a reference signal sent aperiodically.

The output of the second AI model includes CSI recovery information corresponding to the CSI feedback information.

That is, the CSI feedback information is related to the input of the first AI model and the state information of the first AI model, and the CSI recovery information corresponding to the CSI feedback information is related to the input of the second AI model and the state information of the second AI model.

The input of the first AI model includes channel measurement results.

The input of the second AI model includes CSI feedback information. Alternatively, the input of the second AI model is determined based on the CSI feedback information. For example, the CSI feedback information may be obtained by quantizing the output of the first AI model. Before inputting the CSI feedback information into the second AI model, the CSI feedback information may be dequantized. For ease of description, the embodiments of the present application are mainly described using the example of the second AI model including CSI feedback information as the input, and do not limit the embodiments of the present application.

The state information of the first AI model is determined based on historical inputs of the first AI model. Alternatively, the state information of the first AI model is initial state information of the first AI model.

For example, the current first AI model has no historical input, that is, the CSI feedback task is performed for the first time, and the first AI model can be initialized. The state information of the initialized first AI model can be the initial state information.

The state information of the first AI model may be determined based on one or more historical inputs of the first AI model.

For example, the current state information of the first AI model may be determined based on accumulated historical inputs starting from the initial state information of the first AI model.

For another example, if the state information of the first AI model has been reset, the current state information of the first AI model may be determined based on historical input accumulated since the state information of the first AI model was most recently reset.

The channel measurement result of the reference signal currently input to the first AI model is also used to update the state information of the current first AI model for the next CSI feedback task.

The state information of the second AI model is determined based on historical inputs of the second AI model. Alternatively, the state information of the second AI model is initial state information of the second AI model.

For example, the current second AI model has no historical input, that is, the CSI feedback task is performed for the first time, and the second AI model can be initialized. The state information of the initialized second AI model is the initial state information.

The state information of the second AI model may be determined based on one or more historical inputs of the second AI model.

For example, the current state information of the second AI model may be determined based on historical inputs accumulated since the initial state information of the second AI model.

For another example, if the state information of the second AI model has been reset, the current state information of the second AI model may be determined based on historical input accumulated since the state information of the second AI model was most recently reset.

The CSI feedback information currently input to the second AI model is also used to update the status information of the current second AI model for the next CSI feedback task.

The output of the first AI model is related to the current input and current state information of the first AI model.

Taking a reference signal as an example, the current input includes the channel measurement results of the reference signal. The current state information of the first AI model is determined based on the historical input of the first AI model. If the state information of the first AI model has not been reset, the current state information of the first AI model can be determined based on the channel measurement results of historical reference signals received before the reception time of the reference signal, accumulated since the initial state information of the first AI model. Alternatively, the current state information of the first AI model can also be the initial state information.

The output of the second AI model is related to the current input and current state information of the second AI model.

Taking a reference signal as an example, when a second AI model obtains CSI feedback information corresponding to the reference signal, the current input of the second AI model includes the CSI feedback information corresponding to the reference signal. Alternatively, the current input of the second AI model is determined based on the CSI feedback information corresponding to the reference signal. The output of the second AI model includes CSI recovery information corresponding to the CSI feedback information. The current state information of the second AI model is determined based on the historical input of the second AI model. If the state information of the first AI model has not been reset, the current state information of the second AI model may be determined based on historical CSI feedback information received before the moment the CSI feedback information was received, accumulated from the initial state information of the second AI model. Alternatively, the current state information of the second AI model may be the initial state information.

Taking a reference signal as an example, if the second AI model is unable to obtain CSI feedback information corresponding to the reference signal, for example, if the network device does not receive the CSI feedback information corresponding to the reference signal at the expected reception time, the current CSI feedback information reporting fails. The current input of the second AI model may include the last historical CSI feedback information received before the expected reception time of the CSI feedback information corresponding to the reference signal, that is, the last received CSI feedback information. The output of the second AI model includes CSI recovery information corresponding to the CSI feedback information that failed to be reported. The current state information of the second AI model is determined based on the historical input of the second AI model. The historical input includes historical CSI feedback information received before the expected reception time of the CSI feedback information.

Assume that three consecutive reference signals sent in chronological order are reference signal #a, reference signal #b, and reference signal #c. Reference signal #b is the current reference signal.

The channel measurement results for the current reference signal (i.e., reference signal #b) are input into the first AI model. Based on the channel measurement results for reference signal #b and the current state information of the first AI model (e.g., state information #b1), the first AI model obtains CSI feedback information corresponding to reference signal #b and updates the state information of the first AI model. The updated state information (e.g., state information #c1) is used to generate CSI feedback information corresponding to reference signal #c. State information #b1 is obtained by the first AI model updating state information #a1 based on the channel measurement results for reference signal #a. State information #a1 is the state information of the first AI model before generating CSI feedback information corresponding to reference signal #a.

When the second AI model obtains CSI feedback information corresponding to the current reference signal (i.e., reference signal #b), the CSI feedback information corresponding to reference signal #b is input into the second AI model. Based on the CSI feedback information corresponding to reference signal #b and the current state information of the second AI model (e.g., state information #b2), the second AI model recovers the channel information and updates the state information of the second AI model. The updated state information (e.g., state information #c2') can be used to recover the channel information corresponding to reference signal #c. State information #b2 is obtained by the second AI model by updating state information #a2 based on the CSI feedback information corresponding to reference signal #a. State information #a2 is the state information of the second AI model before the channel information corresponding to reference signal #2 is recovered.

If the second AI model cannot obtain the CSI feedback information corresponding to the current reference signal (i.e., reference signal #b), the CSI feedback information corresponding to reference signal #a is input into the second AI model. Based on the CSI feedback information corresponding to reference signal #a and the current state information of the second AI model (e.g., state information #b2), the second AI model recovers the channel information corresponding to reference signal #b and updates the state information of the second AI model. The updated state information (e.g., state information #c2) can be used to recover the channel information corresponding to reference signal #c. State information #b2 is obtained by the second AI model updating state information #a2 based on the CSI feedback information corresponding to reference signal #a. State information #a2 is the state information of the second AI model before the channel information corresponding to reference signal #2 is recovered.

An example of the CSI feedback process based on time domain correlation can be found in FIG5 , which will not be described in detail here.

The second device is the device on the first AI model side.

Optionally, the second device may be an AI entity, and the first AI model may be deployed on the second device.

Exemplarily, the AI entity may be an AI entity on the terminal device side, where the terminal device side includes the terminal device, or other devices that communicate with the terminal device, such as a device controlled by the terminal device or serving the terminal device. The AI entity may be the terminal device itself, or an AI entity that communicates with the terminal device. For example, the second device may be a server, such as an OTT server or a cloud server.

The first device and the third device are devices on the second AI model side.

The third device may be an AI entity, and the second AI model may be deployed on the third device.

Exemplarily, the AI entity may be an AI entity on the network device side, and the first and third devices are devices on the network device side. The network device side includes the network device, or other devices that communicate with the network device, such as devices controlled by or serving the network device. The AI entity may be the network device itself, or an AI entity that communicates with the network device. For example, the third device may be a RIC, OAM, or server, such as an OTT server or a cloud server. The near-real-time RIC is located in a RAN node, such as a CU/DU.

The first device and the third device in method 700 may be the same device or different devices.

For example, the second device is a terminal device, and the first device and the third device may be the same network device.

As shown in FIG. 7 , the method 700 may include the following steps.

710. The first device sends first indication information to the second device. The first indication information is used to determine a time unit in which reset status information of the first AI model takes effect.

The first device sends information to the second device, which may be the first device sending the information directly to the second device, or the first device sending the information to the second device through forwarding by other devices.

The second device sends information to the first device, which may be the case where the second device directly sends the information to the first device, or the second device sends the information to the second device through forwarding by other devices.

For example, the first device may be a network device, and the second device may be a terminal device. In this case, the second AI model may be deployed in the network device or in an AI entity that communicates with the network device.

For example, the first device may be an AI entity (e.g., a near real-time RIC) that communicates with the network device, and the second device may be a terminal device. The network device and the terminal device may communicate with each other. In this case, the second AI model may be deployed in the AI entity that communicates with the network device.

For example, the first device may be a network device, and the second device may be an AI entity (such as an OTT server or a cloud server) that communicates with the terminal device. In this case, the second AI model may be deployed in the network device or in the AI entity that communicates with the network device. The network device may directly send the first indication information to the AI entity that communicates with the terminal device. Alternatively, the AI entity that communicates with the terminal device may obtain the first indication information from the network device through forwarding by other devices. For example, the second device may obtain the indication information from the network device through forwarding by the terminal device.

Exemplarily, the first device may be an AI entity (such as a near real-time RIC) that communicates with a network device, and the second device may be an AI entity (such as an OTT server or a cloud server) that communicates with a terminal device. The network device and the terminal device can communicate with each other. In this case, the second AI model can be deployed in the AI entity that communicates with the network device. The AI entity that communicates with the network device can directly send the first indication information to the AI entity that communicates with the terminal device. Alternatively, the AI entity that communicates with the terminal device can obtain the first indication information from the AI entity that communicates with the network device through forwarding by other devices.

The effectiveness of the reset status information of the first AI model may also be understood as applying the reset status information of the first AI model.

In some possible implementations, the time unit in which the reset status information of the first AI model takes effect can also be understood as the time unit in which the reset operation of the status information of the first AI model ends, or the time unit in which the reset of the status information of the first AI model is completed.

A time unit may also be referred to as a moment. For example, the time unit during which the reset state information of the first AI model takes effect may also be referred to as the moment at which the reset state information of the first AI model takes effect.

Exemplarily, the time unit may include any one of the following: slot, subframe, symbol or TTI, etc.

Alternatively, the time unit may also be a non-air interface time unit, such as a time unit in wired transmission.

For example, the time unit may include any of the following: nanoseconds, microseconds, milliseconds, etc.

The specific description of the AI model can be found in the previous article and will not be repeated here.

The status information of the AI model may also be referred to as any one or more of the following: cache information related to the AI model, storage information related to the AI model, intermediate information (for example, intermediate information generated by the AI model), internal information (for example, internal information of the device on which the AI model is deployed or internal information of the AI model), and parameter information generated or updated by the AI model.

The first AI model can be used to generate CSI feedback information. A second AI model that matches the first AI model can be used to recover channel information corresponding to the CSI feedback information. The specific generation and recovery processes can be referred to the descriptions of the encoder and decoder above, respectively, and will not be repeated here.

