Communication method and communication device

Abstract

The invention provides a communication method and device, and the method comprises the steps that a terminal side obtains a first data set, each sample in the first data set comprises first CSI, CSI feedback information and second CSI, the first CSI and the second CSI represent the channel states of a first resource set and a second resource set respectively, and the first resource set and the second resource set respectively represent the channel states of the first resource set and the second resource set; the second resource set is part or all of the first resource set, the CSI feedback information is obtained by compressing the first CSI or a third CSI, the third CSI represents the channel state of the third resource set, and the third resource set is part of the first resource set or is different from the first resource set; the terminal side performs model training on the model for processing the CSI to be fed back based on the first data set, so that the generalization ability of the model can be improved, and the feedback performance of the CSI can be improved.

Description

Technical Field

[0001] This application relates to the field of wireless communications, and in particular, to a communication method and a communication device. Background Art

[0002] In the feedback of channel state information (CSI) based on artificial intelligence (AI), when the model is deployed on the base station side, the base station side obtains the estimation result of the channel state information-reference signal (CSI-RS) on the user equipment (UE) side as a label for model training. The CSI feedback can be implemented based on an auto-encoder (AE) model. The AE model generally refers to a network structure composed of two sub-models. For example, it can include an encoder and a decoder as two sub-models. The encoder and decoder of the AE model are generally co-trained and can be used in a matching manner. In one implementation, the base station side co-trains the encoder and decoder locally based on the historically collected CSI data. To achieve the pairing of the dual-end models, the base station side constructs a data set based on the input (i.e., the measured CSI) and output (i.e., the feedback CSI) of the encoder and transmits the data set to the UE side. The UE side uses the data set to train the encoder sub-model to enable the pairing of the dual-end models on the base station side and the UE side. In this solution, how to construct the data set transmitted from the network side to the terminal side to improve the CSI feedback performance is an issue that needs to be considered. Summary of the Invention

[0003] This application provides a communication method and a communication device. The terminal side can use the data set constructed by the network side for model training to improve the feedback performance of CSI.

[0004] In a first aspect, a communication method is provided. This method can be applied to the terminal side. For example, it can be executed by a terminal device; alternatively, it can also be executed by components deployed in the terminal device, such as circuits or chips inside the terminal device (such as a modem chip, also known as a baseband chip, or a system on chip (SoC) chip or a system in package (SIP) chip that includes a modem core, etc.); alternatively, it can also be executed by a device outside the terminal device (such as the host of an over the top (OTT) system or a cloud server) or components in the device (such as chips, processors, or circuits inside the device, etc.); or it can also be implemented by a logic module or software that can implement all or part of the functions of the terminal device, and so on. Or rather, this method can be executed by a first device. The first device can be a terminal device, or a component in the terminal device, such as a circuit or chip inside the terminal device (such as a modem chip), or an SoC chip or a SIP chip that includes a modem core, etc.; the first device can also be a device outside the terminal device (such as the host of an OTT system or a cloud server, etc.) or a component in the device (such as a chip, processor, or circuit inside the device); the first device can also be a logic module or software that can implement part or all of the function logic on the terminal side, and so on. This application does not make any limitations in this regard. For the convenience of understanding and explanation, the method provided in the first aspect will be described below by taking the terminal side as an example.

[0005] The method includes: obtaining a first data set. Each sample in the first data set includes the following information: first channel state information (CSI), CSI feedback information, and second CSI. The first CSI characterizes the channel state of a first resource set. The second CSI characterizes the channel state of a second resource set. The second resource set is part or all of the first resource set. The first resource set includes at least one of frequency domain resources or spatial domain resources. The CSI feedback information is obtained by compressing the first CSI or a third CSI. The third CSI characterizes the channel state of a third resource set. The third resource set is part of the first resource set, or alternatively, the third resource set is different from the first resource set. The first data set is used for model training of a first model. The first model is used for performing a first process on the CSI to be fed back. The first process at least includes CSI compression.

[0006] Based on the above solution, the terminal side obtains a first data set, each sample of the first data set includes multiple CSIs (such as a first CSI and a second CSI) and CSI feedback information, and based on the first data set, a model for CSI compression of the CSI to be fed back is trained, and a UE-side model that can be docked with the network-side model can be trained. The terminal side performing CSI feedback based on the trained first model is beneficial to improving the feedback performance of the CSI.

[0007] Combined with the first aspect, in some implementation manners of the first aspect, the first resource set corresponds to a first time-domain resource, the second resource set corresponds to a second time-domain resource and / or the first time-domain resource, the second time-domain resource is earlier than the first time-domain resource, and the third resource set corresponds to the first time-domain resource.

[0008] Based on the above solution, each sample in the first data set may include multiple different CSIs, such as CSIs with different corresponding time-domain resources and / or frequency-domain resources, or the multiple CSIs may further include historical CSIs, such as a second CSI corresponding to a second time-frequency resource, so as to increase the diversity of the samples, enable the model to obtain accurate CSIs according to different CSIs, and thus improve the CSI feedback performance.

[0009] Combined with the first aspect, in some implementation manners of the first aspect, the second resource set includes a first part and a second part, the first part corresponds to the second time-domain resource, and the second part corresponds to the first time-domain resource.

[0010] Based on the above solution, the second CSI in each sample in the first data set may include a part of the first CSI and historical CSI.

[0011] Combined with the first aspect, in some implementation manners of the first aspect, receive the first data set from the network side.

[0012] Combined with the first aspect, in some implementation manners of the first aspect, receive a second data set and a third data set from the network side. The second data set includes multiple first CSIs, the multiple first CSIs correspond to multiple first time-domain resources, and the multiple first time-domain resources do not overlap with each other. The third data set includes multiple CSI feedback information, and one of the multiple first CSIs corresponds to one or more of the multiple CSI feedback information.

[0013] Based on the above solution, compared with the network side sending the first data set, the network side sending the second data set and the third data set to enable the terminal side to determine the first data set can reduce signaling overhead.

[0014] In combination with the first aspect, in some implementations of the first aspect, one CSI feedback information among the multiple CSI feedback information corresponds to the same time domain resource as a first CSI corresponding to the one CSI feedback information.

[0015] In combination with the first aspect, in some implementations of the first aspect, the first data set is determined based on the second data set and the third data set.

[0016] In combination with the first aspect, in some implementations of the first aspect, it is determined that the first CSI included in each sample in the first data set is one of the multiple first CSIs, the second CSI included in each sample is determined according to the first CSI included in each sample and other first CSIs among the multiple first CSIs, and the CSI feedback information included in each sample is the CSI feedback information corresponding to the first CSI included in each sample among the multiple CSI feedback information.

[0017] Based on the above solution, the terminal side can determine the information included in each sample in the first data set based on the second data set and the third data set, which can reduce the signaling overhead of the network side for transmitting the data set.

[0018] In combination with the first aspect, in some implementations of the first aspect, the second CSI included in each sample is determined by partial CSIs in the first CSI included in each sample and partial CSIs in the other first CSIs.

[0019] In combination with the first aspect, in some implementations of the first aspect, the time domain resource corresponding to the other first CSI is earlier than the time domain resource corresponding to the first CSI included in each sample.

[0020] In combination with the first aspect, in some implementations of the first aspect, the use of the first data set for model training of a first model may include: using the first data set for model training of a second model, the second model being used for second processing of compressed CSI feedback information, the second processing at least including CSI reconstruction, and the second model obtained through model training being used for model training of the first model.

[0021] In combination with the first aspect, in some implementations of the first aspect, the method further includes: performing model training on the first model based on the first data set.

[0022] Exemplarily, performing model training on the first model based on the first data set may be: performing model training on a second model based on the first data set, the second model being used for second processing of compressed CSI feedback information, the second processing at least including CSI reconstruction; and performing model training on the first model based on input CSI and the second model obtained through model training.

[0023] Based on the above solution, the terminal side can use the first data set to train the first model, and a UE-side model that can be docked with the network-side model can be trained. The terminal side's CSI feedback based on the trained first model is beneficial to improving the CSI feedback performance.

[0024] In a second aspect, a communication method is provided. This method can be applied to the network side. For example, it can be executed by a network device; or, it can also be executed by a component deployed in the network device, such as a circuit or chip inside the network device (such as a modem chip, also known as a baseband chip, or a SoC chip or SIP chip containing a modem core, etc.); or, it can also be executed by a device outside the network device (such as an intelligent network element on the network side) or a component in the device (such as a chip, processor, or circuit inside the device, etc.); or, it can also be implemented by a logical module or software that can implement all or part of the functions of the network device, and so on. Or rather, this method can be executed by a second device. The second device can be a network device or a component in the network device, such as a circuit or chip inside the network device (such as a modem chip), or a SoC chip or SIP chip containing a modem core, etc.; the second device can also be a device outside the network device (such as an intelligent network element) or a component in the device (such as a chip, processor, or circuit inside the device); the second device can also be a logical module or software that can implement part or all of the functions of the network side, and so on. This application does not make any limitations in this regard. For ease of understanding and explanation, the method provided in the second aspect will be described below with the network side as an example.

[0025] This method includes: indicating a first data set, where the first data set is used for model training of a first model. The first model is used for performing a first process on the channel state information CSI to be fed back. The first process at least includes CSI compression. Each sample in the first data set includes the following information: first CSI, CSI feedback information, and second CSI; where the first CSI represents the channel state of a first resource set, the second CSI represents the channel state of a second resource set, the second resource set is part or all of the first resource set, the first resource set includes at least one of frequency-domain resources or spatial-domain resources, the CSI feedback information is obtained by compressing the first CSI or a third CSI, the third CSI represents the channel state of a third resource set, the third resource set is part of the first resource set, or the third resource set is different from the first resource set. Among them, indicating the first data set can be replaced by sending the first data set, or sending a second data set and a third data set, where the second data set and the third data set are used for obtaining the first data set.

[0026] Based on the above solution, indicating the first data set from the network side to the terminal side, where each sample of the first data set includes multiple CSIs (such as the first CSI and the second CSI) and CSI feedback information, can enable the terminal side to train a model for CSI compression for the CSI to be fed back based on the first data set, and a UE-side model that can be docked with the network-side model can be trained, which is beneficial to improving the feedback performance of CSI.

[0027] Combined with the second aspect, in some implementation manners of the first aspect, the first resource set corresponds to a first time-domain resource, the second resource set corresponds to a second time-domain resource and / or the first time-domain resource, the second time-domain resource is earlier than the first time-domain resource, and the third resource set corresponds to the first time-domain resource.

[0028] Combined with the second aspect, in some implementation manners of the first aspect, the second resource set includes a first part and a second part, the first part corresponds to the second time-domain resource, and the second part corresponds to the first time-domain resource.

[0029] Combined with the second aspect, in some implementation manners of the first aspect, send the first data set.

[0030] Combined with the second aspect, in some implementation manners of the first aspect, send a second data set and a third data set, where the second data set and the third data set are used to determine the first data set, the second data set includes multiple first CSIs, the multiple first CSIs correspond to multiple first time-domain resources, the multiple first time-domain resources do not overlap, the third data set includes multiple CSI feedback information, and one of the multiple first CSIs corresponds to one or more of the multiple CSI feedback information.

[0031] Combined with the second aspect, in some implementation manners of the first aspect, the first CSI included in each sample of the first data set is one of the multiple first CSIs, the second CSI included in each sample is determined according to the first CSI included in each sample and the other first CSIs among the multiple first CSIs, and the CSI feedback information included in each sample is the CSI feedback information corresponding to the first CSI included in each sample among the multiple CSI feedback information.

[0032] Combined with the second aspect, in some implementation manners of the first aspect, the second CSI included in each sample is determined by partial CSIs of the first CSI included in each sample and partial CSIs of the other first CSIs.

[0033] Combined with the second aspect, in some implementation manners of the first aspect, the time-domain resource corresponding to the other first CSI is earlier than the time-domain resource corresponding to the first CSI included in each sample.

[0034] In a third aspect, a device is provided. The device may include functional modules corresponding one by one to the methods / operations / steps / actions described in any possible implementation manner of the first aspect, or include functional modules corresponding one by one to the methods / operations / steps / actions described in any aspect of the second aspect. The module may be a hardware circuit, software, or a combination of a hardware circuit and software.

[0035] In one design, the device may include a processing module and a communication module. Among them, the communication module is used to perform the sending and receiving actions executed by the terminal side in the method described in the first aspect above, and the processing module is used to perform the actions related to processing executed by the terminal side in the method described in the first aspect above.

[0036] In one design, the device may be a terminal device, or a device, module, circuit, or chip configured to be disposed in a terminal device, or a device that can be used in combination with a terminal device, such as the host of an over-the-top (OTT) or a cloud server.

[0037] In one design, the device may include a processing module and a communication module. Among them, the communication module is used to perform the sending and receiving actions executed by the network side in the method described in the second aspect above, and the processing module is used to perform the actions related to processing executed by the network side in the method described in the second aspect above.

[0038] In one design, the device may be a network device, or a device, module, circuit, or chip configured to be disposed in a network device, or a device that can be used in combination with a network device, such as an intelligent network element deployed with a radio access network (RAN) intelligent controller (RIC).

[0039] In a fourth aspect, a device is provided, including a processor and a storage medium. The storage medium stores instructions, and when the instructions are run by the processor, the methods in the first aspect or any possible implementation manner of the first aspect are implemented, or the methods in the second aspect or any possible implementation manner of the second aspect are implemented.

[0040] In a fifth aspect, a device is provided, including a processing circuit. The processing circuit is used to process data and / or information so that the methods in the first aspect or any possible implementation manner of the first aspect are implemented, or the methods in the second aspect or any possible implementation manner of the second aspect are implemented.

[0041] The processing circuit may include one or more processors, or all or part of the circuits in one or more processors for control or processing functions.

[0042] Optionally, the device may further include a memory for storing programs or instructions, and a processor for running the programs or instructions, so that the method in the first aspect or any possible implementation manner of the first aspect is implemented, or the method in the second aspect or any possible implementation manner of the second aspect is implemented.

[0043] Optionally, the device may further include the transceiver circuit or an input / output interface.

[0044] In a sixth aspect, a chip is provided, including a processing circuit for running programs or instructions, so that the method in the first aspect or any possible implementation manner of the first aspect is implemented, or the method in the second aspect or any possible implementation manner of the second aspect is implemented.

[0045] Optionally, the chip may further include a memory for storing programs or instructions.

[0046] Optionally, the chip may further include a transceiver circuit or an input / output interface.

[0047] In a seventh aspect, a computer-readable storage medium is provided, where the computer-readable storage medium includes instructions, and when the instructions are run by a processor, the method in the first aspect or any possible implementation manner of the first aspect is implemented, or the method in the second aspect or any possible implementation manner of the second aspect is implemented.

[0048] In an eighth aspect, a computer program product is provided, where the computer program product includes computer program code or instructions, and when the computer program code or instructions are run, the method in the first aspect and any possible implementation manner of the first aspect is implemented, or the method in the second aspect or any possible implementation manner of the second aspect is implemented.

[0049] In a ninth aspect, a communication system is provided, where the communication system includes a device implementing the method in the first aspect and any possible implementation manner of the first aspect, or includes a device implementing the method in the second aspect and any possible implementation manner of the second aspect.

[0050] It should be understood that the third aspect to the ninth aspect of this application correspond to the technical solutions of the first aspect to the second aspect of this application, and the beneficial effects obtained by each aspect and the corresponding feasible implementation manners are similar and will not be elaborated herein. Description of the Drawings

[0051] Figure 1 is a schematic diagram of a communication system applicable to the communication method of the embodiments of this application.

[0052] Figure 2 It is a schematic diagram of another communication system applicable to the communication method of the embodiments of the present application.

[0053] Figure 3 It is a schematic diagram of a possible application framework in the communication system.

[0054] Figure 4 It is a schematic diagram of another possible application framework in the communication system.

[0055] Figure 5 It is a schematic diagram of frequency-hopping transmitting SRS.

[0056] Figure 6 It is a schematic diagram of using the AE model for CSI feedback.

