Channel state transmission method and apparatus, channel state acquisition method and apparatus, terminal, and network device

Abstract

The present invention provides a channel state transmission method and apparatus, a channel state acquisition method and apparatus, a terminal, and a network device. The method comprises: acquiring a plurality of pieces of first downlink channel state information (CSI), each piece of first downlink CSI corresponding to one network device; compressing the plurality of pieces of first downlink CSI to acquire target information; and sending the target information to at least two network devices, the at least two network devices providing coherent joint transmission to a terminal.

Description

Channel state transmission, acquisition method and device, terminal and network equipment

[0001] The present disclosure claims priority to the Chinese patent publication with the publication number 202411094088.8 and the title "Channel state transmission, acquisition method and device, terminal and network equipment" filed on August 9, 2024 with the Chinese Patent Office, the entire contents of which are hereby incorporated by reference. TECHNICAL FIELD

[0002] The present disclosure relates to the technical field of communication, in particular to a channel state transmission, acquisition method, device, terminal and network equipment. BACKGROUND

[0003] In a distributed multiple-in multiple-out (MIMO) (for example, distributed MIMO, or cell-free massive MIMO, or cooperative communication) system, multiple base stations or transmit and receive points (TRPs) or access points (APs) serve one user equipment (UE, also referred to as a terminal). In order to perform coherent joint transmission (CJT), the mutually cooperating base stations need to obtain the downlink channel state information (CSI) of each cooperating base station to the UE. The current acquisition method of the downlink CSI of each cooperating base station to the UE has the problems of large delay or large signaling overhead. SUMMARY

[0004] Embodiments of the present disclosure provide a channel state transmission, acquisition method, device, terminal and network equipment to solve the problem of large delay or large signaling overhead in the acquisition method of the downlink CSI of each cooperating base station to the UE.

[0005] To solve the above technical problem, the present disclosure provides a channel state information transmission method applied to a terminal, comprising:

[0006] Obtaining a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to one network device;

[0007] Compressing the plurality of first downlink CSIs to obtain target information;

[0008] The target information is sent to at least two network devices, and the at least two network devices provide coherent joint transmission for the terminal.

[0009] In some embodiments, the target information is obtained by compressing the plurality of first downlink CSIs.

[0010] The plurality of first downlink CSIs are input into a first target model to obtain target information output by the first target model.

[0011] The first target model is a double-layer compression model, a first-layer compression model compresses a plurality of first downlink CSIs respectively, and a second-layer compression model compresses a plurality of second downlink CSIs output by the first-layer compression model; the first target model is obtained by training first training data, and the first training data includes at least one of the following: third downlink CSIs corresponding to a plurality of network devices respectively, a number of transmitting antenna ports of the plurality of network devices, a number of subbands occupied by signals transmitted by the plurality of network devices, and label data information.

[0012] The third downlink CSI is a downlink CSI used for model training.

[0013] In some embodiments, the first target model is a model obtained by training a double-sided model, or the first target model is a model obtained by training a single-sided model.

[0014] In some embodiments, the double-sided model includes at least one of the following:

[0015] A self-encoder model, a convolutional neural network model, a generative adversarial network model, and a Transformer structure model.

[0016] In some embodiments, when the first target model is a model obtained by training a double-sided model, the method further includes at least one of the following:

[0017] The second target model is sent to a plurality of network devices, the second target model is a model used for decompressing the target information, and the second target model is obtained by training first training data by a terminal.

[0018] A first target model sent by a network device is received.

[0019] A preconfigured first target model is obtained.

[0020] In some embodiments, when the first target model is a model obtained by training a double-sided model, the label data information included in the first training data is third downlink CSIs corresponding to a plurality of network devices respectively.

[0021] In some embodiments, in a case where the first target model is a model trained by a one-sided model, the label data information included in the first training data comprises first label data and second label data, the first label data is information obtained by compressing third downlink CSI of a plurality of network devices respectively by using a compression sensing principle, and the second label information is information obtained by compressing the first label data information by using a compression sensing principle.

[0022] In some embodiments, the sending of the target information to the at least two network devices comprises one of the following:

[0023] sending the target information to the at least two network devices by a directional beam;

[0024] sending the target information to the at least two network devices by an omnidirectional antenna.

[0025] In some embodiments, the method further comprises:

[0026] receiving indication information sent by a network device, the indication information being used to indicate whether the terminal jointly compresses and jointly feeds back first downlink CSI from a plurality of network devices in a cooperative communication mode;

[0027] The compression of the plurality of first downlink CSI to obtain target information comprises:

[0028] In a case where the indication information indicates that the terminal jointly compresses and jointly feeds back the first downlink CSI from the plurality of network devices in the cooperative communication mode, the plurality of first downlink CSI are compressed to obtain the target information.

[0029] In some embodiments, the method further comprises:

[0030] sending, to the at least two network devices, identification information of a network device associated with the target information.