CSI feedback information may also be referred to as CSI feedback bits. CSI feedback information is indicated by a CSI report. For ease of description, in the embodiments of this application, CSI feedback information may also be referred to as feedback information. One piece of feedback information corresponds to one CSI report.

The second device can determine the time unit in which the reset status information of the first AI model takes effect according to the first indication information, and process the CSI feedback task based on this, such as determining whether the output CSI feedback information is based on the reset status information or the status information before the reset.

For example, the second device may apply the reset state information of the first AI model in the time unit or a time unit after the time unit to generate CSI feedback information, and report the CSI feedback information.

This facilitates enabling the third device to decode CSI feedback information using the state information of the second AI model that matches the state information of the first AI model, thereby ensuring feedback performance and further improving the decoding performance of the second AI model, i.e., the decoder. Specifically, if the CSI feedback information is generated while the reset state information of the first AI model is in effect, the third device can use the reset state information of the second AI model to restore the channel information corresponding to the CSI feedback information. If the CSI feedback information is generated while the reset state information of the first AI model is not in effect, the third device can use the pre-reset state information of the second AI model to restore the channel information corresponding to the CSI feedback information.

Exemplarily, the third device and the first device may be the same device. The first device sends the first indication information to the second device, so that the second device can determine the time unit when the reset state information of the first AI model takes effect based on the indication of the first device, thereby aligning the time when the state information of the first AI model takes effect. This facilitates the first device to determine whether the CSI feedback information was generated when the reset state information of the first AI model took effect, and then use the state information of the second AI model that matches the state information of the first AI model to decode the CSI feedback information, thereby improving the performance of the second AI model in recovering channel information.

Alternatively, if the third device and the first device are different devices, the first device sends the first indication information to the second device, so that the second device can determine the time unit when the reset state information of the first AI model takes effect based on the indication of the first device, thereby aligning the time when the state information of the first AI model takes effect between the first device and the second device. The first device can send the indication information to the third device. For example, the first device can indicate the time unit when the reset state information of the first AI model takes effect to the third device, or the first device can indicate the third device to apply the reset state information of the second AI model or the state information of the second AI model before reset to restore the channel information corresponding to the CSI feedback information, or the first device can indicate the time unit when the reset state information of the second AI model takes effect to the third device. This facilitates enabling the third device to use the state information of the second AI model that matches the state information of the first AI model to decode the CSI feedback information, thereby improving the performance of the second AI model in recovering channel information.

In a possible implementation, method 700 may be applied to a scenario where CSI feedback information reporting fails. CSI feedback information reporting failure means that the CSI feedback information packet is lost and the network device fails to obtain the CSI feedback information sent by the terminal device.

That is, in a scenario where the CSI feedback information fails to be reported, the first device can send a first indication message to the second device so that the second device can determine the time unit in which the reset status information of the first AI model takes effect, and complete the reset based on this.

For example, after the number of times the CSI feedback information is lost reaches a set threshold, the first device may send first indication information to the second device.

The reset state information of the first AI model and the reset state information of the second AI model match.

For example, the reset state information of the first AI model may be any state information of the first AI model in the CSI feedback process, and the reset state information of the second AI model may be state information of the second AI model that matches the reset state information of the first AI model in the CSI feedback process. In other words, the reset state information of the second AI model may be any state information of the second AI model in the CSI feedback process, and the reset state information of the first AI model may be state information of the first AI model that matches the reset state information of the second AI model in the CSI feedback process.

The reset state information of the first AI model and the reset state information of the second AI model may both be their respective initial state information. Alternatively, the reset state information of the first AI model and the reset state information of the second AI model may be based on corresponding historical information.

Exemplarily, the reset state information of the first AI model is based on the channel measurement result of the historical reference signal, and the reset state information of the second AI model is based on the CSI feedback information corresponding to the historical reference signal. The historical reference signal may be a historical reference signal relative to reference signal #A.

The historical reference signal may be one or more. The reset state information of the first AI model may be based on the channel measurement results of multiple historical reference signals accumulated since the initial state information of the first AI model. For example, the reset state information of the first AI model may be based on the channel measurement results of all historical reference signals sent before reference signal #A accumulated since the initial state information of the first AI model. The reset state information of the second AI model may be based on the CSI feedback results corresponding to multiple historical reference signals accumulated since the initial state information of the second AI model. For example, the reset state information of the second AI model may be based on the channel measurement results corresponding to all historical reference signals sent before reference signal #A accumulated since the initial state information of the second AI model.

Taking Figure 5 as an example, the reset state information of the first AI model may be e ₁ , and the reset state information of the second AI model may be d ₁ . For another example, the reset state information of the first AI model may be e ₂ , and the reset state information of the second AI model may be d ₂ . For another example, the reset state information of the first AI model may be e ₀ , and the reset state information of the second AI model may be d ₀ .

Exemplarily, method 700 can be applied to a scenario where CSI feedback information reporting fails. In this case, the reset state information of the first AI model may be the state information of the first AI model that occurred before the CSI feedback information reporting failed, and the reset state information of the second AI model may be the state information of the second AI model that occurred before the CSI feedback information reporting. Taking Figure 6 as an example, the reset state information of the first AI model may be any one of e ₀ , e ₁ , or e ₂ , and the reset state information of the second AI model may be any one of d ₀ , d ₁ , or d ₂ that matches the reset state information of the first AI model. Figure 6 only uses the example of the reset state information of the first AI model being e ₀ and the reset state information of the second AI model being d ₀ for illustration, and does not limit the solution of the embodiment of the present application.

That is, when the CSI feedback information reporting fails, the state information of the first AI model can be reset to the state information before the CSI feedback information reporting failure occurs, and the state information of the second AI model can be reset to the state information before the CSI feedback information reporting failure occurs.

For a detailed description of the reset process in the scenario where the CSI feedback information reporting fails, please refer to the description in FIG6 , which will not be described again here to avoid repetition.

The reset status information of the first AI model and the reset status information of the second AI model may be determined in various ways.

The reset status information of the first AI model may be predefined, preconfigured, determined by the second device, or may be indicated by other devices, for example, by the first device.

For example, the second device receives indication information, where the indication information indicates an identifier of status information of the first AI model.

The second device may determine, according to the indication information, the state information of the first AI model corresponding to the identifier as reset state information of the first AI model.

For another example, the second device receives indication information, where the indication information indicates an identifier of status information of the second AI model.

The second device may determine, according to the indication information, the state information of the first AI model that matches the state information of the second AI model corresponding to the identifier as reset state information of the first AI model.

For another example, the second device receives indication information, which indicates the moment of status information of the first AI model.

The second device may determine, according to the indication information, the state information of the first AI model at the moment as reset state information of the first AI model.

Exemplarily, the reset status information of the second AI model may be predefined, preconfigured, determined by a third device, or may be indicated by other devices, for example, the second device.

For example, the third device receives indication information, where the indication information may indicate any one of the following: an identifier of the state information of the second AI model, an identifier of the state information of the first AI model, or a time of the state information of the second AI model.

The method for determining the reset status information of the second AI model can refer to the method for determining the reset status information of the first AI model, and will not be repeated here.

The embodiment of the present application does not limit the method for determining the reset status information of the first AI model and the reset status information of the second AI model, as long as the two match.

In the embodiments of the present application, resetting the state information of the first AI model helps ensure consistency between the state information of the first AI model and the state information of the second AI model. For example, resetting the state information of the first AI model and the state information of the second AI model to matching state information helps ensure the accuracy of recovered channel information and improves the robustness of the second AI model's channel information recovery performance. For example, in the event of CSI feedback packet loss, resetting the state information of the first AI model and the state information of the second AI model can restore consistency between the state information of the first AI model and the state information of the second AI model, thereby improving the accuracy of the recovered channel information.

At the same time, in the solution of the embodiment of the present application, a first indication information can be sent to the first AI model side (i.e., the second device) so that the first AI model side can determine the time unit in which the reset status information of the first AI model takes effect. This is conducive to the consistency of the effective time of the reset status information of the first AI model and the reset status information of the second AI model, thereby further improving the performance of the second AI model in recovering channel information. For example, before instructing the first AI model to reset its status information, there may still be CSI feedback tasks that have not been completed. There may be a conflict between the unfinished CSI feedback tasks and the reset of the status information, making it difficult for the second AI model side (i.e., the third device) to determine whether the received CSI feedback information is obtained based on the reset status information of the first AI model. This may cause the status information of the second AI model and the first AI model to be unable to be reset synchronously, thereby affecting the accuracy of the recovered channel information. By adopting the solution of the embodiment of the present application, the second device can determine the time unit in which the reset status information of the first AI model takes effect based on the first indication information, so that the first AI model side and the second AI model side can align the time when the status information takes effect, which is beneficial for the second AI model side to determine whether the CSI feedback information is generated when the reset status information takes effect. Then, the channel information can be restored based on the corresponding status information, which is beneficial to ensuring the feedback performance of the AI space-frequency-time codebook when it is long.

Optionally, method 700 may further include step 720 .

720. The second device sends first CSI feedback information to the first device.

Optionally, step 720 may include: the second device sending first CSI feedback information to the first device in a first time unit, the first CSI feedback information being related to reset status information of the first AI model, and the first time unit being no earlier than a time unit in which the reset status information of the first AI model takes effect.

For example, the second device may be a terminal device, the first device may be a network device, and the terminal device may send first CSI feedback information to the network device in a first time unit.

Exemplarily, the second device sending the first CSI feedback information to the first device may be sending the first CSI feedback information to the first device by forwarding the information to the first device through another device.

For example, the second device may be an entity communicating with the terminal device, the first device may be a network device, the terminal device may obtain the first CSI feedback information from the second device, and send the first CSI feedback information to the network device in a first time unit.

The first time unit is no earlier than the time unit in which the reset status information of the first AI model takes effect, which may include: the first time unit is later than the time unit in which the reset status information of the first AI model takes effect, and/or the first time unit includes the time unit in which the reset status information of the first AI model takes effect.

In the embodiment of the present application, A is no earlier than B, which can also be referred to as B is no later than A.

In the embodiment of the present application, A is later than B, which can also be referred to as A being after B, B being earlier than A, or B being before A.

After receiving the first indication information, the second device may determine the time unit in which the reset encoder state information takes effect. The time at which the first CSI feedback information is sent, i.e., the first time unit, must not be earlier than the time unit in which the reset state information of the first AI model takes effect. When the state information of the first AI model takes effect, the second device may apply the reset encoder state information to generate the first CSI feedback information and send the first CSI feedback information in the first time unit.