[0057] Figure 7 It is an example of the neuron structure.

[0058] Figure 8 It is an example of the structure of a deep neural network (DNN).

[0059] Figure 9 It is a schematic diagram of the network side docking with the terminal side for the dataset.

[0060] Figure 10 It is a schematic diagram of the network side docking with the terminal side for the model.

[0061] Figure 11 It is a schematic flowchart of the communication method provided by the present application.

[0062] Figure 12 It is a schematic diagram of the process of the network side performing joint training.

[0063] Figure 13 It is a schematic diagram of the network side determining a sample in the first dataset.

[0064] Figure 14 It is a schematic diagram of the terminal side determining the first dataset.

[0065] Figure 15 It is a schematic diagram of the process of the terminal side performing model training.

[0066] Figure 16 and Figure 17 are schematic block diagrams of the communication device provided by the embodiments of the present application;

[0067] Figure 18 is a possible schematic block diagram of the AI processor provided by the embodiments of the present application;

[0068] Figure 19It is a schematic diagram showing the application of a dual - end model encoder (Encoder) and decoder (Decoder) in a channel information processing scenario provided by an embodiment of the present application. Detailed implementation manners

[0069] Next, the technical solutions in the present application will be described in conjunction with the accompanying drawings.

[0070] To facilitate the understanding of the embodiments of the present application, the following explanations are made first:

[0071] First, in the present application, the terminal side can also be referred to as the UE (user equipment, UE) side (UE side), including: terminal devices, components deployed in terminal devices (such as circuits or chips inside terminal devices), devices deployed outside terminal devices (such as over - the - top (OTT) hosts or cloud servers), or components in devices deployed outside the terminal devices (such as circuits or chips inside the devices). The network (NW) side (NW side) includes: network devices communicating with terminal devices, components deployed in network devices (such as circuits or chips with near - real - time RAN intelligent control functions inside network devices), devices deployed outside network devices (such as intelligent network elements, for example, the intelligent network element has near - real - time RAN intelligent control functions), or components in the intelligent network elements (such as circuits or chips inside the intelligent network elements). Among them, network devices can include: access network devices, core network devices, or operation administration and maintenance (OAM).

[0072] Second, in the present application, indication includes direct indication (also called explicit indication) and indirect indication (also called implicit indication). Among them, directly indicating information A means including the information A; indirectly indicating information A can mean indicating information A through the correspondence between information A and information B and directly indicating information B; or indicating information A through a preset rule that can be used to determine A based on B and directly indicating information B. Among them, the correspondence between information A and information B, and the preset rule can be predefined, pre - stored, pre - burned, or pre - configured.

[0073] Third, in this application, "at least one" means one or more, and "a plurality" means two or more. "And / or" describes the relationship between related objects and indicates that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, or B exists alone, where A and B can be singular or plural. The character " / " generally indicates an "or" relationship between the related objects before and after, but it does not exclude the case where it represents an "and" relationship between the related objects before and after. The specific meaning can be understood in combination with the context. "At least one (item)" or its similar expressions refer to any combination of these items, including any combination of single item(s) or multiple item(s). For example, at least one (item) of a, b, or c can mean: a, b, c; a and b; a and c; b and c; or a, b, and c. Where a, b, and c can be single or multiple.

[0074] Fourth, in this application, the use of prefix words such as "first" and "second" is only for facilitating the differential description of different things belonging to the same name category, and does not restrict the order, size, or quantity of things. For example, "first information" and "second information" are just different information, and do not limit the quantity of information, the order of transmission, or the priority relationship.

[0075] Fifth, in this application, "send" and "receive" represent the direction of signal transmission. For example, "send information to XX" can be understood as the destination of this information is XX, which can include directly sending through the air interface, and also include indirectly sending by other units or modules through the air interface. "Receive information from YY" can be understood as the source of this information is YY, which can include directly receiving from YY through the air interface, and can also include indirectly receiving from YY through the air interface by other units or modules. "Send" can also be understood as the "output" of the chip interface, and "receive" can also be understood as the "input" of the chip interface. In other words, sending and receiving can be carried out between devices, for example, between a terminal device and a computing node, or can be carried out within a device, for example, sending or receiving between components, modules, chips, software modules, or hardware modules within a device through a bus, trace, or interface.

[0076] Sixth, in the embodiments of this application, "when...", "if", and "in case" all refer to the device will make corresponding processing under a certain objective situation, not to limit time, and do not require the device to have a judgment action when implemented, nor does it mean there are other limitations.

[0077] Seventh, in this application, words such as "example", "exemplarily", "for example", or "such as" are used to represent examples, illustrations, or explanations. Any embodiment or design solution described as "example", "exemplarily", "for example", or "such as" in this application should not be construed as more preferred or advantageous than other embodiments or design solutions. Rather, the use of words such as "example", "exemplarily", "for example", or "such as" is intended to present the relevant concepts in a specific manner.

[0078] The technical solutions provided in this application can be applied to various communication systems, such as: the fifth generation (5G) or new radio (NR) system, the long term evolution (LTE) system, the LTE frequency division duplex (FDD) system, the LTE time division duplex (TDD) system, the wireless local area network (WLAN) system, the satellite communication system, future communication systems, or a fusion system of multiple systems, etc. 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 the Internet of Things (IoT) communication system or other communication systems.

[0079] A network element in a communication system can send a signal to another network element or receive a signal from another network element. Among them, the signal can include information, signaling, or data, etc. Among them, the network element can also be replaced by an entity, a network entity, a device, a communication device, a communication module, a node, a communication node, etc. In this disclosure, the network element is used as an example for description. For example, a communication system can include at least one terminal device and at least one network device. The network device can send a downlink signal to the terminal device, and / or the terminal device can send an uplink signal to the network device. It can be understood that the terminal device in this disclosure can be replaced by a first network element, and the network device can be replaced by a second network element, and the two execute the corresponding communication methods in this disclosure.

[0080] Figure 1 is a schematic diagram of a communication system applicable to the communication method of the embodiments of this application. As Figure 1 shown, the communication system 100A can include at least one access network device, such as Figure 1The access network device 110 shown; The communication system 100A may further include at least one terminal device, such as Figure 1 the terminal device 120 and the terminal device 130 shown. The access network device 110 and the terminal devices (such as the terminal device 120 and the terminal device 130) can communicate through a wireless link. Between the communication devices in this communication system, for example, between the access network device 110 and the terminal device 120, communication can be carried out through multi-antenna technology.

[0081] In a wireless communication network, for example, in a mobile communication network, the services supported by the network are becoming more and more diverse, so the requirements to be met are becoming more and more diverse. For example, the network needs to be able to support ultra-high speed, ultra-low latency, and / or ultra-large connections. This feature makes network planning, network configuration, and / or resource scheduling more and more complex. In addition, due to the increasing powerful functions of the network, such as supporting higher and higher frequencies, supporting high-order multiple input multiple output (MIMO) technology, supporting beam forming (BF), supporting new technologies such as beam management, network energy saving has become a hot research topic. These new requirements, new scenarios, and new features have brought unprecedented challenges to network planning, operation and maintenance, and efficient operation. To meet this challenge, artificial intelligence technology can be introduced into the wireless communication network to achieve network intelligence. To support AI technology in the wireless network, AI nodes may also be introduced into the network.

[0082] Figure 2 is a schematic diagram of another communication system suitable for the communication method of the embodiments of the present application. Compared with Figure 1 the communication system 100A shown, Figure 2 the communication system 100B shown further includes an AI network element 140. The AI network element 140 is used to perform AI-related operations, for example, constructing a training data set or training an AI model, etc. Among them, the AI network element can also be simply referred to as an intelligent network element.

[0083] In a possible implementation, the access network device 110 may send data related to the training of the AI model to the AI network element 140. The AI network element 140 constructs a training data set and trains the AI model. For example, the data related to the training of the AI model may include the 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 access network device 110 and forward them to the terminal device through the access network device 110. For example, the results of the operations related to the AI model may include at least one of the following: the AI model that has completed training, the evaluation result of the model, or the test result, etc. Exemplarily, a part of the AI model that has completed training may be deployed on the access network device 110, and another part may be deployed on the terminal device 120 and / or the terminal device 130. Alternatively, the AI model that has completed training may be deployed on the access network device 110. Or, the AI model that has completed training may be deployed on the terminal device 120 and / or the terminal device 130.

[0084] It should be understood that Figure 2 only taking the example that the AI network element 140 is directly connected to the access network device 110 for illustration. In other scenarios, the AI network element 140 may also be connected to the terminal device. Or, the AI network element 140 may be connected to both the access network device 110 and the terminal device at the same time. Or, the AI network element 140 may also be connected to the access network device 110 through a third-party network element. The embodiments of the present application do not limit the connection relationship between the AI network element and other network elements. For example, the AI network element 140 may also be set as a module in the access network device and / or the terminal device. For example, it may be set in Figure 1 the access network device 110 or the terminal device shown.

[0085] It should be noted that Figure 1 and Figure 2 are only simplified schematic diagrams for easy understanding. For example, the communication system may also include other devices, such as wireless relay devices and / or wireless backhaul devices, etc. Figure 1 and Figure 2 are not shown. In actual applications, the communication system may include multiple access network devices or multiple terminal devices. The embodiments of the present application do not limit the number of access network devices and terminal devices included in the communication system.

[0086] In the embodiments of the present application, the terminal device may also be referred to as UE, access terminal, user unit, user station, mobile station, mobile platform, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent, or user device.

[0087] The terminal device can be a device that provides voice / data. For example, it can be a handheld device, a vehicle-mounted device, etc. with wireless connection capabilities. Currently, some examples of terminals are: mobile phone, tablet computer, laptop computer, palmtop computer, mobile internet device (MID), wearable device, virtual reality (VR) device, augmented reality (AR) device, wireless terminal in industrial control, wireless terminal in self-driving, wireless terminal in remote medical surgery, wireless terminal in smart grid, wireless terminal in transportation safety, wireless terminal in smart city, wireless terminal in smart home, cellular phone, cordless phone, session initiation protocol (SIP) phone, wireless local loop (WLL) station, personal digital assistant (PDA), handheld device with wireless communication function, computing device or other processing device connected to a wireless modem, wearable device, terminal device in a 5G network or terminal device in a future evolved public land mobile network (PLMN), etc. The embodiments of this application are not limited thereto.

[0088] By way of example and not limitation, in the embodiments of this application, the terminal device can also be a wearable device. A wearable device can also be referred to as a wearable intelligent device, which is a general term for devices developed by applying wearable technologies to intelligentize daily wear, such as glasses, gloves, watches, clothing, and shoes, etc. A wearable device is a portable device that is either directly worn on the body or integrated into the user's clothes or accessories. A wearable device is not just a hardware device, but more importantly, it realizes powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable intelligent devices include those with complete functions and large sizes that can realize complete or partial functions without relying on a smartphone, such as smart watches or smart glasses, etc., and those that only focus on a certain type of application function and need to cooperate with other devices such as smartphones, such as various smart bracelets and smart jewelry for monitoring physical signs.

[0089] In the embodiments of the present application, the device for implementing the functions of the terminal device may be the terminal device itself, or a device capable of supporting the terminal device to implement such functions, such as a chip system. This device may be installed in the terminal device or used in combination with the terminal device. In the embodiments of the present application, the chip system may be composed of chips, or may include chips and other discrete devices. In the embodiments of the present application, only the case where the device for implementing the functions of the terminal device is the terminal device is taken as an example for illustration, which does not limit the solutions of the embodiments of the present application.

[0090] The access network device in the embodiments of the present application may be a device for communicating with the terminal device. This access network device may also be referred to as a network device. For example, the access network device may be a base station. The access network device in the embodiments of the present application may refer to a RAN node (or device) that connects the terminal device to the wireless network. The base station may generally cover various names as follows, or be replaced with the following names, such as: Node B, evolved Node B (eNB), next generation Node B (gNB), relay station, access point, transmitting and receiving point (TRP), transmitting point (TP), master station, slave station, multi-mode radio (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 may 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 may also refer to a communication module, a modem or a chip disposed in the foregoing device or apparatus. The base station may also be a mobile switching center and a device that undertakes the base station function in D2D, V2X, M2M communications, and a device that undertakes the base station function in future communication systems, etc. The base station may support networks of the same or different access technologies. Optionally, the RAN node may also be a server, a wearable device, a vehicle or an in-vehicle device, etc. For example, the access network device in V2X technology may be a road side unit (RSU). The embodiments of the present application do not limit the specific technologies and specific device forms adopted by the access network device.

[0091] The base station can be fixed or mobile. For example, a helicopter or a drone can be configured to act as a mobile base station, and one or more cells can move according to the position of the mobile base station. In other examples, a helicopter or a drone can be configured to be used as a device for communicating with another base station.

[0092] In some deployments, the access network device mentioned in this application can be a device including a CU, or a DU, or a device including a CU and a DU, or a control plane CU node (Central Unit Control Plane (CU-CP)) and a user plane CU node (Central Unit User Plane (CU-UP)) and a DU node. For example, the access network device can include a gNB-CU-CP, a gNB-CU-UP, and a gNB-DU.

[0093] In some deployments, multiple RAN nodes cooperate to assist a terminal in achieving wireless access, and different RAN nodes respectively implement partial functions of a base station. For example, the RAN node can be a CU, a DU, a CU-CP, a CU-UP, or an RU, etc. The CU and the DU can be separately set, or can also be included in the same network element, such as a BBU. The RU can be included in a radio frequency device or a radio frequency unit, such as included in an RRU, an AAU, or an RRH.

[0094] The RAN node can support one or more types of fronthaul interfaces. For different fronthaul interfaces, the DUs and RUs with different functions are respectively corresponding. 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 of the baseband functions, and the RU is configured to implement one or more of the radio frequency functions. If the fronthaul interface between the DU and the RU is another interface, compared with the CPRI, some of the downlink and / or uplink baseband functions, for example, for the downlink, one or more of precoding, digital beamforming, or inverse fast Fourier transform (IFFT) / cyclic prefix (CP) addition, are moved from the DU to the RU for implementation; for the uplink, one or more of digital beamforming, or fast Fourier transform (FFT) / cyclic prefix (CP) removal, are moved from the DU to the RU for implementation. In a possible implementation, this interface can be the Enhanced Common Public Radio Interface (eCPRI). Under the eCPRI architecture, different splitting methods between the DU and the RU correspond to different categories (Cat) of eCPRI, such as eCPRI Cat A, B, C, D, E, F.

[0095] Taking eCPRI Cat A as an example, for downlink transmission, with layer mapping as the splitting point, the DU is configured to implement one or more functions before layer mapping (i.e., one or more of coding, rate matching, scrambling, modulation, layer mapping), and other functions after layer mapping (such as one or more of resource element (RE) mapping, digital beamforming, or IFFT / CP addition) are moved to the RU for implementation. For uplink transmission, with RE demapping as the splitting point, the DU is configured to implement one or more functions before demapping (i.e., one or more of decoding, rate dematching, descrambling, demodulation, inverse discrete Fourier transform (IDFT), channel equalization, RE demapping), and other functions after demapping (such as one or more of digital BF or FFT / CP removal) are moved to the RU for implementation. It can be understood that for the function descriptions of the DUs and RUs corresponding to various types of eCPRI, reference can be made to the eCPRI protocol, which will not be elaborated here.

[0096] In a possible design, the processing unit used to implement the baseband function in the BBU is called the baseband high (BBH) unit, and the processing unit used to implement the baseband function in the RRU / AAU / RRH is called the baseband low (BBL) unit.

[0097] In different systems, the CU (or CU-CP and CU-UP), DU, or RU may also have different names, but those skilled in the art can understand their meanings. For example, in the open RAN (ORAN) architecture, the CU may also be called the open-CU (O-CU), the DU may also be called the open-DU (O-DU), the CU-CP may also be called the open-CU-CP (O-CU-CP), the CU-UP may also be called the open-CU-UP (O-CU-UP), and the RU may also be called the open-RU (O-RU). Any one 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 a software module and a hardware module.