The input of the first AI model includes a channel measurement result of a first reference signal. The first reference signal corresponds to first CSI feedback information. The first CSI feedback information is the output of the first AI model or is based on the output of the first AI model. For example, the output of the first AI model is quantized to obtain the first CSI feedback information.

The output of the first AI model is related to the measurement result of the first reference signal and the reset state information of the first AI model. The reset state information of the first AI model may be initial state information, or the reset state information of the first AI model may be based on the channel measurement result of a historical reference signal. The historical reference signal may be a historical reference signal relative to reference signal #A. The transmission time of reference signal #A is earlier than the transmission time of the first reference signal.

Specifically, the channel measurement result of the first reference signal is input into the first AI model. The first AI model may output first CSI feedback information based on the channel measurement result of the first reference signal and reset state information of the first AI model. Alternatively, the output of the first AI model is processed to obtain the first CSI feedback information.

Accordingly, the CSI recovery information corresponding to the first CSI feedback information may be related to the reset state information of the second AI model. The second AI model side may recover the channel information corresponding to the first CSI feedback information based on the reset state information of the second AI model.

The input of the second AI model includes the first CSI feedback information. Alternatively, the input of the second AI model is based on the first CSI feedback information. For example, the input of the second AI model includes the result of dequantization processing of the first CSI feedback information.

The output of the second AI model includes CSI recovery information corresponding to the first CSI feedback information. The output of the second AI model is related to the first CSI feedback information and reset state information of the second AI model. The reset state information of the second AI model may be initial state information, or the reset state information of the second AI model may be based on CSI feedback information corresponding to a historical reference signal. The historical reference signal may be a historical reference signal relative to reference signal #A. Reference signal #A is transmitted earlier than the first reference signal.

Specifically, the first CSI feedback information is input into the second AI model. Alternatively, the first CSI feedback information is processed and then input into the second AI model. The second AI model may output CSI recovery information corresponding to the first CSI feedback information based on the first CSI feedback information and reset status information of the second AI model.

The description of CSI feedback based on the dual-end model can be found in the previous article and will not be repeated here.

In the solution of the embodiment of the present application, a first indication information can be sent to the first AI model side so that the first AI model side (i.e., the second device) can determine that the time unit in which the reset state information of the first AI model takes effect is no later than the sending time of the first CSI feedback information (i.e., the first time unit). The first AI model side can apply the reset state information of the first AI model to generate the first CSI feedback information before the first CSI feedback information is sent. In this way, the second AI model side, such as the network device side, can determine that the first CSI feedback information is generated when the reset state information takes effect, and then can restore the channel information corresponding to the first CSI feedback information based on the reset state information of the second AI model, which is conducive to ensuring the accuracy of the restored channel information, thereby improving the performance of the second AI model in restoring the channel information.

Furthermore, the time interval between the first time unit and the time unit in which the reset status information of the first AI model takes effect is greater than or equal to the processing time of the first AI model.

The processing time of the first AI model may also be replaced by the processing time of the terminal device, the CSI processing time, or the CSI calculation time, etc. For example, the processing time of the first AI model may include the time required to generate CSI feedback information through processing of the first AI model.

This helps ensure that the second device has sufficient time to generate the first CSI feedback information based on the reset status information of the first AI model.

The second device can determine the time unit in which the reset status information of the first AI model takes effect in various ways, which are explained below.

In a first possible implementation, the first indication information is used to instruct to send uplink information on a first time resource. The time unit in which the reset state information of the first AI model takes effect is no later than the start time of the first time resource.

After receiving the first indication information, the second device may determine to send uplink information on the first time resource, and make the reset status information of the first AI model effective before sending the uplink information.

A time resource may include one or more time units. The starting time of a time resource may be understood as the starting time of the starting time unit in the time resource.

The time unit in which the reset status information of the first AI model takes effect is no later than the start time of the first time resource, which may include that the time unit in which the reset status information of the first AI model takes effect is earlier than the start time of the first time resource, and/or the time unit in which the reset status information of the first AI model takes effect is the start time of the first time resource.

Optionally, the first time resource may include a first time unit. In other words, the first time resource may include a resource for transmitting the first CSI feedback information, that is, the uplink information sent on the first time resource may include the first CSI feedback information.

For example, the first indication information may be used to indicate that the first CSI feedback information is sent in a first time unit.

After receiving the first indication information, the second device may determine to send the first CSI feedback information at the first time unit, and enable the reset state information of the first AI model before sending the first CSI. In this way, the first CSI feedback information can be generated based on the reset state information of the first AI model.

The uplink information sent on the first time resource may also include other uplink information besides the first CSI feedback information. The embodiment of the present application does not limit the content of the uplink information.

Optionally, the method 700 may further include: the second device determines a time period #1 (an example of the first time period), wherein the end time of the time period #1 is not later than the start time of the first time resource.

The end time of period #1 is no later than the start time of the first time resource, which may include: the end time of period #1 is the start time of the first time resource.

In this case, the second device can determine time period #1 and, based on time period #1, determine the first time resource, i.e., the end time of time period #1 must be no later than the start time of the first time resource. Furthermore, based on the first time resource, the second device can determine the time unit in which the reset state information of the first AI model takes effect, i.e., the time unit in which the reset state information of the first AI model takes effect must be no later than the start time of the first time resource.

The time unit in which the reset status information of the first AI model takes effect is no later than the start time of the first time resource may include: the time unit in which the reset status information of the first AI model takes effect is the start time of the first time resource.

Furthermore, the second device may determine the time period #1 according to any two of the following: the start time of the time period #1, the length of the time period #1, or the end time of the time period #1.

The length of period #1 may also be referred to as a first duration or a first offset.

Exemplarily, the length of period #1 may be predefined.

Alternatively, the length of period #1 may be indicated by the first indication information.

Alternatively, the length of period #1 may be indicated by other indication information (eg, fourth indication information) other than the first indication information. That is, the first device may send fourth indication information to the second device, where the fourth indication information indicates the length of period #1.

Optionally, the first indication information may indicate a first time resource.

For example, the first indication information may be scheduling information of the first CSI feedback information, where the scheduling information indicates a time resource of the first CSI feedback information. The time resource of the first CSI feedback information may indicate a sending time of the first CSI feedback information, that is, a first time unit.

Time resources can be replaced by time domain resources.

In this case, after receiving the first indication information, the second device can determine the first time resource, and determine the time unit in which the reset status information of the first AI model takes effect based on the first time resource, that is, the time unit in which the reset status information of the first AI model takes effect is no later than the start time of the first time resource.

In a second possible implementation, time period #2 (another example of the first time period) can be used to determine the time unit in which the reset status information of the first AI model takes effect.

The second device may determine a time unit in which the reset status information of the first AI model takes effect according to period #2.

Exemplarily, period #2 may include a time unit in which the reset state information of the first AI model takes effect.

In other words, the time unit during which the reset status information of the first AI model takes effect is within time period #2.

For example, the time unit in which the reset status information of the first AI model takes effect may be the end time of period #2. The second device may enable the reset status information of the first AI model to take effect at the end time of period #2. For example, the second device may complete the reset of the status information of the first AI model at the end time of period #2.

For another example, the time unit in which the reset state information of the first AI model takes effect may be earlier than the end time of period #2. The second device may complete the reset of the state information of the first AI model before the end time of period #2.

For another example, the time unit in which the reset status information of the first AI model takes effect is at the latest the end time of period #2. The second device can complete the reset of the status information of the first AI model at the latest the end time of period #2. In this case, the end time of period #2 can be understood as the deadline for completing the reset of the status information of AI model #1.

Alternatively, the time unit during which the reset state information of the first AI model takes effect may be no later than the start time of period #2.

Alternatively, the time unit during which the reset state information of the first AI model takes effect may be no earlier than the end time of period #2.

In the embodiment of the present application, the time unit in which the reset status information of the first AI model takes effect is within time period #2 as an example for explanation, which does not constitute a limitation on the solution of the embodiment of the present application.

The second device may determine the time period #2 according to any two of the following: the start time of the time period #2, the duration of the time period #2, or the end time of the time period #2.

Optionally, the method 700 may further include: obtaining a length of period #2. The length of period #2 is used to determine a time unit during which the reset state information of the first AI model takes effect.

The length of period #2 may also be referred to as a second duration or a second offset.

Exemplarily, the length of period #2 may be predefined.

Alternatively, the length of period #2 may be indicated by the first indication information.

Alternatively, the length of time period #2 may be indicated by other indication information (eg, fourth indication information) other than the first indication information. That is, the first device may send fourth indication information to the second device, where the fourth indication information indicates the length of time period #2.

Exemplarily, the starting time of time period #2 can be any one of the following: the sending time of the first indication information, the receiving time of the first indication information, the time indicated by the first indication information, the sending time of other indication information, the receiving time of other indication information, or the time indicated by other indication information.

For example, the first indication information may indicate time B, and time B may be used as the starting time of time period #2.

For another example, the first indication information may indicate uplink information B, and the sending time of uplink information B may be used as the starting time of period #2. For example, uplink information B may be CSI feedback information B, and the sending time of CSI feedback information B may be used as the starting time of period #2.

For another example, the scheduling information of the first CSI feedback information may be the first indication information or other indication information. The sending time or receiving time of the scheduling information of the first CSI feedback information may be used as the starting time of period #2.

Exemplarily, the end time of period #2 may be the time indicated by the first indication information or other indication information.

Furthermore, the method 700 may further include: sending uplink information on the first time resource. The time unit in which the reset state information of the first AI model takes effect is no later than the start time of the first time resource.

Sending the uplink information on the first time resource may be instructed by the first indication information or by other indication information.

In this case, the second device may determine time period #2, determine a time unit in which the status information of the first AI model is effective according to time period #2, and determine the first time resource according to the time unit in which the status information of the first AI model is effective.

In a third possible implementation manner, the first indication information may indicate a time unit in which the reset status information of the first AI model takes effect.

Exemplarily, the first indication information may indicate an identifier of a time unit in which the reset status information of the first AI model takes effect.

The following describes how to instruct the first AI model to reset its status information.

The first device can send instruction information to the second device to instruct the second device to reset the status information of the first AI model.

In a first possible implementation manner, the first indication information may indicate resetting of status information of the first AI model.

The first indication information may instruct the second device to reset the state information of the first AI model. The second device may determine, based on the first indication information, a time unit in which the reset state information of the first AI model takes effect.