[0098] In the embodiments of this application, the device used to implement the functions of the network device may be the network device; it may also be a device capable of supporting the network device to implement this function, such as a chip system, a hardware circuit, a software module, or a combination of a hardware circuit and a software module. This device may be installed in the network device or used in matching with the network device. In the embodiments of this application, only the case where the device used to implement the functions of the network device is the network device is taken as an example for illustration, which does not limit the solutions of the embodiments of this application.

[0099] The network device and / or the terminal device may be deployed on land, including indoors or outdoors, handheld or vehicle-mounted; they may also be deployed on the water surface; they may also be deployed on airplanes, balloons, and satellites in the air. In the embodiments of this application, the scenarios where the network device and the terminal device are located are not limited. In addition, the terminal device and the network device may be hardware devices, or software functions running on dedicated hardware or software functions running on general hardware. For example, they are virtualized functions instantiated on a platform (such as a cloud platform), or entities including dedicated or general hardware devices and software functions. This application does not limit the specific forms of the terminal device and the network device.

[0100] Optionally, the AI node can be deployed at one or more of the following locations in the communication system: access network devices, terminal devices, or network elements of the core network, etc. Optionally, the AI node can also be deployed independently. For example, it can be deployed at a location outside any of the above devices, such as in the host of an OTT system or a cloud server. The AI node can communicate with other devices in the communication system, and the other devices can be, for example, one or more of the following: access network devices, terminal devices, or network elements of the core network, etc.

[0101] It can be understood that the present application does not limit the number of AI nodes. For example, when there are multiple AI nodes, the multiple AI nodes can be divided based on functions. For example, different AI nodes are responsible for different functions.

[0102] It can also be understood that the AI node can be an independent device, or can be integrated into the same device to implement different functions, or can be a network element in a hardware device, or can be a software function running on dedicated hardware, or can be a virtualized function instantiated on a platform (such as a cloud platform). The present application does not limit the specific form of the above AI node.

[0103] The AI node can be an AI network element or an AI module.

[0104] Figure 3 It is a schematic diagram of a possible application framework in the communication system. As Figure 3 shown, the network elements in the communication system are connected through interfaces (such as the next generation (NG) interface, Xn interface) or the air interface. Among them, the NG interface is the interface between the radio access network and the 5G core network. The Xn interface is the interface between access network devices, and the air interface is the interface between an access network device and a terminal device. One or more AI modules are provided in one or more of these network element nodes, such as core network devices, access network nodes (RAN nodes), terminals, or OAM. (For clarity, Figure 3 only 1 is shown). The access network node can be a separate RAN node or can include multiple RAN nodes. For example, it includes RU, CU, and DU. One or more AI modules can also be provided in one or more of the RU, CU, or DU. Optionally, the CU can also be split into CU-CP and CU-UP. One or more AI modules are provided in CU-CP and / or CU-UP.

[0105] The AI module is used to implement corresponding AI functions. The AI modules deployed in different network elements can be the same or different. Based on different parameter configurations of the model of the AI module, the AI module can implement different functions. The model of the AI module can be based on one or more of the following parameter configurations: structural parameters (such as at least one of the number of neural network layers, the width of the neural network, the connection relationship between layers, the weights of neurons, the activation function of neurons, or the bias in the activation function), input parameters (such as the type and / or dimension of the input parameters), or output parameters (such as the type and / or dimension of the output parameters). Among them, the bias in the activation function can also be referred to as the bias of the neural network.

[0106] An AI module can have one or more models. One model can infer an output, and the output includes one parameter or multiple parameters. The learning process, training process, or inference process of different models can be deployed in different nodes or devices, or can be deployed in the same node or device.

[0107] The network device can be a network device provided with one or more AI modules. The network device can include Figure 3 one or more of the core network device, access network node (RAN node), or OAM shown. For example, the AI module can be Figure 4 the RAN intelligent controller (RIC) shown, such as near-realtime RIC (near-RT RIC) or non-real time RIC (Non-RT RIC), etc. For example, the near-realtime RIC is set in the RAN node (such as in the CU or DU), and the non-real time RIC is set in the OAM, cloud server, core network device, or other access network devices. The RIC can obtain a subset from multiple terminal devices from the RAN node (such as CU, CU-CP, CU-UP, DU, and / or RU), reorganize it into a training data set, and perform training based on the training data set. Exemplarily, the near-realtime RIC and non-real time RIC can also be set separately as a network element, and the access network device can be the near-realtime RIC or non-real time RIC.

[0108] Figure 4 It is a schematic diagram of another possible application framework in the communication system. Figure 4 In the communication system shown, in addition to including access network nodes (CU, DU, and RU are shown in the figure) and terminals, it also includes RIC. For example, the RIC can be Figure 3The AI module shown can be used to implement AI-related functions. The RIC includes a near-real-time RIC and a non-real-time RIC. Among them, the non-real-time RIC mainly processes non-real-time information, such as data that is not sensitive to latency, and the latency of this data can be in seconds. The real-time RIC mainly processes near-real-time information, such as data that is relatively sensitive to latency, and the latency of this data is in the order of tens of milliseconds.

[0109] The near-real-time RIC is used for model training and inference. For example, it is used to train an AI model and perform inference using this AI model. The near-real-time RIC can obtain network-side and / or terminal-side information from RAN nodes (such as CU, CU-CP, CU-UP, DU, and / or RU) and / or terminals. This information can be used as training data or inference data. Optionally, the near-real-time RIC can submit the inference result to the RAN node and / or the terminal. Optionally, the inference result can be exchanged between the CU and the DU, and / or between the DU and the RU. For example, the near-real-time RIC submits the inference result to the DU, and the DU sends it to the RU.

[0110] The non-real-time RIC is also used for model training and inference. For example, it is used to train an AI model and perform inference using this model. The non-real-time RIC can obtain network-side and / or terminal-side information from RAN nodes (such as one or more of CU, CU-CP, CU-UP, DU, or RU) and / or terminals. This information can be used as training data or inference data, and the inference result can be submitted to the RAN node and / or the terminal. Optionally, the inference result can be exchanged between the CU and the DU, and / or between the DU and the RU. For example, the non-real-time RIC submits the inference result to the DU, and the DU sends it to the RU.

[0111] The near-real-time RIC and the non-real-time RIC can also be separately set as a network element. Optionally, the near-real-time RIC and the non-real-time RIC can also be part of other devices. For example, the near-real-time RIC is set in a RAN node (such as in CU or DU), and the non-real-time RIC is set in the OAM, cloud server, core network device, or other network devices.

[0112] With the development of wireless communication technology, the supported services are continuously increasing, and higher requirements are put forward for communication systems in terms of system capacity, communication latency and other indicators. Among them, a massive multiple-input multiple-output (MIMO) system can achieve diversity gain in the spatial domain by configuring a large-scale antenna array at the transceiver end, thereby significantly increasing the system capacity. For example, an access network device can use the same time-frequency resources to simultaneously send data to multiple terminal devices, that is, multi-user MIMO (MU-MIMO); or, the access network device can also simultaneously send multiple data streams to the same terminal device, that is, single-user MIMO (SU-MIMO). The data between the multiple terminal devices or the multiple data streams of the same terminal device are space-division multiplexed, so it has become a key direction in the evolution of communication systems.

[0113] The access network device can obtain the channel state information (CSI) of the downlink channel, which is used to determine one or more of the resources of the downlink data channel for scheduling the terminal device, the modulation and coding scheme (MCS), precoding, and other configurations.

[0114] Taking precoding as an example, in massive MIMO, the access network device needs to precode the downlink data using a precoding matrix. By using precoding technology, the access network device can achieve space-division multiplexing between terminal devices or data streams, that is, isolate the data between different terminal devices or the data between different data streams of the same terminal device in space, so as to reduce the interference between different terminal devices or different data streams and improve the received signal-to-interference plus noise ratio (SINR) of the terminal device. To calculate the precoding matrix, the access network device needs to obtain the CSI of the downlink channel and determine the precoding matrix according to the CSI.

[0115] In a TDD system, due to the reciprocity of the uplink and downlink channels, the access network device can obtain the uplink CSI by measuring the uplink reference signal, and then infer a relatively accurate downlink CSI. For example, the uplink CSI is used as the downlink CSI. However, in an FDD system, the uplink-downlink reciprocity cannot be guaranteed, and the downlink CSI is obtained by the terminal device measuring the downlink reference signal. For example, the terminal device measures the channel state information reference signal (CSI-RS) or the synchronizing signal block (SSB) to obtain the downlink CSI. Therefore, the terminal device needs to generate a CSI report in a manner predefined by the protocol or configured by the access network device, and feedback the generated CSI report to the access network device so that it can obtain the downlink CSI.

[0116] In an FDD system, an important part of CSI feedback is the precoding matrix indicator (PMI), that is, 0-1 bits are used in the CSI to quantify the channel matrix or the precoding matrix. The design of the PMI (which can also be called codebook design) is a basic problem in a mobile communication system. The traditional codebook design method is to predefine (stipulate) a series of precoding matrices and corresponding numbers in the protocol. These precoding matrices are called codewords. The channel matrix or the precoding matrix can be approximated by using a predefined codeword or a linear combination of multiple predefined codewords. Therefore, the terminal device can feedback one or more of the corresponding numbers of the codewords and the weighting coefficients to the access network device through the PMI for the access network device to reconstruct the channel matrix or the precoding matrix.

[0117] As the scale of the antenna array in the MIMO system continues to increase, the number of supported antenna ports increases, and the dimensions of the corresponding channel matrix and precoding matrix grow. To enable the terminal device to estimate (or measure) the downlink channel, the overhead of the access network device for transmitting reference signals increases. At the same time, the error of approximately representing the large-scale channel matrix and precoding matrix with a limited number of predefined codewords will increase. One method to improve the channel reconstruction accuracy is to increase the number of codewords in the codebook, but this will simultaneously increase the overhead of CSI feedback (including one or more of the corresponding numbers of the codewords and the weighting coefficients), thereby reducing the available resources for data transmission and causing a loss of system capacity. In summary, it is necessary to study how to more effectively compress and represent channel information and how to more effectively reconstruct the channel based on the feedback information without increasing the overhead of transmitting reference signals and the CSI feedback overhead. There is a correlation between different elements in the downlink channel matrix between the access network device and the terminal device. In addition, there is a correlation between the downlink channel matrices in different time slots. For example, the correlation between different elements in the channel matrix means that there is a set of bases (which can be represented by matrices U1 and U2), and when the channel matrix H is projected onto this set of bases, a sparse equivalent channel can be obtained, that is, H' = U1H H U2 is a sparse matrix, where the superscript H represents the conjugate transpose operation. In theory, only the non-zero elements in H' need to be estimated and fed back through the transmission of reference signals to reconstruct the channel matrix H. Therefore, there is room for compression in the transmission of reference signals and the overhead of CSI feedback. However, in traditional CSI feedback schemes, such as the above-mentioned codebook-based feedback method, etc., the channel compression space is not fully utilized, and significant information loss may occur during the channel compression process. Machine learning (such as deep learning (DL)) methods have stronger non-linear feature extraction capabilities. Therefore, they can more effectively extract the correlation between channel matrices, and thus can more effectively compress and represent channel information and more effectively reconstruct channel information based on the feedback information compared to traditional schemes.

[0118] In some implementation manners, a neural network model can be used for the compressed feedback of channel information. After the terminal measures the reference signal to obtain channel information, the channel information is input into the neural network model of the terminal to obtain the compressed information of the channel information. The terminal feeds back the compressed information of the channel information to the network device. The network device inputs the compressed information of the channel information into the neural network model of the network device to recover the channel information measured by the terminal. In this way, the non-linear feature extraction ability of the neural network can be utilized to improve the accuracy of channel measurement.

[0119] As an example, such as Figure 19As shown in the figure, it is a schematic diagram of applying a dual - end model encoder and decoder to the channel information processing scenario. Among them, the model of the terminal is the encoder, which is used to compress CSI - RS measurement data, and the terminal sends the compressed CSI - RS measurement data to the network device. After receiving the compressed CSI - RS measurement data, the network device restores the CSI - RS measurement data based on the decoder in the network device to obtain the CSI - RS measurement data. In this way, between the terminal and the network device, large - port and full - band channel measurements can be realized with limited CSI - RS measurement overhead based on the dual - end model. When performing CSI - RS measurements, multi - port channel measurements can be achieved by indicating multiple resource sets. Currently, 1 resource set supports channel measurements of up to 32 antenna ports at most. By increasing multiple resource sets, channel measurements for a larger array can be supported. In the related art, to support large - bandwidth channel measurements, the number of resource blocks (RBs) that the terminal needs to measure can be indicated, and the number of channels measured within 1 RB is indicated by density, and the value can be 0.5, 1, or 3.

[0120] In the embodiments of this application, combined with Figure 19 the CSI - RS measurement scenario shown in the figure, the terminal can measure only part of the CSI - RS and compress the measurement data of this part of the CSI - RS; correspondingly, the network device restores the measurement data of this part of the compressed CSI - RS to obtain the measurement data of this part of the CSI - RS. Or, the terminal can measure only part of the CSI - RS and compress the measurement data of this part of the CSI - RS; however, the network device not only restores the measurement data of this part of the compressed CSI - RS to obtain the measurement data of this part of the CSI - RS, but the network device can also predict the unmeasured CSI - RS measurement data to obtain the full - volume CSI - RS measurement data. Or, the terminal can measure only part of the CSI - RS and compress the measurement data of this part of the CSI - RS; after the terminal sends the compressed measurement data of this part of the CSI - RS multiple times, the network device predicts the full - volume CSI - RS measurement data based on the compressed measurement data of this part of the CSI - RS sent multiple times.

[0121] To facilitate the understanding of the embodiments of this application, the following briefly explains the terms involved in this application.

[0122] 1. Artificial Intelligence: It can refer to enabling machines to have learning capabilities, accumulate experience, and solve problems that humans can solve through experience, such as natural language understanding, image recognition, and chess playing. Artificial intelligence can be understood as the intelligence demonstrated by machines made by humans. Usually, artificial intelligence refers to the technology that presents human intelligence through computer programs. The goals of artificial intelligence include understanding intelligence by constructing computer programs with symbolic reasoning or inference.

[0123] 2. Machine Learning (ML): It is an implementation method of artificial intelligence. Machine learning is a method that can endow machines with learning capabilities, enabling machines to complete functions that cannot be achieved by direct programming. In a practical sense, machine learning is a method of using data to train a model and then using the model for prediction. There are many machine learning methods, such as neural network (NN), decision tree, support vector machine (SVM), etc. Machine learning theory mainly designs and analyzes algorithms that allow computers to learn automatically. Machine learning algorithms are a class of algorithms that automatically analyze and obtain rules from data and use the rules to predict unknown data. Machine learning can be divided into supervised learning, unsupervised learning, and reinforcement learning.

[0124] Supervised learning is based on the collected sample values and sample labels, uses machine learning algorithms to learn the mapping relationship from sample values to sample labels, and uses a machine learning model to represent the learned mapping relationship. The process of training a machine learning model is the process of learning this mapping relationship. For example, in signal detection, the received signal with noise is the sample, and the corresponding true constellation point of the signal is the label. Machine learning expects to learn the mapping relationship between the sample and the label through training, that is, to enable the machine learning model to learn a signal detector. During training, the model parameters are optimized by calculating the error between the predicted value of the model and the true label. Once the mapping relationship is learned, the learned mapping can be used to predict the sample label of each new sample. The mapping relationship learned by supervised learning can include linear mapping and non - linear mapping. According to the type of label, the learning tasks can be divided into classification tasks and regression tasks.

[0125] Unsupervised learning only relies on the collected sample values and uses algorithms to discover the internal patterns of the samples by itself. In unsupervised learning, there is a class of algorithms that use the samples themselves as the supervision signal, that is, the model learns the mapping relationship from samples to samples, which is called self - supervised learning. During training, the model parameters are optimized by calculating the error between the predicted value of the model and the samples themselves. Self - supervised learning can be used in applications such as signal compression and decompression recovery. Common algorithms include autoencoders and adversarial generative networks, etc.

[0126] Reinforcement learning is different from supervised learning. It is a type of algorithm that learns strategies to solve problems by interacting with the environment. Different from supervised and unsupervised learning, there is no clear "correct" action label data in reinforcement learning problems. The algorithm needs to interact with the environment to obtain the reward signal feedback from the environment, and then adjust the decision-making actions to obtain a larger reward signal value. For example, in downlink power control, the reinforcement learning model adjusts the downlink transmission power of each user according to the total system throughput rate feedback by the wireless network, and then expects to obtain a higher system throughput rate. The goal of reinforcement learning is also to learn the mapping relationship between the environmental state and the optimal / better decision-making actions. However, because the label of the "correct action" cannot be obtained in advance, the network cannot be optimized by calculating the error between the action and the "correct action". The training of reinforcement learning is achieved through iterative interaction with the environment.

[0127] 3. AI Model: It can be an algorithm or a computer program that can implement AI functions. The AI model represents the mapping relationship between the input and output of the model. The AI model can be understood as a function model that maps the input of a certain dimension to the output 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. The AI model can also be called a model or an AI function or a function. An AI function can correspond to one or more AI models.

[0128] The types of AI models can be neural networks, linear regression models, decision tree models, SVM, Bayesian networks, Q-learning models, or other machine learning models.

[0129] The implementation of the AI model can be a hardware circuit, software, or a combination of software and hardware, without limitation. Non-limiting examples of software include: program code, program, subroutine, instruction, instruction set, code, code segment, software module, application program, or software application, etc.

[0130] 4. Neural Network (NN): It is a specific implementation form of AI or machine learning. According to the universal approximation theorem, a neural network can theoretically approximate any continuous function, enabling the neural network to have the ability to learn any mapping.

[0131] A neural network can be composed of neural units, and a neural unit can refer to x sAn arithmetic unit that takes the intercept 1 as the input. A neural network is a network formed by connecting many such single neural units together, that is, the output of one neural unit can be the input of another neural unit. The input of each neural unit can be connected to the local receptive field of the previous layer to extract the features of the local receptive field, and the local receptive field can be a region composed of several neural units.

[0132] Taking the type of AI model as a neural network as an example, the AI model involved in this application can be a deep neural network (DNN). The idea of DNN comes from the neuron structure of brain tissue. Each neuron performs a weighted sum operation on its respective input value and generates an output through a non-linear function. DNN generally has a multi-layer structure. Each layer of DNN can contain multiple neurons. After the input layer processes the received numerical values through neurons, it transmits them to the intermediate hidden layer. Similarly, the hidden layer then transmits the calculation results to the final output layer to generate the final output of DNN. DNN generally has more than one hidden layer, and the hidden layer often directly affects the ability to extract information and fit functions. Increasing the number of hidden layers of DNN or expanding the width of each layer can improve the function fitting ability of DNN. The weighted values in each neuron are the parameters of the DNN network model. The model parameters are optimized through the training process, so that the DNN network has the ability to extract data features and express mapping relationships.

[0133] According to the construction method of the network, DNN can include feedforward neural network (FNN), convolutional neural networks (CNN), and recurrent neural network (RNN), etc.

[0134] CNN is a neural network specifically designed to process data with a similar grid structure. For example, time series data (discrete sampling on the time axis) and image data (two-dimensional discrete sampling) can both be considered data with a similar grid structure. CNN does not perform operations using all the input information at once, but instead uses a fixed-size window to intercept part of the information for convolution operations, which greatly reduces the computational amount of model parameters. In addition, according to the different types of information intercepted by the window (such as people and objects in the same picture being different types of information), each window can use different convolution kernels for operations, which enables CNN to better extract the features of the input data.

[0135] RNN is a type of DNN network that utilizes feedback time series information. Its input includes the new input value at the current moment and its own output value at the previous moment. RNN is suitable for obtaining sequence features that are relevant in time and is particularly applicable to applications such as speech recognition and channel coding and decoding.

[0136] The characteristic of the FNN network is that neurons between adjacent layers are fully connected pairwise, which makes FNN usually require a large amount of storage space and results in a high computational complexity.

[0137] The above-mentioned FNN, CNN, and RNN are all constructed based on neurons. As mentioned before, each neuron performs a weighted sum operation on its respective input value and produces an output through a non-linear function. The weights of the weighted sum operation of neurons in a neural network and the non-linear function are called the parameters of the neural network. The parameters of all neurons in a neural network constitute the parameters of this neural network.

[0138] 5. Dataset: It can refer to the data used for model training, validation, and testing in machine learning. The quantity and quality of the data will affect the effect of machine learning.

[0139] In the field of machine learning, ground truth usually refers to the data that is considered accurate or real.

[0140] The training dataset can be used for the training of AI models. The training dataset can include the input of the AI model, or include the input and target output of the AI model. Among them, the training dataset includes one or more training data, and the training data can include the training samples input to the AI model or the target output of the AI model. Among them, the target output can also be called a label, sample label, or labeled sample. The label is the ground truth.

[0141] In the field of communication, the training dataset can include simulation data collected through a simulation platform, or experimental data collected in an experimental scenario, or, can also include the measured data collected in an actual communication network. Due to differences in the geographical environment and channel conditions where the data is generated, for example, differences in indoor, outdoor, moving speed, frequency band, or antenna configuration, etc., when obtaining the data, the collected data can be classified. For example, the data with the same channel propagation environment and antenna configuration is classified into one category.

[0142] Model training essentially involves learning certain features from the training data. During the process of training an AI model (such as a neural network model), since we hope the output of the AI model is as close as possible to the value we truly want to predict, we can compare the predicted value of the current network with the true target value, and then update the weight vector of each layer of the AI model according to the difference between the two. (Of course, there is usually an initialization process before the first update, that is, pre-configuring parameters for each layer in the AI model.) For example, if the predicted value of the network is too high, we adjust the weight vector to make it predict lower, and keep adjusting until the AI model can predict the true target value or a value very close to the true target value. Therefore, it is necessary to pre-define "how to compare the difference between the predicted value and the target value", which is the loss function or the objective function. They are important equations for measuring the difference between the predicted value and the target value. Among them, taking the loss function as an example, the higher the output value (loss) of the loss function, the greater the difference. Then the training of the AI model becomes a process of minimizing this loss as much as possible, making the value of the loss function less than the threshold, or making the value of the loss function meet the target requirements. For example, assuming the AI model is a neural network, adjusting the model parameters of the neural network includes adjusting at least one of the following parameters: the number of layers of the neural network, the width, the weights of the neurons, or the parameters in the activation function of the neurons.

[0143] The inference data can be used as the input of the trained AI model for inference of the AI model. During the model inference process, when the inference data is input into the AI model, the corresponding output can be obtained, which is the inference result.

[0144] 6. CSI feedback: In LTE and NR communication systems, network devices need to obtain downlink CSI to determine resource allocation for scheduling the downlink data channel of the terminal device, modulation and coding scheme (MCS), precoding, and other configurations. It can be understood that CSI belongs to a type of channel information that can reflect channel characteristics and channel quality.

[0145] CSI measurement may refer to the receiver solving the channel information based on the reference signal sent by the transmitter, that is, estimating the channel information by using the channel estimation method. Exemplarily, 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), etc. CSI-RS, SSB, and DMRS, etc. can be used to measure the downlink CSI. SRS and DMRS, etc. can be used to measure the uplink CSI.

[0146] In a time division duplex (TDD) system, due to the reciprocity of the uplink and downlink channels, the network device can obtain the uplink CSI by measuring the uplink reference signal (such as SRS), and then infer a more accurate downlink CSI. For example, the uplink CSI is used as the downlink CSI. In a frequency division duplex (FDD) system, the uplink and downlink reciprocity cannot be guaranteed, and the downlink CSI is obtained by the terminal device measuring the downlink reference signal. Therefore, the terminal device needs to generate a CSI report in a manner predefined by the protocol or configured by the base station, and feedback the CSI to the base station so that it can obtain the downlink CSI. In the following of this application, the uplink reference signal is taken as SRS as an example for description. It can be understood that the SRS in this application can be replaced by the uplink reference signal, or other uplink reference signals used for channel measurement, which are not limited herein.

[0147] In the NR protocol, the configuration and reporting process of downlink CSI is roughly as follows: The network device sends CSI reporting configuration (CSI-ReportConfig) to the terminal device, specifying the reporting type (reportConfigType), reporting quantity (reportQuantity), etc. Among them, the reporting type can be periodic, semi-persistent, or aperiodic; the reporting quantity can be rank indicator (RI), channel quality indicator (CQI), precoding matrix indicator (PMI), reference signal received power (RSRP), etc.

[0148] Exemplarily, the network device sends CSI-RS to the terminal device, and the terminal device performs channel measurement and interference measurement, etc. based on the CSI-RS to obtain measurement results. The terminal device determines each reporting quantity to be configured and reported according to the measurement results, and reports the downlink CSI to the network device. The downlink CSI includes one or more of RI, CQI, PMI, RSRP, etc. measured by the terminal device.

[0149] Among them, RI is used to indicate the number of layers of downlink transmission recommended by the terminal device; CQI is used to indicate the modulation and coding scheme that the terminal device determines can be supported by the current channel conditions; PMI is used to indicate the precoding recommended by the receiving end of the reference signal, such as the terminal device. The number of layers of the precoding indicated by PMI corresponds to RI.

[0150] It should be understood that the RI, CQI, PMI, etc. indicated in the above CSI report are the recommended values of the terminal device, and the network device can perform downlink transmission according to part or all of the information indicated in the CSI report. Or, the network device can also not perform downlink transmission according to the information indicated in the CSI report.

[0151] In the embodiments of the present application, the meaning of CSI is broader than that in traditional solutions and is not limited to CQI, PMI, RI, or CSI-RS resource indicator (CRI). It can also be a channel response (such as a channel response matrix), a channel matrix, a channel feature matrix, a precoding matrix, RSRP, signal to interference plus noise ratio (SINR) (signal to interference plus noise ratio can also be referred to as signal-to-interference and noise ratio), the identity (ID) of the optimal beam, or the ID of the top K beams, etc., one or more of them. For example, the optimal beam can be the beam with the maximum channel quality (such as RSRP, SINR, etc.) in the beam set. The top K beams can be K beams in the beam set whose channel quality (such as RSRP, SINR, etc.) is greater than or equal to a certain threshold, or the K beams ranked at the top when the channel quality is sorted from large to small, where K is a positive integer.

[0152] Among them, the channel response and the channel matrix represent the channel itself, and the channel feature matrix and the precoding matrix are matrices composed of the features extracted from the channel.

[0153] The descriptions of several CSI-related terms involved in the present application are as follows:

[0154] Target CSI: It can also be referred to as the full amount of CSI information, the ground-truth CSI. For the convenience of distinction and description in the following text, the target CSI is represented by H4.

[0155] CSI feedback information: It can also be referred to as the feedback information of CSI, the feedback information of channel measurement results, the feedback information of channel information, CSI feedback information, compressed information, the compressed information of channel information, the compressed information of CSI, the compressed channel information, or the compressed CSI, etc. For the convenience of distinction and description in the following text, the CSI feedback information is represented by C2.

[0156] Input CSI: It can also be referred to as measured CSI, original CSI, the information of CSI before compression, the original information of CSI, etc. The input CSI can be understood as the CSI input into the AI model (such as the first model mentioned below), that is, the input CSI can be the input of the AI model. For the convenience of distinction and description in the following text, the input CSI is represented by H1.

[0157] It should be understood that in this application, during the training phase, the CSI feedback information may refer to the information obtained after compressing the input CSI, which can be simply referred to as compressed CSI. The CSI feedback information may also refer to the information obtained after compressing and quantizing the input CSI, which can be simply referred to as quantized CSI. This application does not make any restrictions on this. During the model online phase, the CSI feedback information may be the information sent from the terminal side to the network side. To save the feedback overhead, the CSI feedback information may refer to the information obtained after compressing and quantizing the measured CSI (which can also be called the input CSI), that is, the quantized CSI.

[0158] It should also be understood that quantizing the compressed CSI can obtain the quantized CSI; de-quantizing the quantized CSI can obtain the compressed CSI. The compressed CSI obtained by de-quantizing the quantized CSI is the compressed CSI with quantization loss.

[0159] Reconstructed CSI: It can also be called recovered CSI, reconstructed CSI, recovered CSI, reconstructed information of channel information, decompressed CSI, decompressed channel information, or output CSI, etc. The reconstructed CSI can be understood as the CSI output by the AI model, that is, the reconstructed CSI can be the output of the AI model (such as the second model mentioned below). For the convenience of distinction and description below, the reconstructed CSI is represented by H3 or indicated.

[0160] Exemplarily, the relationship between the reconstructed CSI and the input CSI can refer to Figure 19 the corresponding description.

[0161] 7. Sounding reference signal (SRS): It is an uplink channel sounding signal sent by the terminal device and received by the network device. In the SRS sending method, including the time-frequency resources for sending SRS, the sending beam, the sending power, etc., are generally configured by the network device for the terminal device. In the relevant protocol framework of the 3rd Generation Partnership Project (3GPP), the network device can configure one or more SRS resource sets for the terminal device, and each SRS resource set has one or more SRS resources. rd generation partnership project, 3GPP), the network device can configure one or more SRS resource sets (SRS resource set) for the terminal device, and each SRS resource set has one or more SRS resources (SRS resource).

[0162] In addition, in the 3GPP-related protocols, different SRS resource sets can perform different functions. An SRS resource set generally supports four functions: {beam management (BM), codebook (CB), non-codebook (NCB), antenna switching (AS)}. Among them, antenna switching can also be referred to as antenna selection. The network device can configure the usage of each SRS resource set through radio resource control (RRC) signaling and then notify the terminal device of the function of the corresponding SRS resource set. For example, when the usage of the SRS resource set is antenna switching, the SRS corresponding to this SRS resource set is generally used to obtain complete uplink channel information. Assuming that in a time division duplexing (TDD) system, the channel has the property of uplink-downlink reciprocity, that is, the uplink channel and the downlink channel are consistent, then the SRS corresponding to this SRS resource can also obtain the channel for downlink transmission (or the precoding for downlink transmission) through uplink channel measurement.

[0163] In a possible implementation manner, the terminal device can send SRS in a frequency hopping manner, that is, multiple SRSs sent by a terminal device can be switched between different frequency bands within the frequency domain resources. Taking two SRS transmissions of the terminal device as an example, the terminal device can send SRS on sub-band 1 in a specific bandwidth, then switch from sub-band 1 in the specific bandwidth to sub-band 2 in the specific bandwidth, and send SRS again on sub-band 2 in the specific bandwidth.

[0164] In some cases, such as in the NR protocol, the uplink power for the terminal device to send SRS to the network device is limited, resulting in low accuracy of the channel state information obtained by the network device based on the received SRS reference signal. To improve the channel estimation accuracy obtained by the network device based on SRS, the bandwidth of a single SRS transmission by the terminal device can be reduced to increase the frequency power spectral density of SRS, so as to ensure the uplink power of a single SRS and improve the accuracy of the channel state information obtained by the network device.

[0165] Figure 5 is a schematic diagram of sending SRS in a frequency hopping manner. Figure 5The schematic diagrams of single - bandwidth SRS transmission, two - sub - band hopping SRS transmission, and four - sub - band hopping SRS transmission are shown. Among them, when the terminal device transmits SRS in a non - hopping manner (i.e., single - bandwidth SRS transmission), the SRS transmitted by the terminal device once can cover the configured bandwidth of the SRS resource. When the terminal device transmits SRS in a hopping manner (i.e., multiple - sub - band hopping SRS transmission), the SRS transmitted by the terminal device each time can cover a part of the configured bandwidth of the SRS resource (i.e., one hopping sub - band), and the terminal device transmitting SRS multiple times within a hopping period can cover the configured bandwidth of the SRS resource.

[0166] The hopping characteristics of SRS can be jointly determined by parameters in both the time domain and the frequency domain.