For example, the time unit during which the reset state information of the first AI model takes effect is related to period #2. For example, the length of period #2 is predefined, and the start time of period #2 is the time when the first indication information is sent. After receiving the first indication information, the second device can determine period #2 and, thereby, the time unit during which the reset state information of the first AI model takes effect.

In a second possible implementation, method 700 may further include: obtaining second indication information, where the second indication information indicates resetting the status information of the first AI model.

The second indication information may come from the first device.

The first indication information and the second indication information are different indication information. The first indication information and the second indication information can be carried in the same signaling or in different signaling.

For example, the time unit in which the reset status information of the first AI model takes effect is related to time period #2. For example, the time when the second indication information is sent or received is the start time of time period #2, and the length of time period #2 may be indicated by the first indication information. After receiving the first indication information and the second indication information, the second device may determine time period #2 based on the length and start time of time period #2, and further determine the time unit in which the reset status information of the first AI model takes effect.

For example, the time unit in which the reset status information of the first AI model takes effect is related to the first time resource. For example, the second indication information is sent or received at the start time of time period #1, and the length of time period #1 is predefined. After receiving the first indication information and the second indication information, the second device can determine time period #1 based on the length and start time of time period #1, determine the first time resource based on time period #1, and then determine the time unit in which the reset status information of the first AI model takes effect based on the first time resource. In other words, the reset status information of the first AI model takes effect no later than the first time resource.

As described above, the first indication information used to determine the time unit in which the status information of the first AI model takes effect may include the following forms: the first indication information may indicate time information related to the determination of the time unit in which the status information of the first AI model takes effect (such as the length of time period #1 or the length of time period #2, etc.), the sending time or receiving time of the first indication information may be used as time information related to the determination of the time unit in which the status information of the first AI model takes effect (such as the starting time of time period #1 or the starting time of time period #2, etc.), or, the first indication information may also be used to trigger the determination of the time unit in which the status information of the first AI model takes effect.

Resetting the AI model's state information requires a certain amount of processing time. During this time, both the network device and the terminal device do not want to use the AI model for inference to avoid degradation in the quality of the obtained CSI feedback information. The following describes a solution for ensuring the quality of CSI feedback information when the AI model's state information is reset.

In one possible implementation, method 700 may further include: the second device obtaining scheduling information for second CSI feedback information. The time resource of the second CSI feedback information indicated by the scheduling information overlaps with time period #3 (an example of the first time period). The second device ignores or skips sending the second CSI feedback information. The second CSI feedback information is related to the first AI model.

Exemplarily, the first device may be a network device, and the second device may be a terminal device. The network device may send scheduling information of the second CSI feedback information to the terminal device.

Exemplarily, the first device may be a network device, and the second device may be an AI entity serving the terminal device. For example, the terminal device may receive scheduling information of the second CSI feedback information from the network device, and the AI entity may obtain the scheduling information of the second CSI feedback information from the terminal device.

Exemplarily, the scheduling information of the second CSI feedback information may be sent before the first indication information is sent.

The time resource of the CSI feedback information indicated by the scheduling information can also be replaced by the time resource used for CSI feedback information transmission, the sending time of the CSI feedback information indicated by the scheduling information, the time when the CSI feedback information is scheduled to be sent, the time when the uplink (UL) grant is scheduled, or the time unit for sending CSI feedback information, etc.

The second CSI feedback information may be the output of the first AI model, or may be based on the output of the first AI model. For example, the output of the first AI model is quantized to obtain the second CSI feedback information.

Specifically, the channel measurement result of the second reference signal is input into the first AI model for processing, and the first AI model outputs the second CSI feedback information, or the second CSI feedback information is determined based on the output of the first AI model.

For a specific description of generating the second CSI feedback information by using the first AI model, please refer to the CSI feedback information generation process in the previous article, which will not be repeated here.

Period #3 may include a time unit in which the reset state information of the first AI model is effective.

For example, period #3 may be period #1.

For another example, period #2 may include a time unit in which the reset status information of the first AI model takes effect. In this case, period #3 may be period #2.

Time period #3 includes the time period during which the reset state information of the first AI model takes effect. That is, the reset state information of the first AI model takes effect during time period #3. If the time resource for the second CSI feedback information indicated by the scheduling information overlaps with time period #3, the second CSI feedback information may be generated with the reset state information of the first AI model taking effect, or it may be generated before the reset state information of the first AI model takes effect. If the second AI model receives the second CSI feedback information, it cannot determine whether the second CSI feedback information was generated with the reset state information of the first AI model taking effect. Therefore, it cannot determine whether to use the reset state information of the second AI model to restore the channel information corresponding to the second CSI feedback information, which may affect the accuracy of the restored channel information.

In the solution of the embodiment of the present application, if the transmission time of CSI feedback information (such as the second CSI feedback information) indicated by the scheduling information overlaps with time period #3, the transmission of this CSI feedback information is ignored or skipped. This prevents the second AI model from being unable to determine whether the second CSI feedback information was generated when the reset state information of the first AI model was in effect, thereby preventing the second AI model from recovering channel information based on mismatched state information, which helps ensure the accuracy of the recovered channel information.

In another possible implementation, method 700 may further include: the first device sending scheduling information for third CSI feedback information, where the time resource for the third CSI feedback information indicated by the scheduling information does not include time period #3; and the first device receiving the third CSI feedback information. The third CSI feedback information is related to the first AI model.

Exemplarily, the first device may be a network device, and the second device may be a terminal device. The network device may send scheduling information of the third CSI feedback information to the terminal device.

The third CSI feedback information may be the output of the first AI model, or may be based on the output of the first AI model. For example, the output of the first AI model is quantized to obtain the third CSI feedback information.

Specifically, the channel measurement result of the third reference signal is input into the first AI model for processing, and the first AI model outputs the third CSI feedback information, or the third CSI feedback information is determined based on the output of the first AI model.

For a specific description of generating the third CSI feedback information by using the first AI model, reference may be made to the CSI feedback information generation process described above, which will not be repeated here.

The description of period #3 can be found in the previous article and will not be repeated here.

In the embodiments of the present application, the transmission time of CSI feedback information (such as the third CSI feedback information) indicated by the scheduling information does not include time period #3. This prevents the second AI model from being unable to determine whether the third CSI feedback information was generated while the reset state information of the first AI model was in effect, thereby preventing the second AI model from recovering channel information based on mismatched state information, which helps ensure the accuracy of the recovered channel information. For example, if the time resource of the third CSI feedback information indicated by the scheduling information is before time period #3, the reset state information of the first AI model has not yet taken effect. If the third CSI feedback information was generated while the reset state information of the first AI model was not in effect, the second AI model can determine that the third CSI feedback information was generated while the reset state information of the first AI model was not in effect, and can then recover the channel information, which helps ensure the accuracy of the recovered channel information.

The following describes the relationship between resetting the status information of an AI model and taking effect on the reset status information.

In one possible implementation, the time interval between the start time and the target time for resetting the state information of the first AI model may be greater than or equal to the duration required to reset the state information of the first AI model. The target time is the time unit in which the reset state information of the first AI model, determined based on the first indication information, takes effect.

The starting time for resetting the state information of the first AI model is the time when resetting the state information of the first AI model begins.

For example, the second device may determine the time to start resetting the state information of the first AI model according to the target time and the duration required for resetting the state information of the first AI model.

This can ensure that the second device has enough time to complete the reset of the status information of the first AI model, so that the reset status information of the first AI model can take effect at the target time.

The time required to reset the status information of the first AI model can be determined in various ways.

Optionally, the time required to reset the state information of the first AI model may be predefined.

Optionally, method 700 may further include: the second device sending third indication information to the first device or other devices other than the first device, where the third indication information indicates the time required to reset the status information of the first AI model.

Exemplarily, the first device may be a network device, and the second device may be a terminal device. The terminal device may send third indication information to the network device, informing the network device of the time required to reset the state information of the first AI model. Alternatively, the terminal device may send third indication information to the core network device, informing the core network device of the time required to reset the state information of the first AI model. The network device may obtain the time required to reset the state information of the first AI model from the core network device.

Optionally, the third indication information is carried in the signaling sent by the terminal device that carries the terminal device capabilities.

That is, the duration required to reset the state information of the first AI model is reported in the terminal device capability report. For example, the first device may be a network device, and the second device may be a terminal device. The terminal device may report its capabilities to the network device. Alternatively, the terminal device may report its capabilities to the core network device. The network device may obtain the terminal device capabilities from the core network device.

Terminal device capabilities may also be referred to as UE capabilities.

Optionally, the third indication information is carried in signaling that carries configuration information of the first AI model. For example, the first device may be a network device, and the second device may be a terminal device. During the process of model recognition and model matching between the terminal device and the network device, the duration required to reset the state information of the first AI model may be reported in the model configuration information.

For example, the second device may report the configuration information through a two-end offline interaction method, where the two ends are the sending end and the receiving end of the third indication information.

For another example, the second device may report the configuration information through high-layer signaling interaction such as RRC signaling.

The configuration information of the model may include one or more of the processing delay of the model, application function, structural information of the model, parameter information of the model, etc. Optionally, the first indication information is related to the time required to reset the state information of the first AI model.

The first indication information may be determined based on the time required to reset the state information of the first AI model.

Exemplarily, when period #2 includes a time unit in which the reset state information of the first AI model takes effect, the length of period #2 may be greater than or equal to the duration required for resetting the state information of the first AI model.

Taking the example where the length of period #2 is indicated by the first indication information, for example, the first device may determine the length of period #2 based on the duration required to reset the status information of the first AI model, and indicate the length of period #2 to the second device.

Furthermore, the second device may determine a starting time for resetting the status information of the first AI model based on the end time of period #2 and the duration required to reset the first AI model. The time interval between the starting time for resetting the status information of the first AI model and the end time of period #2 may be greater than or equal to the duration required to reset the first AI model.

This helps ensure that the second device has sufficient time to complete the reset of the status information of the first AI model, so that the reset status information of the first AI model takes effect.

The first device may indicate the resetting and/or validation of the model status information of the second device in a variety of indication ways, that is, the first indication information may be indicated in a variety of indication ways.

FIG18 is a schematic diagram showing four types of indication signaling of the first indication information.

Optionally, the first indication information may be carried in the first DCI and/or in higher-layer signaling. The higher-layer signaling may include one or more of radio resource control (RRC) signaling or medium access control element (MAC-CE) signaling.

Higher-layer signaling may also be referred to as higher-layer configuration signaling.

For example, as shown in FIG18( a ), the first indication information may indicate the length of period #2. The length of period #2 may be indicated by higher-layer signaling. The higher-layer signaling may also be used to indicate other information (others).

Optionally, the first DCI is also used to trigger first CSI feedback information, and the first CSI feedback information belongs to a non-periodic CSI report.

For example, as shown in FIG18( b ), the first DCI may use DCI format A. In this DCI format, the first indication information may indicate the length of period #2. The length of period #2 may be indicated by the DCI that triggers an aperiodic CSI report (A-CSI report). The A-CSI report includes the first CSI feedback information.

Optionally, the first DCI is also used to schedule or configure an uplink shared channel (UL-SCH), which does not include CSI feedback information.

For example, as shown in (c) of Figure 18, the first DCI may adopt DCI format B, in which the first indication information may indicate the length of period #2. The length of period #2 may be indicated by the DCI used to schedule or configure the UL-SCH. The UL-SCH does not include CSI feedback information. In this case, the first DCI is not used to trigger A-CSI. The first DCI may be a UL grant, which is used to schedule or configure the UL-SCH without triggering A-CSI (UL-SCH without indicating A-CSI).

Optionally, the first DCI is also used to carry first indication information of other terminal devices.

That is, the first DCI can carry the first indication information of multiple terminal devices.

Exemplarily, the first DCI may not be used for scheduling or configuring UL-SCH and is not used to trigger A-CSI. The first DCI may be used only to carry first indication information of one or more terminal devices. The first DCI may be a UL grant that is not used for scheduling or configuring UL-SCH and is not used to trigger A-CSI (UL grant without UL-SCH without A-CSI).

For example, as shown in (d) of Figure 18 , the first DCI may use DCI format C, in which the first indication information may indicate the length of period #2. The length of period #2 may be indicated by a DCI that is not used for scheduling or configuring the UL-SCH and is not used to trigger A-CSI.

Method 700 of an embodiment of the present application can be applied to a CSI feedback process between multiple terminal devices and a network device. For example, the first device can be a network device, the second device can be a terminal device, and the network device can send first indication information to multiple terminal devices. The content indicated by the first indication information can be the same or different for different terminal devices. For example, the first indication information can indicate the length of period #2, and the length of period #2 can be the same or different for different terminal devices.

The network device may send a first DCI to the multiple terminal devices, where the first DCI may carry first indication information of the multiple terminal devices. The multiple terminal devices may respectively obtain their respective first indication information from the first DCI.

Figure 8 shows a schematic flow chart of another communication method according to an embodiment of the present application. The method 800 shown in Figure 8 can be regarded as a specific implementation of the method 700 shown in Figure 7. For related descriptions, reference can be made to method 700. To avoid repetition, some descriptions of method 800 are omitted as appropriate.

As shown in FIG. 8 , method 800 may include the following steps.

810. The first device sends first indication information to the second device. The first indication information is used to determine a time unit in which reset status information of the first AI model takes effect.

820. The second device determines a time unit in which the reset status information of the first AI model takes effect.

The second device may determine, based on the first indication information, a time unit in which the reset status information of the first AI model takes effect.

830. The second device resets the status information of the first AI model.

The second device may reset the state information of the first AI model according to the time unit determined in step 820 .

840. The second device generates first CSI feedback information.

The second device may generate first CSI feedback information by using the reset state information of the first AI model.

850. The second device sends first CSI feedback information to the first device.

For example, the second device may send the first CSI feedback information to the first device in a first time unit. The time unit in which the reset state information of the first AI model takes effect is no later than the first time unit.

It should be understood that the step numbers of the method in the embodiments of the present application are only for convenience of description and do not limit the order in which the steps are executed.

As mentioned above, in different CSI feedback scenarios, the first device and the third device may be different devices. The following exemplifies the application of the method of the embodiment of the present application in different CSI feedback scenarios with reference to FIG9 to FIG13 .

Figure 9 shows a schematic flow chart of another communication method according to an embodiment of the present application. The method 900 shown in Figure 9 can be regarded as a specific implementation of the method shown in Figure 7 or Figure 8. For related descriptions, please refer to the previous text. To avoid repetition, some descriptions are appropriately omitted when describing method 900.

As shown in Figure 9, in method 900, the first device is a network device, the second device is a terminal device, the encoder (an example of the first AI model) is deployed on the terminal device, and the decoder (an example of the second AI model) is deployed on the network device. Method 900 may include the following steps.

910. The network device sends a first reference signal to the terminal device.

920. The network device sends first indication information to the terminal device. The first indication information is used to determine the time unit in which the reset status information of the first AI model takes effect.

930. The terminal device determines, based on the first indication information, a time unit in which the reset status information of the encoder takes effect.

For example, the first indication information indicates the length of period #2, and the start time of period #2 may be the time when the first indication information is sent. The terminal device may determine period #2 based on the time when the first indication information is sent and the length of period #2. The time unit in which the reset encoder status information takes effect falls within period #2.

940. The terminal device resets the status information of the encoder.

The terminal device can complete the reset of the encoder status information within time period #2.

950. The terminal device performs measurement of the first reference signal to obtain channel information.

960. The terminal device generates first CSI feedback information according to the status information of the reset encoder.

Channel information is input into the encoder for processing to obtain first CSI feedback information corresponding to the channel information. The first CSI feedback information is feedback information about the channel information. At this point, the encoder state information has been reset. The first CSI feedback information is CSI feedback information generated when the reset encoder state information is in effect.

970. The terminal device sends first CSI feedback information to the network device.

The terminal device sends first CSI feedback information to the network device in uplink control information (UCI) at a first time unit. The first time unit is no earlier than the time unit determined in step 930.

980. The network device restores the channel information corresponding to the first CSI feedback information according to the reset decoder state information.

The network device knows that the first CSI feedback information is generated when the reset state information of the encoder is effective, and can restore the channel information corresponding to the first CSI feedback information according to the reset state information of the decoder.

The first CSI feedback information is input into the decoder for processing, or the processed first CSI feedback information is input into the decoder for processing to obtain CSI recovery information corresponding to the first CSI feedback information. At this time, the state information of the decoder has been reset. The CSI recovery information corresponding to the first CSI feedback information is the channel information recovered when the reset state information of the decoder is effective.

Figure 10 shows a schematic flow chart of another communication method according to an embodiment of the present application. The method 1100 shown in Figure 11 can be regarded as a specific implementation of the method shown in Figure 7 or Figure 8.

As shown in Figure 10, in method 1000, the first device is a network device, the second device is a terminal device, the encoder (an example of the first AI model) is deployed in the terminal device, and the decoder (an example of the second AI model) is deployed in an AI entity that communicates with the network device, for example, a near real-time RIC (an example of the third device).

The main difference between Figure 10 and Figure 9 is that in method 1000 shown in Figure 10, channel information recovery is performed by the decoder in the near-real-time RIC. Specifically, steps 1001 through 1007 in method 1000 are consistent with steps 910 through 970 in method 900 in Figure 9. For related descriptions, please refer to method 900 and will not be repeated here. Method 1000 also includes steps 1008, 1009, and 1010.

1008. The network device sends the first CSI feedback information to the near real-time RIC.

1009 : The near real-time RIC restores the channel information corresponding to the first CSI feedback information according to the reset decoder state information.

The first CSI feedback information is input into the decoder for processing, or the processed first CSI feedback information is input into the decoder for processing to obtain CSI recovery information corresponding to the first CSI feedback information. At this time, the state information of the decoder has been reset. The CSI recovery information corresponding to the first CSI feedback information is the channel information recovered when the reset state information of the decoder is effective.

1010. The near real-time RIC sends CSI recovery information corresponding to the first CSI feedback information to the network device.

In addition, the first indication information may also be sent to the terminal device by a near-real-time RIC. That is, the first device may be a near-real-time RIC, and the second device may be a terminal device. The decoder may be deployed in the near-real-time RIC. For example, the network device in steps 920, 970, and 980 of method 900 may be replaced with a near-real-time RIC. After executing step 980, the near-real-time RIC sends CSI recovery information corresponding to the first CSI feedback information to the network device.

Figure 11 shows a schematic flow chart of another communication method according to an embodiment of the present application. The method 1100 shown in Figure 11 can be considered as a specific implementation of the method shown in Figure 7 or Figure 8. For related descriptions, please refer to the previous text. To avoid repetition, some descriptions of method 1100 will be omitted when describing it.

As shown in Figure 11, in method 1100, the first device is a network device, the second device is an AI entity on the terminal device side, such as a server, such as an OTT device, or an OAM, the encoder is deployed on the AI entity on the terminal device side, such as a server, such as an OTT device, or an OAM, and the decoder is deployed on the network device (an example of a third device). Method 1100 may include the following steps.

1101. The network device sends a first reference signal to the terminal device.

1102. The network device sends first indication information to the terminal device. The first indication information is used to determine a time unit in which reset status information of the first AI model takes effect.

1103. The terminal device sends a first indication message to the OTT server.

For example, the first indication information may indicate the length of period #2.

1104. The OTT server determines, based on the first indication information, a time unit in which the reset encoder status information takes effect.

For example, the network device may send a DCI to the terminal device, where the DCI may be used to carry the first indication information, and the DCI may also indicate the scheduling of the first CSI feedback information. The OTT server may determine, based on the sending time of the DCI and the length of period #2, that the time unit in which the reset encoder status information takes effect is no later than the end time of period #2.

For example, the time when the first CSI feedback information is scheduled to be fed back may be the end time of time period #2.

The above is only an example. For other determination methods, please refer to the previous text and will not be repeated here.

1105 , the OTT server performs a reset of the encoder's status information.

The OTT server may complete resetting of the encoder's status information according to the time unit determined in step 1104 .

1106. The terminal device performs measurement of the first reference signal to obtain channel information.

1107. The terminal device sends the channel information to the OTT server.

1108. The OTT server generates first CSI feedback information corresponding to the channel information according to the state information of the reset encoder.

The channel information is input to the encoder for processing to obtain first CSI feedback information, at which point the encoder state information has been reset. The first CSI feedback information is CSI feedback information generated when the reset encoder state information is effective.

1109. The OTT server sends first CSI feedback information to the terminal device.

1110. The terminal device sends first CSI feedback information to the network device.

For example, the terminal device may send the first CSI feedback information to the network device at the end of time period #2.

1111. The network device restores the channel information corresponding to the first CSI feedback information according to the reset decoder status information.

The network device knows that the first CSI feedback information is generated when the reset state information of the encoder is effective, and can restore the channel information corresponding to the first CSI feedback information according to the reset state information of the decoder.