[0167] For example, the determination process of the time - domain position of SRS is as follows:

[0168] In the time domain, SRS occupies N S symbols (such as 1, 2, 4) within a time slot, the repetition factor \(R\in\{1, 2, 4\}\), and \(R\leq N\) S , that is, it is repeated \(R\) times on each symbol.

[0169] It can be seen from the repetition factor that:

[0170] When \(R = N\) S , that is, SRS is not supported to be transmitted in a hopping manner within a time slot;

[0171] When \(R = 1\) and \(N\) S = 2, 4, SRS is supported to be transmitted in a hopping manner within a time slot, and specifically, it can hop in units of one OFDM symbol;

[0172] When \(R = 2\) and \(N\) S = 4, SRS is supported to be transmitted in a hopping manner within a time slot, and specifically, it hops in units of a pair of OFDM symbols (i.e., 2 OFDM symbols);

[0173] Among them, for periodic SRS and semi - static SRS, corresponding period and time - domain offset parameters need to be configured. Periodic SRS and semi - static SRS can be transmitted in a hopping manner within a time slot or in a hopping manner between time slots (i.e., according to the period of SRS). Aperiodic SRS hopping is performed within a time slot (i.e., it hops completely once triggered);

[0174] A network device can configure SRS resources for a terminal device through RRC signaling. The RRC signaling can indicate information such as the number of ports included in the SRS resources, the frequency-domain position and time-domain position occupied by the SRS resources, the period used, the comb, the cyclic shift value, the sequence ID, etc. Among them, the frequency-domain position of the SRS resources is determined by a set of frequency-domain parameters in the RRC signaling (for example, in the existing 3GPP protocol, the frequency-domain parameters include n RRC , n shift , B SRS , C SRS , b hop ). Exemplarily, the terminal device determines the bandwidth occupied by the SRS (i.e., the configured bandwidth of the SRS resources), the starting position in the frequency domain, and the bandwidth occupied by the SRS in each symbol (or the bandwidth occupied by the frequency-hopping subband) through these frequency-domain parameters and the rules predetermined by the protocol.

[0175] 8. Auto-encoder (AE) model: It can generally refer to a network structure composed of two AI models, such as a network structure composed of an encoder and a decoder. Among them, each model can be an AI model. The AE model can also be called a bilateral model, a two-terminal model, a collaborative model, etc. The encoder and decoder of the AE are usually co-trained and can be used in a matching manner.

[0176] When AI technology is introduced into a wireless communication network, a CSI feedback method based on an AI model is generated. The terminal device can use the AI model to perform compressed feedback on the CSI, and the network device can use the AI model to recover the compressed CSI. Exemplarily, the feedback of the CSI based on the AI model can be implemented based on the AE model.

[0177] Figure 6 is a schematic diagram of using the AE model for CSI feedback. As Figure 6 shown, the terminal side compresses the target CSI through the encoder, and the network side reconstructs the CSI through the decoder. Exemplarily, the terminal side can use the target CSI (i.e., V) as the input of the encoder, and the encoder can compress the target CSI to obtain the compressed CSI (i.e., C). The terminal side can quantize the compressed CSI to obtain CSI feedback information. The terminal side sends the CSI feedback information to the network side, for example, by sending it to the network side through a CSI report.

[0178] The network side can first dequantize the CSI feedback information to obtain the compressed CSI with quantization loss (i.e., ). The network side uses the compressed CSI as the input of the decoder, and the decoder performs CSI decompression based on the compressed CSI to obtain the reconstructed CSI (i.e., )。

[0179] Among them, the quantizer used to perform quantization can be predefined, such as predefined by a protocol; it can also be indicated by the network side, and there is no limitation on this.

[0180] In another implementation, the encoder on the terminal side can also compress and quantize the target CSI and output CSI feedback information. The decoder on the network side can also dequantize and decompress the CSI feedback information to obtain the reconstructed CSI.

[0181] It should be understood that the encoder on the terminal side can be deployed inside the terminal device or in other devices outside the terminal device, such as the host of the aforementioned OTT or a cloud server, etc.; the decoder on the network side can be deployed inside the network device or in other devices outside the network device, such as the aforementioned intelligent network element.

[0182] It should also be understood that although the encoder and decoder are shown in the figure, this is only a model division from a functional perspective. The encoder can also be called the first model, and the decoder can also be called the second model. In addition, this application does not limit the number of models included in the AE model.

[0183] It should also be understood that the AE model is only a possible model for implementing the above functions and should not impose any limitation on this application. This AE model can also be replaced by other AI models that can achieve the same or similar functions.

[0184] Among them, the encoder can be understood as a CSI generator, a CSI generation part, a CSI generation model, a CSI compressor, a CSI compression model, etc.; the decoder can be understood as a CSI reconstructor, a CSI reconstruction part, a CSI reconstruction model, a CSI restorer, a CSI restoration model, a CSI decompressor, a CSI decompression model.

[0185] 9. Model files and model parameters: Model files and / or model parameters can be used to determine the model. The model file can be used to indicate the model structure, and the model structure includes, for example, but is not limited to: FNN, CNN, and RNN. The model file can have a fixed format, such as a standard predefined format, or a format negotiated in advance by both ends of the docking. Model parameters can refer to the parameters in a neural network model, for example, including but not limited to, the number of layers of the neural network, the types and weights of neurons in each neural network layer, etc. This application does not limit the method of sending down the parameters of the reference model.

[0186] Taking DNN as an example. The idea of DNN comes from the neuron structure of the brain tissue. Each neuron can perform a weighted summation operation on its input and generate an output through a non-linear function of the result of the weighted summation operation. Figure 7 is an example of the neuron structure. Figure 7 The input of the neuron shown is x = [x0 x1 … x N-1 , and the weights corresponding to the input are w = [w0 w1 … w N-1 , the bias of the weighted summation is b, and the form of the non-linear function f() can be diversified. For example, the non-linear function f() is the maximum function max{0, x}. Then the effect of a neuron's execution is where N is a positive integer; n is a positive integer greater than or equal to 0 and less than or equal to (N - 1).

[0187] DNN generally has multiple neural network layers, including an input layer, one or more hidden layers (or, implicit layers), and an output layer. Generally speaking, the first layer is the input layer, the last layer is the output layer, and the middle layers are hidden layers. Each layer includes multiple neurons. The layers are fully connected. That is to say, any neuron in the i-th layer is connected to any neuron in the (i + 1)-th layer. The input layer can pass the received values (i.e., the input of the DNN) to the middle hidden layer after being processed by the neurons. Similarly, the hidden layer can pass the calculation result to the final output layer to generate the output of the DNN. Figure 8 shows an example of DNN. Figure 8 The DNN model shown has three neural network layers, namely an input layer, a hidden layer, and an output layer.

[0188] It should be understood that the examples combined with Figure 7 and Figure 8 are only shown for easy understanding and should not constitute any limitation to this application. This application does not limit the structure and parameters used in the AI model.

[0189] One of the model structure or model parameters can be predefined, and the other can be sent by the sender (such as the network side), or both the model structure and model parameters can be sent by the sender (such as the network side). This application does not make any limitation on this.

[0190] In the embodiments of this application, the sending model can refer to sending the model file and / or model parameters, and the receiving model can refer to receiving the model file and / or model parameters.

[0191] 10. Dual - end docking: It refers to the docking between the sender (such as the network side) and the receiver (such as the terminal side), which can be dataset - based docking or model - based docking. Taking the inference task of CSI compression as an example below, the dataset - based docking and model - based docking will be described separately.

[0192] Among them, the dataset - based docking mainly means that the sender provides a dataset for the receiver for the receiver's model training.

[0193] Figure 9 It is a schematic diagram of the dataset docking between the network side and the terminal side. As Figure 9 shown, the network side can first obtain the dataset through joint training. Exemplarily, the network side uses a CSI compression model (such as an encoder) to compress the target CSI in the dataset and a CSI reconstruction model (such as a decoder) to decompress the target CSI. That is, the input of the CSI compression model can be the target CSI, and the output can be the compressed CSI; the input of the CSI reconstruction model can be the compressed CSI, and the output can be the reconstructed CSI. In this way, the dataset obtained by the network side includes the target CSI and the compressed CSI.

[0194] Optionally, the network side further performs quantization processing on the compressed CSI to obtain CSI feedback information. Correspondingly, the dataset can also include: the target CSI and the CSI feedback information. The dataset can also include a quantizer. The quantizer can also be replaced by a quantization method. The quantizer can be included in the CSI compression model. In other words, the CSI compression model can also have the function of performing quantization processing on the compressed CSI.

[0195] Optionally, if the compressed CSI is quantized to obtain CSI feedback information, before the network side inputs the CSI feedback information into the CSI reconstruction model, it can perform de - quantization processing on the CSI feedback information to obtain the input of the CSI reconstruction model. Among them, the de - quantizer used for de - quantizing the CSI feedback information can be predefined, such as protocol - predefined, or sent by the sender. The de - quantizer can be included in the CSI reconstruction model. In other words, the CSI reconstruction model can also have the function of performing de - quantization processing on the CSI feedback information.

[0196] Furthermore, the network side can send the dataset to the terminal side. The terminal side performs model training based on the received dataset to obtain the CSI compression model of the terminal side.

[0197] Model-based docking mainly means that the sender provides the receiver with model files and / or model parameters for the receiver's model deployment or for the receiver's model development (such as model training) and model deployment. In this application, the sender can provide the receiver with model files and / or model parameters for the receiver to perform model training.

[0198] Figure 10 is a schematic diagram of model docking between the network side and the terminal side. As Figure 10 shown, the network side can first obtain the CSI compression model through joint training. The specific process of the network side's joint training can refer to the above description in combination with Figure 9 and will not be elaborated here. The network side can send the CSI compression model or CSI reconstruction model obtained through joint training to the terminal side. For example, the network side sends the model file and / or model parameters of the CSI compression model to the terminal side, or sends the model file and / or model parameters of the CSI reconstruction model to the terminal side. The terminal side can perform model training according to the received model file and / or model parameters.

[0199] It should be understood that as described above in combination with Figure 9 and Figure 10 the CSI compression model shown can be an encoder, and the CSI reconstruction model can be a decoder.

[0200] As mentioned above, the dual-end models are usually co-trained and can be used in a matching manner. To support the docking, model development, or deployment of the dual-end models, it is necessary to provide a dataset or a model for one or both of the dual-end models. For example, the network side can provide a dataset for the terminal side for model development and / or deployment. As can be seen from the above in combination, the data included in the dataset is not fixed, and there may be multiple types of datasets, and different types of datasets contain different data. Therefore, how the network constructs a dataset for model training to improve the performance of the dual-end models on the network side and the terminal side is an issue that needs to be considered.

[0201] In view of this, this application provides a communication method. The data transmitted from the network side to the terminal side can include multiple CSIs. If the multiple CSIs include historical CSIs, the terminal side performs model training on the model for CSI compression according to the received dataset, which is beneficial to improving the performance of CSI feedback.

[0202] For ease of understanding, before introducing the method provided in this application, the resources involved in the following embodiments are first described.

[0203] The resources involved in this application can include one or more of time-domain resources, frequency-domain resources, or space-domain resources.

[0204] The time domain resources may include one or more time domain units. The time domain unit may be any of the following: one or more time slots, one or more subframes, one or more frames, one or more orthogonal frequency division multiplexing (OFDM) symbols, 1 second (s) or several seconds (s), 1 millisecond (ms) or several milliseconds (ms), etc.

[0205] The same or different time domain resources mentioned in this application may refer to the same or different time domain units included in the time domain resources. Taking time domain resource A and time domain resource B as examples, if time domain resource A is different from time domain resource B, the time domain units included in time domain resource A are different from those included in time domain resource B. For example, time domain resource A includes time slot #a and time domain resource A includes time slot #b. If time domain resource A is the same as time domain resource B, time domain resource A and time domain resource B include the same time domain units. For example, both time domain resource A and time domain resource B include time slot #a. If time domain resource A includes time domain resource B, time domain resource A includes at least the time domain units belonging to time domain resource B. For example, if time domain resource B includes time slot #a, time domain resource A includes at least time slot #a.

[0206] The frequency domain resources may include one or more frequency domain units. The frequency domain unit may be any of the following: subcarrier, component carrier (CC), resource block (RB), subchannel, resource pool, bandwidth, subband, bandwidth part (BWP), channel, physical resource block (PRB), resource block group (RBG), PRB bundling, or an interlaced RB, etc.

[0207] The same or different frequency-domain resources mentioned in this application may refer to the same or different frequency-domain units included in the frequency-domain resources. That one frequency-domain resource includes another frequency-domain resource may mean that the one frequency-domain resource includes at least the frequency-domain units belonging to the other frequency-domain resource, and the other frequency-domain resource may also be referred to as a partial frequency-domain resource in the one frequency-domain resource. Taking frequency-domain resource A and frequency-domain resource B as an example, if frequency-domain resource A is different from frequency-domain resource B, then the frequency-domain units included in frequency-domain resource A are different from those included in frequency-domain resource B. For example, frequency-domain resource A includes sub-band #a and frequency-domain resource A includes sub-band #b. If frequency-domain resource A is the same as frequency-domain resource B, then frequency-domain resource A and frequency-domain resource B include the same frequency-domain units. For example, both frequency-domain resource A and frequency-domain resource B include sub-band #a. If frequency-domain resource A includes frequency-domain resource B, then frequency-domain resource A includes at least the frequency-domain units belonging to frequency-domain resource B, or in other words, frequency-domain resource B is a partial frequency-domain resource in frequency-domain resource A. For example, if frequency-domain resource B includes sub-band #a and frequency-domain resource A includes frequency-domain resource B, then frequency-domain resource A includes at least sub-band #a. That frequency-domain resource A includes frequency-domain resource B may also be said that resource B is a subset of resource A.

[0208] Spatial-domain resources are related to one or more of the following: transmit ports, receive ports, transmit streams, or receive streams. For example, if spatial-domain resource A is different from spatial-domain resource B, then at least one of the following corresponding to spatial-domain resource A and spatial-domain resource B is different: transmit ports, receive ports, transmit streams, or receive streams. For example, if spatial-domain resource A includes spatial-domain resource B, then the transmit port / receive port corresponding to spatial-domain resource A includes the transmit port / receive port corresponding to spatial-domain resource B, and / or, the transmit stream / receive stream corresponding to spatial-domain resource A includes the transmit stream / receive stream corresponding to spatial-domain resource B. If spatial-domain resource A includes spatial-domain resource B, then each of the above at least one item corresponding to spatial-domain resource A includes at least the corresponding item corresponding to spatial-domain resource B. For example, if the transmit port corresponding to spatial-domain resource B is transmit port A and the transmit stream is transmit stream A, then the transmit port corresponding to spatial-domain resource A includes at least transmit port A, and the transmit stream corresponding to spatial-domain resource A includes at least transmit stream A.

[0209] If the resource includes multiple items among time-domain resources, frequency-domain resources, or spatial-domain resources, resource M being different from resource N means that at least one of the above-mentioned multiple items included in resource M and resource N is different; resource M being the same as resource N means that each of the above-mentioned multiple items included in resource M and resource N is the same; resource M including resource N means that each of the above-mentioned multiple items included in resource M includes at least the corresponding item included in resource N. For example, resource M includes time-domain resource M and frequency-domain resource M, and resource N includes time-domain resource N and frequency-domain resource N. Resource M being different from resource N means that time-domain resource M is different from time-domain resource N, and / or frequency-domain resource M is different from frequency-domain resource N. Resource M being the same as resource N means that time-domain resource M is the same as time-domain resource N, and frequency-domain resource M is the same as frequency-domain resource N. Resource M including resource N means that time-domain resource M includes at least time-domain resource N, and frequency-domain resource M includes at least frequency-domain resource N.

[0210] In this application, the resource can also be referred to as a resource set, and the resource set can include at least one of the above-mentioned time-domain resources, and / or at least one of the above-mentioned frequency-domain resources, and / or at least one of the above-mentioned spatial-domain resources.

[0211] It should also be noted that in the following embodiments, different numbered data sets are only for facilitating the distinction of different types of data sets, and do not restrict the size or quantity of the data sets. For example, the "first data set", "second data set", and "third data set" in the following embodiments are only data sets of different types or for different purposes, and do not limit the quantity, size relationship, or priority relationship of the data sets. Exemplarily, the first data set, the second data set, or the third data set can each include multiple data, and a single data can also be referred to as a sample, a channel measurement result, or channel information, etc.

[0212] It should also be noted that in the following embodiments, the "model" can be replaced with names such as an encoder, a decoder, a compressor, a decompressor, a quantizer, or a dequantizer, etc., and this application does not make any restrictions in this regard. For example, if the inference task corresponding to the model (such as the first model) is to compress data, then the model can be replaced with an encoder, a compressor, or a quantizer, etc. Another example is that if the inference task corresponding to the model (such as the second model) is to decompress data, then the model can be replaced with names such as a decoder, a decompressor, or a dequantizer, etc.

[0213] Figure 11 It is a schematic flowchart of the communication method 1100 provided by the embodiments of this application. This method can include the following steps.

[0214] S1110, the terminal side obtains a first data set.

[0215] Wherein, the first data set can be used for the terminal device to perform model training.

[0216] In this application, the device that performs model training on the terminal side can be referred to as a training device. The training device can be a terminal device or a device other than the terminal device, such as the aforementioned OTT system server, etc., and there is no limitation in this regard.

[0217] Exemplarily, each sample in the first dataset includes the following information: first CSI, CSI feedback information, and second CSI.

[0218] Among them, the first CSI can characterize the channel state of the first resource set, or rather, the first CSI characterizes the channel information of the first channel, and the first channel corresponds to the first resource set (for example, the channel information of the first channel is obtained by measuring the reference signal transmitted on the first resource set), and the first channel can be the channel between the terminal side and the network side.

[0219] The second CSI characterizes the channel state of the second resource set, or rather, the second CSI characterizes the channel information of the second channel, and the second channel corresponds to the second resource set (for example, the channel information of the second channel is obtained by measuring the reference signal transmitted on the second resource set). The second resource set is part or all of the first resource set.

[0220] Exemplarily, the first resource set includes at least one of frequency domain resources or spatial domain resources. It can be understood that the type of resources included in the second resource set (i.e., frequency domain resources or spatial domain resources) is the same as the type of resources included in the first resource set, that is, the second resource set includes at least one of frequency domain resources or spatial domain resources.

[0221] Taking the resources included in the resource set as frequency domain resources as an example, when the second resource set is all of the first resource set, it can be understood that the frequency domain resources (or frequency domain units) included in the second resource set are the same as the frequency domain resources (or frequency domain units) included in the first resource set. When the second resource set is part of the first resource set, it can be understood that the frequency domain resources included in the second resource set are a subset of the frequency domain resources included in the first resource set. It should be understood that when the first resource set and the second resource set are frequency domain resources and / or spatial domain resources, when describing the relationship between the first resource set and the second resource set, the relationship between the time domain resources corresponding to the first resource set and the time domain resources corresponding to the second resource is not limited.

[0222] Optionally, the first resource set corresponds to a first time-domain resource, the second resource set corresponds to a second time-domain resource and / or the first time-domain resource, and the second time-domain resource is earlier than the first time-domain resource. In other words, the time-domain resource corresponding to the second resource set is earlier than the time-domain resource corresponding to the first resource set, or the second resource set corresponds to the same time-domain resource as the first resource set, or the second resource set includes frequency-domain resources and / or spatial-domain resources corresponding to two time-domain resources, and one of the two time-domain resources is the same as the time-domain resource corresponding to the first resource set, and the other time-domain resource is earlier than the time-domain resource corresponding to the first resource set.

[0223] For example, when the second resource set is a part of the first resource set, the time-domain resource corresponding to the second resource set is the same as the time-domain resource corresponding to the first resource set. If the first resource set is the full band corresponding to the first time-domain resource, the second resource set can be some sub-bands in the full band, that is, the first CSI can be the full-band CSI on the first time-domain resource, and the second CSI can be the CSI corresponding to some sub-bands in the full band on the first time-domain resource. The second CSI can also be the channel information obtained by measuring an uplink reference signal (such as SRS), that is, the second CSI can also be the SRS measurement value (SRS measurements), such as the SRS measurement value is obtained by measuring the SRS on the part of the sub-bands.

[0224] For another example, when the second resource set is a part of the first resource set, the time-domain resource corresponding to the second resource set is earlier than the time-domain resource corresponding to the first resource set. If the first resource set is the full band corresponding to the first time-domain resource, the second resource set can be some sub-bands in the full band, that is, the first CSI can be the full-band CSI on the first time-domain resource, and the second CSI can be the CSI corresponding to some sub-bands in the full band on the second time-domain resource. Exemplarily, the second CSI can be the SRS measurement result corresponding to the part of the sub-bands. The part of the sub-bands can be divided into multiple sub-bands, and the time-domain resources corresponding to each of the multiple sub-bands are different. For example, each sub-band corresponds to a time-domain unit included in the second time-domain resource, and the time-domain positions of the time-domain units corresponding to each sub-band can be different. The multiple time-domain units corresponding to the multiple sub-bands can be located at the time-domain positions of the frequency-hopping transmitted SRS.

[0225] For another example, when the second resource set is all of the first resource set, the second resource set may include a first partial resource set and a second partial resource set. The first partial resource set corresponds to the first time-domain resource, and the second partial resource set corresponds to the second time-domain resource. The first partial resource set and the second partial resource set do not overlap or partially overlap. For example, if the first resource set is the full band corresponding to the first time-domain resource, the second resource set may be the full band. The first partial resource set in the second resource set may be the first partial sub-band in the full band, and the first partial sub-band corresponds to the first time-domain resource; the second partial resource set may be the second partial sub-band in the full band, and the second partial sub-band is the other part of the sub-band except the first partial sub-band, or the second partial sub-band partially overlaps with the first partial sub-band. The second partial sub-band corresponds to the second time-domain resource. Or rather, the first CSI may be the full-band CSI on the first time-domain resource, and the second CSI includes the CSI corresponding to the first partial sub-band in the full band on the first time-domain resource and the CSI corresponding to the second partial sub-band on the second time-domain resource. The second CSI may also be the SRS measurement corresponding to the full band, and the SRS measurement may characterize the channel information of multiple sub-bands. Some of the multiple sub-bands (i.e., the first partial sub-bands) correspond to the first time-domain resource, and the other part of the sub-bands corresponds to the second time-domain resource. The frequency-domain positions of the multiple sub-bands may refer to the frequency-domain resources for transmitting SRS by frequency hopping, and the first time-domain resource and the second time-domain resource may each include one or more time-domain units, and the time-domain positions of the one or more time-domain units may refer to the time-domain positions for transmitting SRS by frequency hopping.

[0226] For another example, when the second resource set is all of the first resource set, the time-domain resource corresponding to the second resource set is earlier than the time-domain resource corresponding to the first resource set. For example, if the first resource set is the full band corresponding to the first time-domain resource, the second resource set may be the full band, and the second resource set is composed of multiple partial sub-bands, and the time-domain resource corresponding to each partial sub-band is earlier than the time-domain resource corresponding to the first resource set. The multiple time-domain units corresponding to the multiple sub-bands may be located at the time-domain positions for transmitting SRS by frequency hopping.

[0227] In this application, the time-domain resources corresponding to the first resource set are the same as those corresponding to the second resource set. It can be understood that the first resource set and the second resource set correspond to the same time-domain unit, or the corresponding time-domain units are relatively close. Since the first resource set is the resource set corresponding to CSI-RS and the second resource set is the resource set corresponding to SRS, CSI-RS and SRS may not be sent in the same time slot, but when their time-domain intervals are relatively close, it can be considered that the corresponding time-domain resources are the same. For example, if the first resource set corresponds to the first time-domain unit, the second resource set that is closest to the first time-domain unit in the time domain can be considered to have the same time-domain resources as the first resource set. The time-domain unit closest in time can be before the first time-domain unit, after the first time-domain unit, or before and after the first time-domain unit, without limitation.

[0228] In this application, the time-domain resources may include one or more time-domain units. For example, the time-domain unit can be a time slot or a frame, and the time-domain unit can also be of other granularities, which can specifically refer to the description in the above text.

[0229] Full-band CSI refers to CSI that includes CSI for all sub-bands. Obtaining CSI for a partial sub-band from the full-band CSI means obtaining the CSI for the frequency-domain resources corresponding to the partial sub-band from the full-band CSI. Specifically, the full-band CSI can be represented as a matrix or tensor with a frequency-domain dimension, and a vector or matrix or tensor corresponding to the CSI for the frequency-domain resources corresponding to the sub-band to be extracted is obtained therefrom.

[0230] The first CSI and the second CSI included in each sample in the first dataset are introduced above. Next, the CSI feedback information included in each sample is introduced. The CSI feedback information is obtained by compressing (or compressing and quantizing) the first CSI or the third CSI. In other words, the CSI feedback information included in each sample corresponds to the same time-domain resources as the first CSI included in each sample, or the third CSI.

[0231] Among them, each sample includes the first CSI described above. The third CSI can be used to characterize the channel state of the third resource set. The third resource set is a part of the first resource set, or the third resource set is different from the first resource set. For example, when the first resource set is the full band, the third resource set can be a part of the subbands in the full band, that is, the CSI feedback information can be obtained by compressing the CSI corresponding to the full band, or can be obtained by compressing the CSI corresponding to some subbands in the full band. For another example, the CSI feedback information can also be obtained by compressing the CSI (denoted as CSI#1, an example of the third CSI) whose similarity or difference degree with the first CSI is less than or equal to a threshold. The resource set corresponding to CSI#1 is resource set #1 (an example of the third resource set), and the resource set #1 can be different from the first resource set, that is, the resource set #1 and the first resource set include different frequency domain resources and / or spatial domain resources. There is no restriction on the time domain resources corresponding to the third resource set. For example, the third resource set corresponds to the same time domain resources as the first resource set, such as the first time domain resource.

[0232] Among them, exemplarily, the similarity or difference degree between CSIs can be characterized by one or more of the mean square error (MSE), normalized mean square error (NMSE), mean absolute error (MAE), square generalized cosine similarity (SGCS), or generalized cosine similarity (GCS) between CSIs.

[0233] It should be understood that the resources included in the resource set described by taking the frequency domain unit as the subband above are only examples and do not constitute a limitation to this application. The frequency domain unit can also be of other granularities. For example, the frequency domain unit can also be an RB.

[0234] Each sample in the first data set can also include multiple groups of the following information: the first CSI, the CSI feedback information, and the second CSI. The relationship between each group of the first CSI, the CSI feedback information, and the second CSI is as described above.

[0235] Exemplarily, the terminal side obtains the first data set through the following two possible implementation manners.

[0236] In Method 1, the network side sends the first data set to the terminal side (S1110a). Correspondingly, the terminal side receives the first data set. Or rather, one way for the terminal side to obtain the first data set is to receive the first data set from the network side.

[0237] Exemplarily, before the network side sends the first data set, the dual - end model 1 is first trained to obtain the first data set. The dual - end model 1 can be understood as the dual - end model on the network side, and the dual - end model 1 can include an encoder 1 and a decoder 1. The network side training the dual - end model 1 can be a joint training of the encoder 1 and the decoder 1.

[0238] Figure 12 It is a schematic diagram of the process of the network side for joint training. As Figure 12 shown, the network side takes the first CSI (or the third CSI) (which can be denoted as H1 NW ) as the input of the encoder 1, and compresses the first CSI (or the third CSI) through the encoder 1 to obtain the first CSI feedback information (which can be denoted as C2). The first CSI feedback information can be used as the input of the decoder 1. The input of the decoder 1 also includes the second CSI (such as the SRS measurement value, which can be denoted as H2 NW ). The decoder 1 can decompress the first CSI feedback information to obtain the reconstructed CSI (which can be denoted as H3 NW ). Furthermore, the network side can use the target CSI (which can be denoted as H4 NW ) as a label to train the encoder 1 and the decoder 1. Through the training of the encoder 1 and the decoder 1, the reconstructed CSI output by the decoder 1 approaches the target CSI. Among them, the target CSI can be the first CSI.

[0239] Optionally, the encoder 1 can compress and quantize the first CSI to obtain the first CSI feedback information. Correspondingly, the decoder 1 can decompress and de - quantize the first CSI feedback information to obtain the reconstructed CSI.

[0240] Optionally, the encoder 1 can compress the first CSI to obtain the compressed CSI, and the compressed CSI is quantized to obtain the first CSI feedback information. Correspondingly, the decoder 1 can decompress the information after de - quantizing the first CSI feedback information to obtain the reconstructed CSI. Among them, the information after de - quantizing the first CSI feedback information corresponds to the compressed CSI.

[0241] Optionally, before the network side trains the dual - end model 1, the network side determines the second CSI input to the decoder 1. The determination method of the second CSI can correspond to the different correlation relationships between the first CSI and the second CSI in the above text. Exemplarily, the determination of the second CSI can include the following methods:

[0242] In Method 1, determine that the second CSI is the CSI corresponding to a partial resource set (or resource) in the first resource set on the first time-domain resource, where the time-domain resource corresponding to the partial resource set is the same as the time-domain resource corresponding to the first resource set. The channel state corresponding to the first resource set is characterized by the first CSI. That is, when the network side determines a first CSI, the network side can determine the second CSI based on the first CSI.

[0243] For example, if the first resource set is the full band corresponding to the first time-domain resource, the second CSI can be the CSI corresponding to partial sub-bands in the full band on the first time-domain resource. The second CSI can also be the SRS measurement value obtained by measuring the SRS on the partial sub-bands.

[0244] In Method 2, determine that the second CSI is the CSI corresponding to a partial resource set (or resource) in the first resource set, where the time-domain resource corresponding to the partial resource set is earlier than the time-domain resource corresponding to the first resource set.

[0245] For example, if the first resource set is the full band corresponding to the first time-domain resource, the second CSI can be the CSI corresponding to partial sub-bands in the full band. The partial sub-bands include one or more sub-bands, and each of the one or more sub-bands corresponds to a time-domain resource, and the time-domain resource corresponding to each sub-band is earlier than the first time-domain resource. For example, the second CSI can be the SRS measurement corresponding to the partial sub-bands in the SRS measurement corresponding to the full band. The position of the time-domain resource corresponding to the SRS measurement corresponding to the full band can refer to the time-domain position of the frequency-hopping transmitted SRS. The SRS measurement corresponding to the full band also includes the SRS measurements corresponding to other partial resource sets in the first resource set on the first time-domain resource.

[0246] In Method 3, determine that the second CSI includes a first part of CSI and a second part of CSI. The first part of CSI can be used to characterize the channel state corresponding to a partial resource set (or resource) in the first resource set on the first time-domain resource, and the second part of CSI can be used to characterize the channel state corresponding to other partial resource sets (or resources) in the first resource set on the second time-domain resource, where the second time-domain resource is earlier than the first time-domain resource.

[0247] For example, the first resource set is the full band corresponding to the first time-domain resource. The first part of CSI can be the CSI corresponding to some sub-bands in the full band, and these sub-bands correspond to the first time-domain resource. The second part of CSI is the CSI corresponding to other sub-bands in the full band. The other sub-bands include one or more sub-bands, and each of the one or more sub-bands corresponds to a time-domain resource, and the time-domain resource corresponding to each sub-band is earlier than the first time-domain resource. For example, the first part of CSI can be the SRS measurement corresponding to these sub-bands in the SRS measurement corresponding to the full band, and the second part of CSI can be the SRS measurement corresponding to the other sub-bands in the SRS measurement corresponding to the full band. The position of the time-domain resource corresponding to the SRS measurement corresponding to the full band can refer to the time-domain position of the frequency-hopping transmitted SRS.