Figure 12 shows a schematic flow chart of another communication method according to an embodiment of the present application. The method 1200 shown in Figure 12 can be regarded as a specific implementation of the method shown in Figure 7 or Figure 8.

As shown in Figure 12, in method 1200, the first device is a network device, the second device is an AI entity on the terminal device side, such as a server, such as an OTT device, or OAM, the encoder is deployed in an AI entity on the terminal device side, such as a server, such as an OTT device, or OAM, and the decoder is deployed in an AI entity on the network device side, such as a near real-time RIC (an example of a third device) communicating with the network device.

The main difference between Figure 12 and Figure 11 is that in method 1200 shown in Figure 12, channel information recovery is performed by the decoder in the near-real-time RIC. Specifically, steps 1201 through 1210 in method 1200 are identical to steps 1101 through 1110 in method 1100 in Figure 11. For related descriptions, reference can be made to method 1100 and will not be repeated here. Method 1200 also includes steps 1211 through 1213. For descriptions of steps 1211 through 1213, reference can be made to steps 1008 through 1010 in Figure 10 and will not be repeated here.

Figure 13 shows a schematic flow chart of another communication method according to an embodiment of the present application. The method 1300 shown in Figure 13 can be considered as a specific implementation of the method shown in Figure 7 or Figure 8. For related descriptions, please refer to the previous text. To avoid repetition, some descriptions of method 1300 will be omitted as appropriate.

As shown in Figure 13, in method 1300, the first device is an AI entity on the network device side, such as a near-real-time RIC communicating with the network device, and the second device is an AI entity on the terminal device side, such as a server (such as an OTT device) or OAM communicating with the terminal device. The encoder is deployed on the OTT server, and the decoder is deployed on the near-real-time RIC. Method 1300 may include the following steps.

1301. A network device sends a first reference signal to a terminal device.

1302. The near real-time RIC sends first indication information to the OTT server. The first indication information is used to determine the time unit in which the reset status information of the first AI model takes effect.

Exemplarily, the first indication information may be carried in a transparent transmission instruction.

For example, the near real-time RIC can instruct the OTT server on the length of period #2 via transparent instructions.

1303. The OTT server determines, based on the first indication information, a time unit in which the reset encoder status information takes effect.

For example, the network device may send a DCI to the terminal device to instruct the scheduling of the first CSI feedback information. The OTT server may determine that the reset encoder status information takes effect at a time unit no later than the end time of period #2 based on the sending time of the DCI and the length of period #2.

For another example, the OTT server may determine, based on the sending time of the first indication information and the length of period #2, that the time unit in which the reset encoder status information takes effect is no later than the end time of period #2.

For example, the time when the first CSI feedback information is scheduled to be fed back may be the end time of time period #2.

The above is only an example. For other determination methods, please refer to the previous text and will not be repeated here.

1304. The OTT server performs a reset of the encoder's status information.

The OTT server may complete resetting of the encoder's status information according to the time unit determined in step 1303 .

1305. The terminal device performs measurement of the first reference signal to obtain channel information.

1306. The terminal device sends the channel information to the OTT server.

1307. The OTT server generates first CSI feedback information corresponding to the channel information according to the state information of the reset encoder.

The process of generating the first CSI feedback information can be referred to above and will not be repeated here.

1308. The OTT server sends first CSI feedback information to the near real-time RIC.

For example, the OTT server may send the first CSI feedback information to the near real-time RIC at the end of period #2. In this case, the end of period #2 may be considered as an example of the first time unit.

1309 , the near real-time RIC restores the channel information corresponding to the first CSI feedback information according to the reset decoder state information.

The recovery process of the CSI recovery information corresponding to the first CSI feedback information can be referred to above and will not be repeated here.

1310. The near real-time RIC sends CSI recovery information corresponding to the first CSI feedback information to the network device.

Alternatively, the first indication information may be sent to the OTT server by the network device. For example, if the decoder is deployed on the network device, step 1310 is not required, and the near-real-time RIC in method 1300 is replaced by the network device. For another example, if the decoder is deployed on the near-real-time RIC, the near-real-time RIC in steps 1301 to 1308 may be replaced by the network device, with the network device sending the first CSI feedback information to the near-real-time RIC, and steps 1309 and 1310 are continued.

Figure 14 is a schematic diagram of a CSI feedback process according to an embodiment of the present application. Figure 15 is a schematic flow chart of the communication method employed in this feedback process. The method shown in Figure 15 can be considered a specific implementation of the method shown in Figure 7 or Figure 8. For a detailed description, please refer to the previous text. To avoid repetition, some descriptions will be omitted when describing method 1500.

For ease of description, in Figures 14 and 15, the first device is a network device and the second device is a terminal device as an example for illustration, wherein the encoder is deployed on the terminal device and the decoder is deployed on the network device. In the case where the first device and the second device are other devices, adaptive modifications can be made with reference to Figures 10 to 13. In Figures 14 and 15, the length of time period #2 indicated by the first indication information is used as an example for illustration. In other implementations, the first indication information may also indicate other content, that is, the second device may also determine the time unit in which the reset status information of the first AI model takes effect by other means. For a specific description, please refer to method 700, which will not be repeated here.

As shown in FIG. 15 , method 1500 may include the following steps.

1501. The network device sends CSI-RS#1 to the terminal device.

Before time T0 of the network device, the network device sends CSI-RS#1 to the terminal device.

1502. The network device sends DCI#1.

At time T0 of the network device, the network device sends DCI#1 to the terminal device in a downlink (DL), indicating the feedback information corresponding to the scheduling CSI-RS#1, for example, CSI-1.

1503. The terminal device feeds back CSI-1.

The terminal device, following the instructions of the network device, feeds back CSI-1 in the uplink UCI. CSI-1 packets are lost in the uplink. For example, due to poor uplink transmission conditions, CSI-1 packets are lost. This means that the terminal device has sent CSI-1, but the network device has not received it.

1504. The network device sends CSI-RS#2 to the terminal device.

Before time T1 of the network device, the network device sends CSI-RS#2 to the terminal device.

1505. The network device sends DCI#2.

At time T1 of the network device, the network device sends DCI#2 to the terminal device, indicating the feedback information CSI-2 corresponding to the scheduling CSI-RS#2.

1506: The network device detects CSI-1 packet loss.

At time T2, the network device discovers that CSI-1 is lost.

1507. The network device sends CSI-RS#3 to the terminal device.

Before time T3 of the network device, the network device sends CSI-RS#3 to the terminal device.

1508. The network device sends DCI#3.

At time T3, the network device sends DCI#3 to the terminal device, instructing it to reset the encoder status information and indicating the length of period #2, Z1. Z1 is a positive number. This means the first indication is carried in DCI#3. The reset encoder status information takes effect no later than the end time of period #2. Period #2 starts at time T3 and ends at time T5 = T3 + Z1.

The terminal device determines, according to the instruction of the network device, that the time unit at which the reset encoder status information takes effect is no later than time T5.

DCI#3 also indicates the feedback information CSI-3 corresponding to the scheduled CSI-RS#3. The time when CSI-3 is scheduled for feedback is time T5.

The terminal device determines that the time when CSI-3 is scheduled to be fed back is time T5 according to the instruction of the network device.

1509. The terminal device feeds back CSI-2.

Before T5, the terminal device, following the instructions of the network device, feeds back CSI-2 in the uplink UCI. Upon receiving the CSI-2, the network device recognizes that it was generated before the reset encoder status information took effect. The network device can use the decoder status information that has not been reset to recover the channel information corresponding to the CSI-2.

1510. The terminal device feeds back CSI-3.

The terminal device completes resetting the encoder status information no later than time T5 according to the instruction of the network device, i.e., the reset is completed at time T5. The reset encoder status information is used to generate CSI-3.

At time T5 of the network device, the terminal device, in accordance with the instructions of the network device, feeds back CSI-3 in the uplink UCI. Time T5 can be considered an example of the first time unit. CSI-3 can be considered an example of first CSI feedback information. Upon receiving CSI-3, the network device recognizes that the CSI-3 was generated with the reset encoder status information in effect. The network device can use the reset decoder status information to recover the channel information corresponding to the CSI-3.

FIG14 and FIG15 do not show the process of generating the CSI feedback information and the process of recovering the channel information corresponding to the CSI feedback information. For detailed description, please refer to method 700 and will not be repeated here.

FIG21 shows a schematic diagram of another CSI feedback process according to an embodiment of the present application.

For ease of description, FIG21 takes the first device as a network device and the second device as a terminal device as an example for illustration, wherein the encoder is deployed on the terminal device and the decoder is deployed on the network device. In the case where the first device and the second device are other devices, adaptive modifications can be made with reference to FIG10 to FIG13. In FIG21, the length of time period #2 indicated by the first indication information is used as an example for illustration. In other implementations, the first indication information may also indicate other content, that is, the second device may also determine the time unit in which the reset status information of the first AI model takes effect by other means. For a specific description, please refer to method 700.

The communication method used in the feedback process shown in FIG21 differs primarily from method 1500 shown in FIG15 in that the network device sends DCI#2' to the terminal device, indicating the scheduling of feedback information CSI-2' corresponding to CSI-RS#2. If the time resource of CSI-2' indicated by DCI#2' overlaps with time period #2, as shown in FIG21, the time when CSI-2' is scheduled to be sent falls within time period #2, then the terminal device skips or ignores the reporting of CSI-2'. For other descriptions, please refer to FIG15. To avoid repetition, some descriptions are omitted when describing FIG21.

As shown in FIG21 , at time T1 of the network device, the network device sends DCI#2’ to the terminal device, indicating the feedback information CSI-2’ corresponding to the scheduling CSI-RS#2.

The terminal device determines, based on the instructions of the network device, that the reset encoder status information takes effect no later than the end of period #2, i.e., time T5 = T3 + Z1. If CSI-2' is scheduled to be transmitted between T3 and T5, the terminal device skips or ignores reporting of CSI-2'.

Figure 16 shows a schematic flow chart of another communication method according to an embodiment of the present application. The method 1600 shown in Figure 16 can be considered as a specific implementation of the method shown in Figure 7 or Figure 8. For related descriptions, please refer to the previous text. To avoid repetition, some descriptions are omitted when describing method 1600.

As shown in FIG. 16 , method 1600 may include the following steps.

1601. The second device sends third indication information to the first device. The third indication information indicates the duration required to reset the status information of the first AI model.

1602. The first device sends first indication information to the second device. The first indication information is used to determine a time unit in which reset status information of the first AI model takes effect.