[0248] Figure 13 The figure shows a schematic diagram of the network side determining a sample in the first dataset. As Figure 13 shown, a first CSI determined by the network side can be the CSI corresponding to a transmission time interval (TTI), that is, the time-domain resource corresponding to the CSI belongs to this TTI3. For example, the network side determines that the first CSI in a sample is the CSI corresponding to TTI3. The first CSI is used to characterize the channel state corresponding to the first resource set. The first resource set can include sub-band #1 to sub-band #4. The CSI feedback information in the sample can be obtained by the encoder 1 compressing (or compressing and quantizing) the first CSI. The second CSI in the sample can include the SRS measurement corresponding to some sub-bands (such as the SRS measurement corresponding to sub-band #1) on the time-domain resource (i.e., the first time-domain resource) corresponding to the first CSI, and the SRS measurements corresponding to other sub-bands (such as sub-band #2 to sub-band #4) on the time-domain resource earlier than the first time-domain resource. The SRS measurements corresponding to sub-band #2 to sub-band #4 respectively correspond to different time-domain resources and are earlier than the first time-domain resource. The position of the time-domain resource corresponding to the SRS measurement can refer to the time-domain position of the frequency-hopping transmitted SRS.

[0249] After obtaining the samples included in the first dataset, the network side sends the first dataset to the terminal device.

[0250] In a possible implementation, the network side sends the first dataset by carrying it in one or more messages, and there is no limit on the number of messages carrying the first dataset.

[0251] Correspondingly, the terminal side receives the first dataset.

[0252] Exemplarily, if the training device on the terminal side is a terminal device, the terminal side receiving the first data set may include: the terminal device receives the data set from a network device, such as received through the air interface; or, the terminal device obtains the data set from an OTT system server, and the data set is received by the OTT system from an intelligent network element on the network side, such as received through a wired network.

[0253] If the training device on the terminal side is an OTT system server, the terminal side receiving the first data set may include: the OTT system server receives the first data set from an intelligent network element, such as received through a wired network; or, the OTT system server obtains the first data set from a terminal device, and the first data set is received by the terminal device from a network device, such as received through the air interface.

[0254] In the second method, the terminal side receives a second data set and a third data set from the network side (S1110b), and determines the first data set based on the second data set and the third data set (S1110c), or rather, the terminal side obtains the first data set based on the second data set and the third data set.

[0255] Among them, the second data set may include multiple first CSIs, or rather, the second data set may be a CSI sample sequence. The time-domain resources corresponding to the multiple first CSIs do not overlap. Exemplarily, the multiple first CSIs may correspond to multiple first time-domain resources, and the first time-domain resources corresponding to each first CSI do not overlap with each other.

[0256] For example, the multiple first time-domain resources may be correlated in the time domain. For example, if the multiple first time-domain resources are multiple time-domain units (or moments, time periods), the multiple time-domain units are continuous in the time domain, or arranged in a specific order, such as the identifiers (such as indexes) corresponding to the multiple time-domain units are continuous, or arranged in ascending or descending order. For example, the multiple first CSIs may correspond to the time-domain resources of multiple consecutive periodic or semi-persistent CSI-RSs. For another example, the multiple first time-domain resources are not correlated in the time domain, or rather, the multiple first time-domain resources do not have correlation in the time domain.

[0257] The third data set includes multiple CSI feedback messages, or rather, the third data set may be a CSI feedback message sample sequence. One or more of the multiple CSI feedback messages in the third data set correspond to one of the multiple first CSIs included in the second data set, or rather, one first CSI in the second data set may correspond to one or more CSI feedback messages in the third data set.

[0258] Exemplarily, the one or more CSI feedback information is obtained by compressing (or compressing and quantizing) a first CSI. Further, the one or more CSI feedback information may correspond to the same time domain resource as the first CSI.

[0259] In a possible implementation, the terminal determines that a sample in the first dataset includes a first CSI in the second dataset. The sample further includes a CSI feedback information, and the CSI feedback information may be one of the one or more CSI feedback information corresponding to the first CSI.

[0260] The sample further includes a second CSI, where the second CSI is determined based on the first CSI, or the second CSI is determined based on a part of the CSIs in the plurality of CSIs. The part of the CSIs may include the first CSI or may not include the first CSI.

[0261] For example, the terminal determines the second CSI based on the first CSI, and the second CSI may be the CSI corresponding to a part of the resource sets (or resources) in the first resource set corresponding to the first CSI.

[0262] As another example, the terminal determines the second CSI based on a part of the first CSIs in the plurality of first CSIs, where the part of the CSIs does not include the first CSI, and the time domain resource corresponding to the part of the CSIs may be earlier than the time domain resource corresponding to the first CSI. The terminal may determine that the second CSI includes the CSI corresponding to a part of the resource sets in the resource set corresponding to each first CSI in the part of the first CSIs, and the part of the resource sets corresponding to each first CSI do not overlap with each other.

[0263] As another example, the terminal determines the second CSI based on a part of the first CSIs in the plurality of first CSIs, where the part of the CSIs includes the first CSI, and the time domain resource corresponding to the other CSIs except the first CSI in the part of the CSIs may be earlier than the time domain resource corresponding to the first CSI. The terminal may determine that the second CSI includes the CSI corresponding to a part of the resource sets in the resource set corresponding to each first CSI in the part of the CSIs, and the part of the resource sets corresponding to each first CSI do not overlap with each other.

[0264] Optionally, the terminal side determines the second CSI according to the SRS hopping pattern. For example, if the SRS has 17 hops and a period of 40 ms, and the interval between two adjacent first CSIs is 20 ms, then the terminal side determines the second CSI according to 17 first CSIs with an interval of 1 first CSI. Which subbands are selected for each of the 17 first CSIs is determined by the SRS hopping pattern. For example, the first first CSI selects subband 1, the second first CSI selects subband 2, and so on. The time-domain resource corresponding to the second CSI is the time-domain resource corresponding to the last first CSI used to determine the second CSI.

[0265] In a possible implementation, the SRS hopping pattern is the SRS hopping pattern configured for the current terminal side.

[0266] In another possible implementation, the base station side configures an SRS hopping pattern for the terminal side to determine the second CSI from the first CSIs. This SRS hopping pattern can be different from the SRS hopping pattern previously configured for the terminal side.

[0267] In another possible implementation, the terminal side can use a random SRS hopping pattern to determine the second CSI.

[0268] Figure 14 The figure shows a schematic diagram of the terminal side determining samples in the first dataset. As Figure 14 shown, the terminal side determines the first CSI included in a sample (such as sample #1) in the first dataset. The first CSI is a first CSI in the second dataset. For example, the first CSI included in sample #1 is the first CSI corresponding to TTI3, that is, the time-domain resource corresponding to this first CSI belongs to this TTI3. The terminal side determines that sample #1 also includes a CSI feedback message in the third dataset. This CSI feedback message can be obtained by compressing the first CSI. The terminal side determines that sample #1 also includes a second CSI. For example, the second CSI can include the CSI corresponding to some subbands (such as the CSI corresponding to subband #1) on the time-domain resource corresponding to the first CSI, and the CSIs corresponding to other subbands (such as subbands #2 to #4) on the time-domain resources earlier than this first time-domain resource. The CSIs corresponding to subbands #2 to #4 correspond to different time-domain resources.

[0269] Similarly, the terminal side determines that another sample in the first dataset (such as sample #2) includes first CSI, second CSI, and CSI feedback information. The first CSI included in this sample #2 can be a first CSI in the second dataset, such as the first CSI corresponding to TTI1, that is, the time domain resource corresponding to this first CSI belongs to this TTI1. The third CSI feedback information included in sample #2 is a CSI feedback information in the third dataset, and this CSI feedback information can be obtained by compressing the first CSI included in sample #2. The second CSI included in sample #2 can include the CSI corresponding to some subbands on the time domain resource (i.e., TTI1) corresponding to the first CSI included in sample #2 (such as the CSI corresponding to subband #2), and the CSI corresponding to other subbands on the time domain resources earlier than this first time domain resource respectively. For example, the time domain resource earlier than this first time domain resource can include TTI0, the other subbands on the time domain resource earlier than this first time domain resource can include subband #4, and the time domain resource earlier than this first time domain resource can also include the time domain resources before TTI0 (not shown in the figure), and the other subbands on the time domain resource earlier than this first time domain resource can also include: the CSI corresponding to other subbands on the time domain resources before TTI0 (such as subband #1 and subband #3) respectively.

[0270] It should be understood that the first CSI included in two samples in the first dataset can be the same or different, and there is no limitation on this. Exemplarily, when the first CSI included in two samples is the same, at least one of the CSI feedback information or the second CSI included in these two samples is different. For example, the CSI feedback information included in one sample can be obtained by compressing this first CSI, and the CSI feedback information in another sample can be obtained by compressing some of the CSI feedback information in the first CSI, or by compressing a CSI different from this first CSI (refer to the third CSI in the above text). Another example is that the second CSI in one sample is the CSI corresponding to some resource sets (or resources) in the first resource set on the first time domain resource, and the second CSI in another sample is the CSI corresponding to some resource sets (or resources) in the first resource set, and the time domain resource corresponding to this part of the resource set is earlier than the time domain resource corresponding to the first resource set.

[0271] S1120. The terminal side performs model training on the first model based on this first dataset.

[0272] Among them, the first model is used to perform a first process on the CSI to be fed back, and the first process at least includes CSI compression. For example, the first process is compression, or the first process includes compression and quantization. The first model can be understood as the encoder (denoted as encoder 2) in the dual - end model (denoted as dual - end model 2) on the terminal side. Among them, the CSI to be fed back can be the currently measured CSI on the terminal side, or the predicted CSI.

[0273] In this application, whether the encoder has a quantization function (or whether it includes a quantizer) can be predefined, such as predefined by a protocol, or it can also be indicated by the network side, or determined by the terminal side itself, and this is not limited.

[0274] In a possible implementation, the terminal side trains the decoder (an example of the second model, denoted as decoder 2) in the dual - end model 2 based on the first dataset, and trains the encoder 2 based on the decoder 2 obtained from the model training. The decoder 2 can be used to perform a second process on the CSI feedback information, and the second process at least includes decompression. For example, the second process is decompression, or the second process includes decompression and de - quantization.

[0275] The model training in this application can refer to initial training, model update, or fine - tuning. For example, the model parameters of the decoder 2 can be initialized parameters, and the terminal side can perform initial training on the decoder 2 based on the first dataset to update the initialized parameters to the trained model parameters. For another example, the decoder 2 can be pre - trained, and the terminal side can perform fine - tuning on the pre - trained decoder based on the first dataset.

[0276] Figure 15 It is a schematic diagram of the process of model training on the terminal side. As Figure 15 shown, the process of model training on the terminal side can include two stages, namely stage 1 and stage 2.

[0277] Stage 1: The terminal - side training device trains the decoder 2 alone based on the first dataset. The decoder 2 can be a nominal decoder. The terminal side uses the CSI feedback information (i.e., C2) and the second CSI (i.e., H2 NW ) in the first dataset as the input of the decoder 2, and uses the first CSI (i.e., H1 NW ) in the first dataset as the label to train the decoder 2. The decoder 2 can be used to decompress the CSI feedback information (i.e., C2) and output the reconstructed CSI (i.e., ). Through the training of the decoder 2, the output of the decoder 2 (i.e., ) is made to match the label (i.e., H1 NW)Tends to approach. At the output of decoder 2 and the label H1 NW When the performance metric is met, the terminal side can stop training the model of decoder 2, and thus the trained decoder 2 can be obtained.

[0278] It should be understood that if the CSI feedback information in the first dataset is obtained by compressing and quantizing the first CSI, this decoder 2 can be used to dequantize and decompress this CSI feedback information (i.e., C2).

[0279] Phase 2: The terminal side can fix the parameters of decoder 2 (i.e., the model parameters of decoder 2 obtained through training in Phase 1) (i.e., adopt the frozen decoder 2, in other words, do not change the parameters of decoder 2 during the process of training encoder 2), and use the input CSI (i.e., H UE ) as the input of encoder 2 to perform end-to-end training on encoder 2 and decoder 2, so that the output of decoder 2 (i.e., ) and the label (i.e., H UE ) tend to approach.

[0280] Optionally, the terminal side can fix the parameters of decoder 2, and use the input CSI (such as all or part of H UE ) as the input of encoder 2, and use the second CSI (such as H2 NW ) as the input of decoder 2 to perform end-to-end training on encoder 2 and decoder 2, so that the output of decoder 2 (i.e., ) and the label (i.e., H UE ) tend to approach.

[0281] At the output of decoder 2 and the H input to encoder 2 UE When the performance metric is met, the terminal side can stop training the model of encoder 2, and thus the trained encoder 2 can be obtained.

[0282] Exemplarily, the end-to-end training process is as follows: Input the second input CSI (i.e., H UE ) into encoder 2. Encoder 2 can be used to compress this second input CSI and output CSI feedback information; Use the output of encoder 2 as the input of decoder 2. Decoder 2 can be used to decompress the CSI feedback information output by encoder 2 and output the reconstructed CSI (i.e., ). Performing end-to-end training on encoder 2 and decoder 2 can also be referred to as jointly training encoder 2 and decoder 2.

[0283] It should be understood that if the encoder 2 is used to compress and quantize the input CSI, the CSI feedback information output by the encoder 2 is the quantized CSI. At this time, the decoder 2 can be used to dequantize and decompress the CSI feedback information.

[0284] It should also be understood that in the training phase 2, the input CSI used by the training device can be the first CSI in the first dataset from the network side (i.e., H UE can be H1 NW ), or the CSI obtained by the terminal side itself, and there is no limitation on this. The ways for the terminal side to obtain CSI include but are not limited to: measurement, simulation, training generation, etc., and there is no limitation.

[0285] It should be understood that the dual - end model on the network side or terminal side of the present application can also be referred to as an asymmetric dual - end model. In the asymmetric dual - end model, the input of the decoder is not exactly the same as the output of the encoder. Optionally, the label corresponding to the output of the encoder can be different from the input of the encoder.

[0286] Based on the above technical solutions, the terminal side can obtain a dataset for model training of the asymmetric dual - end model, and perform model training on the dual - end model on the terminal side based on this dataset. The CSI feedback information and the second CSI in this dataset can be used as the input of the decoder in the dual - end model, and the first CSI can be used as the label of the output of the decoder, which is beneficial to improving the decoding performance of the decoder. The terminal device can improve the encoding performance of the encoder by training the encoder in the dual - end model based on the decoder obtained through model training, thereby improving the feedback performance of the CSI.

[0287] It should be understood that the magnitudes of the sequence numbers of the above - mentioned processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0288] It should also be understood that in each embodiment of the present application, if there is no special description and logical conflict, the terms and / or descriptions between different embodiments are consistent and can be mutually referred to. The technical features in different embodiments can be combined to form new embodiments according to their internal logical relationships.

[0289] It should also be understood that in some of the above - mentioned embodiments, mainly the devices in the existing network architecture are used as examples for illustrative purposes. It should be understood that the specific form of the device is not limited in the embodiments of the present application. For example, devices that can achieve the same functions in the future are applicable to the embodiments of the present application.

[0290] It can be understood that in the above method embodiments, the methods and operations implemented by a device (such as a terminal device or a network device) can also be implemented by components of the device (such as a chip or a circuit).

[0291] Those skilled in the art should be able to realize that, for the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein, this application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed in the way of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described function for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0292] As above, the methods provided in the embodiments of this application have been described in detail in conjunction with multiple drawings. Below, the apparatuses provided in the embodiments of this application will be described in conjunction with the drawings.

[0293] Figure 16 and Figure 17 are schematic block diagrams of possible apparatuses provided in the embodiments of this application. These apparatuses can be used to implement the functions on the terminal side or the network side in the above method embodiments, and thus can also achieve the beneficial effects possessed by the above method embodiments.

[0294] Figure 16 is a schematic block diagram of the apparatus provided in the embodiment of this application. Figure 16 The illustrated apparatus 1600 may include a processing module 1910 and a communication module 1920.

[0295] In a possible design, the apparatus 1600 can be used to implement Figure 11 the communication method implemented by the terminal side in the illustrated embodiment. For example, the processing module 1610 is used to implement the processing-related steps executed by the terminal device in each method embodiment, such as performing model training and determining the first data set, etc.; the communication module 1620 is used to implement the sending and / or receiving and other steps executed by the terminal device in each method embodiment.

[0296] Exemplarily, the communication module 1620 is configured to: obtain a first data set, where each sample in the first data set includes the following information: first channel state information (CSI), CSI feedback information, and second CSI. The first CSI characterizes the channel state of a first resource set, the second CSI characterizes the channel state of a second resource set, the second resource set is part or all of the first resource set, the first resource set includes at least one of frequency-domain resources or spatial-domain resources, the CSI feedback information is obtained by compressing the first CSI or a third CSI, the third CSI characterizes the channel state of a third resource set, the third resource set is part of the first resource set, or the third resource set is different from the first resource set; the processing module 1610 is configured to: perform model training on a first model based on the first data set, and the first model is used to perform a first process on the CSI to be fed back, and the first process includes at least CSI compression.

[0297] Among them, for the first resource set, the second resource set, and the frequency-domain position relationship and / or corresponding time-domain resources between the first resource set and the second resource set, reference may be made to the descriptions in the foregoing text, and details are not described herein again.

[0298] Specifically, the communication module 1620 may be configured to: receive the first data set from the network side.

[0299] Optionally, the communication module 1620 is specifically configured to: receive a second data set and a third data set from the network side. The second data set includes multiple first CSIs, the multiple first CSIs correspond to multiple first time-domain resources, and the multiple first time-domain resources do not overlap with each other. The third data set includes multiple CSI feedback information, and one of the multiple first CSIs corresponds to one or more of the multiple CSI feedback information.

[0300] Optionally, the processing module 1610 is further configured to: determine the first data set based on the second data set and the third data set.

[0301] Specifically, the processing module 1610 may be configured to: determine that the first CSI included in each sample in the first data set is one of the multiple first CSIs, the second CSI included in each sample is determined according to the first CSI included in each sample and other first CSIs among the multiple first CSIs, and the CSI feedback information included in each sample is the CSI feedback information corresponding to the first CSI included in each sample among the multiple CSI feedback information.

[0302] Exemplarily, the second CSI included in each sample is determined by a part of the CSI in the first CSI included in each sample and a part of the CSI in the other first CSI. The time-domain resources corresponding to the other first CSI may be earlier than the time-domain resources corresponding to the first CSI included in each sample.

[0303] The processing module 1610 may specifically be configured to: perform model training on a second model based on the first data set, where the second model is used to perform a second process on the compressed CSI feedback information, and the second process at least includes CSI reconstruction; perform model training on the first model based on the input CSI and the second model obtained through model training.

[0304] For a more detailed description of the above processing module 1610 and communication module 1620, reference may be directly made to Figure 11 the relevant descriptions in the method embodiments shown, which will not be elaborated here.

[0305] In another possible design, the apparatus 1600 may be used to implement Figure 11 the communication method implemented by the network device in the shown embodiment. For example, the processing module 1610 is used to implement the processing-related steps performed by the network device in each method embodiment; the communication module 1620 is used to implement the sending and / or receiving and other steps performed by the network device in each method embodiment.

[0306] Exemplarily, the communication module 1620 is configured to: indicate a first data set, where the first data set is used to perform model training on a first model, the first model is used to perform a first process on the channel state information CSI to be fed back, and the first process at least includes CSI compression. Each sample in the first data set includes the following information: first CSI, CSI feedback information, and second CSI; where the first CSI characterizes the channel state of a first resource set, the second CSI characterizes the channel state of a second resource set, the second resource set is part or all of the first resource set, the first resource set includes at least one of frequency-domain resources or spatial-domain resources, the CSI feedback information is obtained by compressing the first CSI or a third CSI, the third CSI characterizes the channel state of a third resource set, the third resource set is part of the first resource set, or the third resource set is different from the first resource set.

[0307] For a more detailed description of the above processing module 1610 and communication module 1620, reference may be directly made to Figure 11 the relevant descriptions in the method embodiments shown, which will not be elaborated here.

[0308] It should be noted that the communication module can also be referred to as a transceiver module, a transceiver unit, a transceiver, a transceiver machine, or a transceiver device, etc. The processing module can also be referred to as a processor, a processing board, a processing unit, or a processing device, etc. Optionally, the communication module is used to perform the sending operation and receiving operation of the terminal device or the network device in the above method. The devices used to implement the receiving function in the communication module can be regarded as the receiving module, and the devices used to implement the sending function in the communication module can be regarded as the sending module. That is, the communication module can include a receiving module and a sending module.

[0309] It should also be noted that in a possible design, the foregoing processing module and / or communication module can be implemented through virtual modules. For example, the processing module can be implemented through a software functional unit or a virtual device, and the communication module can be implemented through a software function or a virtual device. In another possible design, the processing module or the communication module can also be implemented through an entity device. For example, if the device is implemented using a chip / chip circuit, the communication module can be an input / output circuit and / or a communication interface, which performs input operations (corresponding to the foregoing receiving operations) and output operations (corresponding to the foregoing sending operations); the processing module can be an integrated processor or a microprocessor or an integrated circuit.

[0310] The division of modules in the embodiments of the present application is illustrative, and is only a logical function division. In actual implementation, there may be other division methods. In addition, in each example of the embodiments of the present application, each functional module can be integrated in one processor, or can exist independently physically, or two or more modules can be integrated in one module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules.

[0311] Figure 17 It is a schematic structural diagram of a communication device provided in another embodiment of the present application. As Figure 17 shown, the device 1700 includes a processing circuit 1710 and a communication circuit 1720. The processing circuit 1710 and the communication circuit 1720 are coupled to each other.

[0312] It can be understood that the processing circuit 1710 can be one or more processors, or can be all or part of the circuits with processing functions in one or more processors.

[0313] It can be understood that the communication circuit 1720 can be a transceiver or an input / output interface.

[0314] Optionally, the device 2000 can also include a memory 1730, which is used to store the instructions executed by the processing circuit 1710 or store the input data required for the processing circuit 1710 to run the instructions or store the data generated after the processing circuit 1710 runs the instructions.

[0315] It can be understood that the memory 1730 can be located outside the processing circuit 1710 or inside the processing circuit 1710.

[0316] As an example, the processing circuit 1710 is used to implement the functions of the above-mentioned processing module 1610, and the communication circuit 1720 is used to implement the functions of the above-mentioned communication module 1620.

[0317] As an example, the device 1700 can be a communication device (such as a terminal device or a network device), or a chip applied to a communication device.

[0318] When the device 1700 is a communication device, the communication circuit can be a transceiver; when the device 1700 is a chip, the communication circuit can be an input / output circuit, a bus, a pin, or other types of communication interfaces. Among them, the input circuit in the input / output circuit can be used for receiving, and the output interface can be used for sending.

[0319] In a possible implementation manner, the device 1700 is used to implement each process and step corresponding to the terminal device in the above method embodiment. In another possible implementation manner, the device 1700 is used to implement each process and step corresponding to the network device in the above method embodiment.

[0320] It can be understood that the device 1700 can be specifically the terminal device or the network device in the above embodiment, or a chip or a chip system. Correspondingly, the communication circuit can be the interface circuit of the chip, or an input / output circuit, which is not limited here. Specifically, the device 1700 can be used to execute each step and / or process corresponding to the terminal device or the network device in the above method embodiment.

[0321] When the above communication device is a chip or an OTT device applied to a terminal device, the chip or the OTT device of the terminal device implements the functions of the terminal device in the above method embodiment. For example, it implements the processing function of the terminal device. The chip or the OTT device of the terminal device receives information from a network device. It can be understood that the information is first received by other modules (such as a radio frequency module or an antenna) in the terminal device, and then sent by these modules to the chip or the OTT device of the terminal device. The chip or the OTT device of the terminal device sends information to the network device. It can be understood that the information is first sent by the chip or the OTT device of the terminal device to other modules (such as a radio frequency module or an antenna) in the network device, and then sent by these modules to the network device.

[0322] When the above communication device is a chip applied to a network device or an OTT device, the chip of the network device or the OTT device implements the functions of the network device in the above method embodiments. For example, it implements the processing function of the network device. The chip of the network device or the OTT device receives information from the terminal device. It can be understood that the information is first received by other modules (such as a radio frequency module or an antenna) in the network device and then sent by these modules to the chip of the network device or the OTT device. The chip of the network device or the OTT device sends information to the terminal device. It can be understood that the information is first sent by the chip of the network device or the OTT device to other modules (such as a radio frequency module or an antenna) in the network device and then sent by these modules to the terminal device.

[0323] The present application also provides a computer program product. When the computer program product runs on a processor, it can implement the communication method executed by the terminal device or the communication method executed by the network device in the above method embodiments.

[0324] The present application also provides a computer-readable storage medium. The computer-readable storage medium contains computer instructions. When the computer instructions run on a processor, they can implement the communication method executed by the terminal device or the communication method executed by the network device in the above method embodiments.

[0325] The present application also provides a communication system, including the foregoing terminal device and network device.

[0326] It can be understood that the processor in the embodiments of the present application may be the following device or all or part of the circuits for processing functions in the following devices: a central processing unit (CPU), and may also be other general-purpose processors, a digital signal processor (DSP), a processor for AI, a field programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The general-purpose processor may be a microprocessor or any conventional processor.

[0327] Among them, by way of example, the processor for AI may be one or more of a graphics processing unit (GPU), a neural processing unit (NPU), a tensor processing unit (TPU), and a data processing unit (DPU).

[0328] Exemplarily, a possible implementation of a processor for AI may be Figure 18 the AI processor 1500 shown.

[0329] As Figure 18 shown, the AI processor 1800 may include one or more of the following: an AI core, a digital vision pre-processing (DVPP) module, a task scheduler (TS), an L3 cache, an AI central processing unit (CPU), a control CPU, an L2 cache, a universal serial bus (USB) interface, a network card, a peripheral component interconnect express (PCIe) interface (PCIe is a high-speed serial computer expansion bus standard), a double data rate (DDR) / high bandwidth memory (HBM) interface, a general purpose input / output (GPIO) / inter-integrated circuit (I2C), etc. It can be understood that the specific meanings of these terms are well-known to those skilled in the art and will not be elaborated here.

[0330] The terms "unit", "module", etc. used in this specification may be used to represent computer-related entities, hardware, firmware, combinations of hardware and software, software, or software in execution.

[0331] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of this application.

[0332] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here.

[0333] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0334] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0335] In addition, in each embodiment of the present application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0336] In the above embodiments, the functions of each functional unit 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 (programs). When the computer program instructions (programs) are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. 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 wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a digital video disc (DVD)), or a semiconductor medium (such as a solid-state disk (SSD)), etc.

[0337] If the above function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0338] As described above, the above are only the specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A communication method, characterized in that Including: Obtain a first data set, where each sample in the first data set includes the following information: first channel state information (CSI), CSI feedback information, and second CSI. The first CSI characterizes the channel state of a first resource set, the second CSI characterizes the channel state of a second resource set, the second resource set is part or all of the first resource set, the first resource set includes at least one of frequency domain resources or spatial domain resources, the CSI feedback information is obtained by compressing the first CSI or a third CSI, the third CSI characterizes the channel state of a third resource set, the third resource set is part of the first resource set, or the third resource set is different from the first resource set; Wherein, the first data set is used for training a first model, and the first model is used for performing a first process on the CSI to be fed back, and the first process at least includes CSI compression.

2. The method according to claim 1, characterized in that, The first resource set corresponds to a first time domain resource, the second resource set corresponds to a second time domain resource and / or the first time domain resource, the second time domain resource is earlier than the first time domain resource, and the third resource set corresponds to the first time domain resource.

3. The method according to claim 2, wherein The second resource set includes a first part and a second part, the first part corresponds to the second time domain resource, and the second part corresponds to the first time domain resource.

4. The method according to any one of claims 1 to 3, characterized in that, The obtaining of the first data set includes: Receiving the first data set from the network side.

5. The method according to any one of claims 1 to 3, characterized in that The method further includes: Receiving a second data set and a third data set from the network side. The second data set includes a plurality of first CSIs, the plurality of first CSIs correspond to a plurality of first time domain resources, and the plurality of first time domain resources do not overlap with each other. The third data set includes a plurality of CSI feedback information, and one of the plurality of first CSIs corresponds to one or more of the plurality of CSI feedback information.

6. The method according to claim 5, wherein The obtaining of the first data set includes: Determining the first data set based on the second data set and the third data set.

7. The method according to claim 6, characterized in that, The determining of the first data set based on the second data set and the third data set includes: Determining that the first CSI included in each sample in the first data set is one of the plurality of first CSIs, the second CSI included in each sample is determined according to the first CSI included in each sample and the other first CSIs among the plurality of first CSIs, and the CSI feedback information included in each sample is the CSI feedback information corresponding to the first CSI included in each sample among the plurality of CSI feedback information.

8. The method according to claim 7, characterized in that The second CSI included in each sample is determined by partial CSIs in the first CSI included in each sample and partial CSIs in the other first CSIs.

9. The method according to claim 7 or 8, characterized in that The time domain resource corresponding to the other first CSI is earlier than the time domain resource corresponding to the first CSI included in each sample.

10. The method according to any one of claims 1 to 9, characterized in that The first data set being used for training a first model includes: The first data set is used for training a second model, and the second model is used for performing a second process on the compressed CSI feedback information. The second process includes at least CSI reconstruction. The second model obtained through model training is used for training the first model.

11. A communication method, characterized in that Comprising: Indicating a first data set, where the first data set is used for training a first model. The first model is used for performing a first process on the channel state information CSI to be fed back. The first process includes at least CSI compression. Each sample in the first data set includes the following information: first CSI, CSI feedback information, and second CSI; Wherein, the first CSI characterizes the channel state of a first resource set, the second CSI characterizes the channel state of a second resource set, and the second resource set is part or all of the first resource set. The first resource set includes at least one of frequency domain resources or spatial domain resources. The CSI feedback information is obtained by compressing the first CSI or a third CSI, and the third CSI characterizes the channel state of a third resource set. The third resource set is part of the first resource set, or the third resource set is different from the first resource set.

12. The method according to claim 11, wherein The first resource set corresponds to a first time domain resource, the second resource set corresponds to a second time domain resource and / or the first time domain resource, the second time domain resource is earlier than the first time domain resource, and the third resource set corresponds to the first time domain resource.

13. The method according to claim 12, wherein The second resource set includes a first part and a second part. The first part corresponds to the second time domain resource, and the second part corresponds to the first time domain resource.

14. The method according to any one of claims 11 to 13, characterized in that, The indicating the first data set includes: Sending the first data set.

15. The method according to any one of claims 11 to 13, characterized in that The indicating the first data set includes: Sending a second data set and a third data set. The second data set and the third data set are used to determine the first data set. The second data set includes a plurality of first CSI, and the plurality of first CSI correspond to a plurality of first time domain resources. The plurality of first time domain resources do not overlap with each other. The third data set includes a plurality of CSI feedback information, and one of the plurality of first CSI corresponds to one or more of the plurality of CSI feedback information.

16. The method according to claim 15, wherein The first CSI included in each sample in the first data set is one of the plurality of first CSI. The second CSI included in each sample is determined according to the first CSI included in each sample and the other first CSI among the plurality of first CSI. The CSI feedback information included in each sample is the CSI feedback information corresponding to the first CSI included in each sample among the plurality of CSI feedback information.

17. The method according to claim 16, characterized in that, The second CSI included in each sample is determined by a part of the CSI in the first CSI included in each sample and a part of the other first CSI.

18. The method according to claim 16 or 17, characterized in that The time domain resource corresponding to the other first CSI is earlier than the time domain resource corresponding to the first CSI included in each sample.

19. A communication device, characterized in that, Including functional modules for implementing the method according to any one of claims 1 to 18.

20. A communication device, characterized in that, Comprising one or more processors and communication circuitry, the communication circuitry being configured for at least one of input or output of signals by the communication device; the one or more processors being configured to implement the method according to any one of claims 1 to 18.

Images