The first device can determine the time unit in which the reset status information of the first AI model takes effect based on the duration required to reset the status information of the first AI model, and send first indication information to the second device, so that the second device can determine the time unit in which the reset status information of the first AI model takes effect.

For example, the first device may determine the length of period #2 based on the duration required to reset the state information of the first AI model, and indicate the length of period #2 to the second device through the first indication information. The length of period #2 is greater than or equal to the duration required to reset the state information of the first AI model.

1603. The second device determines a time unit in which the reset status information of the first AI model takes effect.

The second device may determine, based on the first indication information, a time unit in which the reset status information of the first AI model takes effect.

For example, the first indication information indicates the length of period #2, and the time unit in which the reset status information of the first AI model takes effect is no later than the end time of period #2.

1604. The second device performs a reset of the status information of the first AI model.

The second device may determine a time to start resetting the status information of the first AI model based on the time unit determined in step 1603 and the duration required to reset the status information of the first AI model. The time interval between the time to start resetting the status information of the first AI model and the time unit determined in step 820 is greater than or equal to the duration required to reset the status information of the first AI model.

For example, the time interval between the start time of resetting the state information of the first AI model and the end time of period #2 is greater than or equal to the duration required for resetting the state information of the first AI model.

1605. The second device generates first CSI feedback information.

The second device may generate first CSI feedback information by using the reset state information of the first AI model.

1606. The second device sends first CSI feedback information to the first device.

FIG17 shows a schematic diagram of a CSI feedback process according to an embodiment of the present application.

For ease of description, FIG17 illustrates an example in which the first device is a network device and the second device is a terminal device, wherein the encoder is deployed on the terminal device and the decoder is deployed on the network device. In the case where the first device and the second device are other devices, adaptive modifications can be made with reference to FIG10 to FIG13. In FIG17, the length of time period #2 indicated by the first indication information is used as an example for illustration. In other implementations, the first indication information may also indicate other content, that is, the second device may also determine the time unit in which the reset status information of the first AI model takes effect by other means. For a specific description, reference may be made to method 700.

The main difference between the communication method adopted in the feedback process shown in Figure 17 and the method 1500 shown in Figure 15 is that the terminal device reports the time R1 required to reset the encoder status information to the network device. For other descriptions, please refer to Figure 15. To avoid repetition, some descriptions are appropriately omitted when describing Figure 17.

(1) At time T0 of the network device, the network device sends DCI#1 to the terminal device, indicating the feedback information CSI-1 corresponding to the scheduling CSI-RS#1.

Before time T0 of the network device, the network device has sent CSI-RS#1 to the terminal device.

(2) The terminal device feeds back CSI-1 in the uplink UCI according to the instructions of the network device.

CSI-1 packet loss in uplink.

(3) At time T1 of the network device, the network device sends DCI#2 to the terminal device, indicating the feedback information CSI-2 corresponding to the scheduling CSI-RS#2.

Before time T1 of the network device, the network device has sent CSI-RS#2 to the terminal device.

(4) At time T2, the network device discovers that CSI-1 is lost.

(5) At time T3 of the network device, the network device sends DCI #3 to the terminal device, instructing the terminal device to reset the status information of the encoder and indicating the length Z1 of time period #2.

The network device can determine Z1 based on R1. Z1 is greater than or equal to R1. Z1 is a positive number and R1 is a positive number.

The terminal device determines, according to the instruction of the network device, that the time unit in which the reset encoder status information takes effect is no later than the end time of period #2, that is, time T5=T3+Z1.

DCI#3 also indicates the feedback information CSI-3 corresponding to the scheduled CSI-RS#3. The time when CSI-3 is scheduled for feedback is time T5.

Before time T3 of the network device, the network device has sent CSI-RS#3 to the terminal device.

The terminal device determines that the time when CSI-3 is scheduled to be fed back is time T5 according to the instruction of the network device.

The terminal device may also determine the time to start resetting the state information of the encoder based on the time T5 and R1. For example, the terminal device may determine to start resetting the state information of the encoder at the time T4 = T5 - R1.

(6) Before time T5 of the network device, the terminal device feeds back CSI-2 in the uplink UCI according to the instruction of the network device.

The network device receives the CSI-2 and knows that the CSI-2 was generated when the reset encoder state information was not yet effective. The network device can use the state information of the decoder that has not been reset to restore the channel information corresponding to the CSI-2.

(7) At time T4 of the network device, the terminal device begins to reset the status information of the encoder.

After the reset is completed, the state information of the reset encoder can be used to generate CSI-3.

(8) At time T5 of the network device, the terminal device feeds back CSI-3 in the uplink UCI according to the instruction of the network device.

The network device receives the CSI-3 and knows that the CSI-3 was generated with the reset encoder state information in effect. The network device can use the reset decoder state information to restore the channel information corresponding to the CSI-3.

It can be understood that in some of the above embodiments, the information names involved are merely examples and do not limit the scope of protection of the embodiments of the present application.

It is also understood that the formulas involved in the various embodiments of the present application are merely illustrative and do not limit the scope of protection of the embodiments of the present application. In the process of calculating the various parameters involved above, calculations can also be performed according to the above formulas, or based on variations of the above formulas, or calculations can be performed in other ways to meet the results of the formula calculations.

It can also be understood that some optional features in the various embodiments of the present application may not depend on other features in certain scenarios, and may also be combined with other features in certain scenarios, without limitation.

It can also be understood that the solutions in the various embodiments of the present application can be reasonably combined and used, and the explanations or descriptions of the various terms appearing in the embodiments can be referenced or explained with each other in the various embodiments, without limitation to this.

It can also be understood that the sizes of the various numerical serial numbers in the embodiments of the present application do not mean the order of execution, but are only distinguished for the convenience of description and should not constitute any limitation on the implementation process of the embodiments of the present application.

It can also be understood that in the above-mentioned various method embodiments, the methods and operations implemented by the device can also be implemented by components (such as chips or circuits) of the device.

Corresponding to the methods provided in the above method embodiments, embodiments of the present application also provide corresponding apparatuses, which include modules for executing the corresponding methods in the above method embodiments. The modules may be software, hardware, or a combination of software and hardware. It is understood that the technical features described in the above method embodiments are also applicable to the following apparatus embodiments.

Figure 19 is a schematic diagram of a communication device 1900 provided in an embodiment of the present application. Device 1900 includes a transceiver unit 1910 and a processing unit 1920. Transceiver unit 1910 can be used to implement corresponding communication functions. Transceiver unit 1910 can also be referred to as a communication interface or communication unit. Processing unit 1920 can be used to implement corresponding processing functions, such as configuring resources.

Optionally, the device 1900 also includes a storage unit, which can be used to store instructions and/or data, and the processing unit 1920 can read the instructions and/or data in the storage unit so that the device implements the actions of the device or network element in the aforementioned method embodiments.

The device 1900 can be a second device, or a communication device that is applied to a second device or used in combination with a second device and can implement a communication method executed on the second device side; or, the device 1900 can be a first device, or a communication device that is applied to a first device or used in combination with a first device and can implement a communication method executed on the first device side.

When the apparatus 1900 is applied to a second device, the apparatus 1900 can implement the steps or processes corresponding to those performed by the second device in the above method embodiments. The transceiver unit 1910 can be used to perform the transceiver-related operations of the second device in the above method embodiments, and the processing unit 1920 can be used to perform the processing-related operations of the second device in the above method embodiments.

When the device 1900 is applied to the first device, the device 1900 can implement the steps or processes corresponding to those performed by the first device in the above method embodiment, wherein the transceiver unit 1910 can be used to perform the transceiver-related operations of the first device in the above method embodiment, and the processing unit 1920 can be used to perform the processing-related operations of the first device in the above method embodiment.

It should be understood that the specific process of each unit executing the above corresponding steps has been described in detail in the above method embodiments, and for the sake of brevity, it will not be repeated here.

It should also be understood that the device 1900 herein is embodied in the form of a functional unit. The term "unit" herein may refer to an application specific integrated circuit (ASIC), an electronic circuit, a processor (e.g., a shared processor, a dedicated processor, or a group processor, etc.) and memory for executing one or more software or firmware programs, a combined logic circuit, and/or other suitable components that support the described functions. In an optional example, those skilled in the art will understand that the device 1900 may be specifically the first device in the above-mentioned embodiment, and may be used to execute the various processes and/or steps corresponding to the first device in the above-mentioned method embodiments; or, the device 1900 may be specifically the second device in the above-mentioned embodiment, and may be used to execute the various processes and/or steps corresponding to the second device in the above-mentioned method embodiments. To avoid repetition, these will not be described in detail here.

The apparatus 1900 of each of the above-mentioned solutions has the function of implementing the corresponding steps performed by the device (such as the first device, and the second device) in the above-mentioned method. The functions can be implemented by hardware, or the corresponding software can be implemented by hardware. The hardware or software includes one or more modules corresponding to the above-mentioned functions; for example, the transceiver unit can be replaced by a transceiver (for example, the sending unit in the transceiver unit can be replaced by a transmitter, and the receiving unit in the transceiver unit can be replaced by a receiver), and other units, such as the processing unit, can be replaced by a processor to respectively perform the sending and receiving operations and related processing operations in each method embodiment.

In addition, the transceiver unit 1910 may also be a transceiver circuit (for example, including a receiving circuit and a transmitting circuit), and the processing unit 1920 may be a processing circuit. The processing circuit may include one or more processors, or circuits in one or more processors for processing functions.

It should be noted that the apparatus in FIG19 may be a network element or device in the aforementioned embodiments, or may be a chip or chip system, such as a system on a chip (SoC). The transceiver unit may be an input/output circuit or a communication interface, and the processing unit may be a processor, microprocessor, or integrated circuit integrated on the chip. This is not limited here.

Figure 20 is a schematic diagram of another communication device 2000 provided in an embodiment of the present application. Device 2000 includes a processor 2010, which is configured to execute computer programs or instructions stored in memory 2020, or read data/signaling stored in memory 2020, to perform the methods described in the above method embodiments. Optionally, there may be one or more processors 2010.

Optionally, as shown in FIG20 , the apparatus 2000 further includes a memory 2020 for storing computer programs or instructions and/or data. The memory 2020 may be integrated with the processor 2010 or may be separately provided. Optionally, there may be one or more memories 2020.

Optionally, as shown in FIG20 , the apparatus 2000 further includes a transceiver circuit 2030, which is configured to receive and/or transmit signals. For example, the processor 2010 is configured to control the transceiver circuit 2030 to receive and/or transmit signals. The processor 2010 may also be replaced by a processing circuit.

The device 2000 may be a network element or device in the aforementioned embodiments, or may be a chip or chip system. When the device 2000 is a network element or device in the aforementioned embodiments, the transceiver circuit 2030 may be a transceiver. When the device 2000 is a chip or chip system, the transceiver circuit 2030 may be an interface circuit or an input/output interface.

As a solution, the apparatus 2000 can be applied to a second device. Specifically, the apparatus 2000 can be the second device, or a device that can support the second device and implement the functions of the second device in any of the above-mentioned examples. The apparatus 2000 is used to implement the operations performed by the second device in each of the above-mentioned method embodiments.

For example, the processor 2010 is configured to execute computer programs or instructions stored in the memory 2020 to implement relevant operations of the second device in each of the above method embodiments.

As another solution, the apparatus 2000 can be applied to a first device. Specifically, the apparatus 2000 can be the first device, or a device that can support the first device and implement the functions of the first device in any of the above-mentioned examples. The apparatus 2000 is used to implement the operations performed by the first device in each of the above-mentioned method embodiments.

For example, the processor 2010 is configured to execute computer programs or instructions stored in the memory 2020 to implement relevant operations of the first device in each of the above method embodiments.

It should be understood that the processor mentioned in the embodiments of the present application may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), ASICs, field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

It should also be understood that the memory mentioned in the embodiments of the present application can be a volatile memory and/or a non-volatile memory. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM). For example, RAM can be used as an external cache. By way of example and not limitation, RAM includes the following forms: static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM).

It should be noted that when the processor is a general-purpose processor, DSP, ASIC, FPGA or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, the memory (storage module) can be integrated into the processor.

It should also be noted that the memory described herein is intended to include, but is not limited to, these and any other suitable types of memory.

An embodiment of the present application further provides a computer-readable storage medium storing computer instructions for implementing the methods executed by the communication device in the above-mentioned method embodiments.

For example, when the computer program is executed by a computer, the computer can implement the method performed by the first device in each embodiment of the above method.

For another example, when the computer program is executed by a computer, the computer can implement the method performed by the second device in each embodiment of the above method.

An embodiment of the present application further provides a computer program product comprising instructions, which, when executed by a computer, implement the methods performed by a device (such as the first device or the second device) in the above-mentioned method embodiments.

The embodiment of the present application further provides a communication system, including the aforementioned first device and second device. The first device and second device can implement the communication method shown in any of the aforementioned examples.

Optionally, the system further includes a device for communicating with the first device and/or the second device.

The explanation of the relevant contents and beneficial effects of any of the above-mentioned devices can be referred to the corresponding method embodiments provided above, which will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. For example, the computer can be a personal computer, a server, or a network device, etc. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrations. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state disk (SSD)). For example, the aforementioned available medium includes, but is not limited to, various media that can store program codes, such as a USB flash drive, a mobile hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims (38)

A communication method, comprising: First indication information is obtained from a first device, where the first indication information is used to determine a time unit in which reset status information of a first artificial intelligence (AI) model takes effect. The method according to claim 1, further comprising: First channel state information (CSI) feedback information is sent to the first device in a first time unit, where the first CSI feedback information is related to the reset state information of the first AI model, and the first time unit is no earlier than a time unit when the reset state information of the first AI model takes effect. The method according to claim 1 or 2 is characterized in that the first indication information is used to indicate that uplink information is sent on a first time resource, and the time unit in which the reset status information of the first AI model takes effect is no later than the start time of the first time resource. The method according to claim 3, further comprising: Determine a first time period, where the end time of the first time period is no later than the start time of the first time resource, and the length of the first time period is predefined, or indicated by the first indication information, or indicated by other information other than the first indication information. The method according to claim 1 or 2, characterized in that the method further comprises: Obtain the length of a first time period, where the length of the first time period is used to determine the time unit in which the reset status information of the first AI model takes effect. The length of the first time period is predefined, or indicated by the first indication information, or indicated by other information other than the first indication information. The method according to claim 5 is characterized in that the starting time of the first time period is the sending time of the first indication information, the receiving time of the first indication information, or the time indicated by the first indication information. The method according to any one of claims 4 to 6, further comprising: receiving scheduling information for second CSI feedback information, where a time resource of the second CSI feedback information indicated by the scheduling information overlaps with the first time period, ignoring or skipping sending the second CSI feedback information, and the second CSI feedback information is related to the first AI model. The method according to any one of claims 1 to 7, characterized in that the first indication information indicates the resetting of the status information of the first AI model. The method according to any one of claims 1 to 7, further comprising: Second indication information is obtained from the first device, where the second indication information indicates resetting status information of the first AI model. The method according to any one of claims 1 to 9, characterized in that the first indication information is related to a duration required for resetting the status information of the first AI model, and the duration required for resetting the status information of the first AI model is predefined. The method according to any one of claims 1 to 9, wherein the first indication information is related to a duration required to reset the state information of the first AI model, and the method further comprises: Send third indication information to the first device or other devices other than the first device, where the third indication information indicates the time required to reset the status information of the first AI model. The method according to claim 11 is characterized in that the third indication information is carried in signaling sent by the terminal device and carrying the terminal device capabilities, or the third indication information is carried in signaling carrying configuration information of the first AI model. The method according to any one of claims 1 to 12 is characterized in that the first indication information is carried in first downlink control information DCI, and/or higher layer signaling, and the higher layer signaling includes radio resource control signaling or media access layer control unit signaling. The method according to claim 13, wherein the first DCI is any one of the following: The first DCI is further used to trigger the first CSI feedback information, where the first CSI feedback information belongs to an aperiodic CSI report; The first DCI is further used to schedule or configure an uplink shared channel, and the uplink shared channel does not include CSI feedback information; or The first DCI is also used to carry first indication information of other terminal devices. The method according to any one of claims 2 to 14, characterized in that the input of the first AI model includes a channel measurement result of a first reference signal, the first CSI feedback information is the output of the first AI model or is based on the output of the first AI model, the output of the first AI model is related to the measurement result of the first reference signal and reset state information of the first AI model, and the reset state information of the first AI model is initial state information, or the reset state information of the first AI model is state information of the first AI model before the transmission time of the first reference signal. A communication method, comprising: First indication information is sent to the second device, where the first indication information is used to determine a time unit in which reset status information of the first artificial intelligence AI model takes effect. The method according to claim 16, further comprising: First channel state information (CSI) feedback information is received from the second device in a first time unit, where the first CSI feedback information is related to the reset state information of the first AI model, and the first time unit is no earlier than a time unit when the reset state information of the first AI model takes effect. The method according to claim 16 or 17 is characterized in that the first indication information is used to indicate that uplink information is sent on a first time resource, and the time unit in which the reset status information of the first AI model takes effect is no later than the start time of the first time resource. The method according to claim 18, wherein the first indication information indicates a length of the first time period, or the method further comprises: sending fourth indication information to the second device, where the fourth indication information indicates the length of the first time period; The end time of the first time period is no later than the start time of the first time resource. The method according to claim 16 or 17, wherein the first indication information indicates a length of the first time period, or the method further comprises: sending fourth indication information to the second device, where the fourth indication information indicates the length of the first time period; The length of the first time period is used to determine the time unit in which the reset status information of the first AI model takes effect. The method according to claim 20 is characterized in that the starting time of the first time period is the sending time of the first indication information, the receiving time of the first indication information, or the time indicated by the first indication information. The method according to any one of claims 19 to 21, further comprising: Sending scheduling information of third CSI feedback information, where a time resource of the third CSI feedback information indicated by the scheduling information does not include the first time period; Receive the third CSI feedback information, where the third CSI feedback information is related to the first AI model. The method according to any one of claims 16 to 22, characterized in that the first indication information indicates the resetting of the status information of the first AI model. The method according to any one of claims 16 to 22, further comprising: Send second indication information to the second device, where the second indication information indicates resetting the status information of the first AI model. The method according to any one of claims 16 to 24 is characterized in that the first indication information is related to a duration required for resetting the status information of the first AI model, and the duration required for resetting the status information of the first AI model is predefined. The method according to any one of claims 16 to 24, wherein the first indication information is related to a duration required to reset the state information of the first AI model, and the method further comprises: Receive third indication information from the second device or other device other than the second device, where the third indication information indicates the time required to reset the status information of the first AI model. The method according to claim 26 is characterized in that the third indication information is carried in signaling sent by the terminal device and carrying the terminal device capabilities, or the third indication information is carried in signaling carrying configuration information of the first AI model. The method according to any one of claims 16 to 27 is characterized in that the first indication information is carried in first downlink control information DCI, and/or higher-layer signaling, and the higher-layer signaling includes radio resource control signaling or media access layer control unit signaling. The method according to claim 28, wherein the first DCI is any one of the following: The first DCI is further used to trigger the first CSI feedback information, where the first CSI feedback information belongs to an aperiodic CSI report; The first DCI is further used to schedule or configure an uplink shared channel, and the uplink shared channel does not include CSI feedback information; or The first DCI is also used to carry first indication information of other terminal devices. The method according to any one of claims 17 to 29, characterized in that the input of the first AI model includes a channel measurement result of a first reference signal, the first CSI feedback information is the output of the first AI model or is based on the output of the first AI model, the output of the first AI model is related to the measurement result of the first reference signal and reset state information of the first AI model, and the reset state information of the first AI model is initial state information, or the reset state information of the first AI model is state information of the first AI model before the transmission time of the first reference signal. A computer-readable storage medium, characterized in that the computer-readable storage medium includes instructions, and when the instructions are executed by a processor, the method according to any one of claims 1 to 15 or 16 to 30 is implemented. A communication device, characterized in that the communication device includes a processor coupled to a storage medium, the storage medium storing instructions, and when the instructions are executed by the processor, the communication device executes the method as described in any one of claims 1 to 15 or 16 to 30. A communication device, characterized by comprising a module for executing the method according to any one of claims 1 to 15. A communication device, characterized by comprising a module for executing the method according to any one of claims 16 to 30. A communication device, characterized by comprising one or more processors, wherein the one or more processors are configured to process data and/or information so that the method according to any one of claims 1 to 15 or 16 to 30 is implemented. A chip, characterized in that it comprises a processor, wherein the processor is used to run a program or instruction so that the method according to any one of claims 1 to 15 or 16 to 30 is implemented. A computer program product, characterized in that the computer program product comprises computer program codes or instructions, and when the computer program codes or instructions are executed, the method according to any one of claims 1 to 15 or 16 to 30 is implemented. A communication system, characterized by comprising the communication device according to claim 33 and/or the communication device according to claim 34.