[0031] The embodiments of the present disclosure further provide a channel state information acquisition method applied to a network device, comprising:

[0032] receiving target information sent by a terminal, the target information being obtained by compressing a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to one network device;

[0033] decompressing the target information to obtain a plurality of decompressed first downlink CSI.

[0034] In some embodiments, the decompression of the target information to obtain a plurality of decompressed first downlink CSI comprises:

[0035] inputting the target information into a second target model, and obtaining a plurality of decompressed first downlink CSIs output by the second target model;

[0036] The second target model is a double-layer decompression model, a first layer decompression model decompresses the target information to obtain fourth downlink CSIs, and a second layer of a plurality of decompression models respectively decompress the fourth downlink CSIs.

[0037] The second target model is trained by a bilateral model and first training data, or the second target model is trained by a unilateral model and second training data. The second training data includes training data of the first layer decompression model, label data of the first layer decompression model, training data of the second layer decompression model, and label data of the second layer decompression model. The training data of the first layer decompression model is second label data information, the label data of the first layer decompression model is first label data information, the training data of the second layer decompression model is the first label data information, and the label data of the second layer decompression model is third downlink CSIs from a plurality of network devices.

[0038] The first training data includes at least one of third downlink CSIs corresponding to a plurality of network devices respectively, a number of transmitting antenna ports of the plurality of network devices, a number of subbands occupied by signals transmitted by the plurality of network devices, and label data information.

[0039] The first label data is information obtained by compressing third downlink CSIs of a plurality of network devices respectively by using a compression sensing principle, and the second label information is information obtained by compressing the first label data information by using the compression sensing principle.

[0040] In some embodiments, the bilateral model includes at least one of:

[0041] a self-encoder model, a convolutional neural network model, a generative adversarial network model, and a Transformer structure model.

[0042] In some embodiments, when the second target model is trained by a bilateral model and first training data, the method further includes at least one of:

[0043] sending the first target model to a terminal, the first target model being a model for compressing a plurality of first downlink CSIs, and the first target model being obtained by training using first training data by a network device;

[0044] receiving a second target model sent by a terminal.

[0045] In some embodiments, the method further comprises:

[0046] sending indication information to the terminal, the indication information being used to indicate whether the terminal jointly compresses and jointly feeds back the first downlink CSI from the plurality of network devices in the cooperative communication mode.

[0047] In some embodiments, the method further comprises one of:

[0048] receiving identification information of the network device associated with the target information sent by the terminal;

[0049] receiving identification information of the network device providing cooperative communication for the terminal sent by the control center.

[0050] The embodiments of the present disclosure also provide a terminal, comprising a memory, a transceiver, and a processor:

[0051] The memory is used to store a computer program; the transceiver is used to transceive data under the control of the processor; and the processor is used to read the computer program in the memory and perform the following operations:

[0052] obtaining a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to a network device;

[0053] compressing the plurality of first downlink CSI to obtain target information;

[0054] sending the target information to at least two network devices through the transceiver, the at least two network devices providing coherent joint transmission for the terminal.

[0055] In some embodiments, the processor is used to read the computer program in the memory and perform the following operations:

[0056] inputting the plurality of first downlink CSI into a first target model to obtain target information output by the first target model;

[0057] The first target model is a double-layer compression model, a first layer compression model compresses a plurality of first downlink CSI respectively, and a second layer compression model compresses a plurality of second downlink CSI output by the first layer compression model; the first target model is obtained by training first training data, the first training data comprising at least one of third downlink CSI corresponding to the plurality of network devices respectively, a number of transmitting antenna ports of the plurality of network devices, a number of subbands occupied by signals transmitted by the plurality of network devices, and label data information.

[0058] The third downlink CSI is a downlink CSI used for model training.

[0059] In some embodiments, the first target model is a model trained by a bilateral model, or the first target model is a model trained by a unilateral model.

[0060] In some embodiments, the bilateral model comprises at least one of:

[0061] a self-encoder model, a convolutional neural network model, a generative adversarial network model, a Transformer structure model.

[0062] In some embodiments, in the case where the first target model is a model trained by a bilateral model, the processor for reading the computer program in the memory further performs at least one of the following operations:

[0063] sending a second target model to a plurality of network devices, the second target model being a model for decompressing the target information, the second target model being obtained by training the first training data by a terminal;

[0064] receiving a first target model sent by a network device;

[0065] obtaining a preconfigured first target model.

[0066] In some embodiments, in the case where the first target model is a model trained by a bilateral model, the label data information included in the first training data is third downlink CSI corresponding to the plurality of network devices respectively.

[0067] In some embodiments, in the case where the first target model is a model trained by a unilateral model, the label data information included in the first training data comprises first label data and second label data, the first label data being information obtained by compressing third downlink CSI of the plurality of network devices respectively using a compression sensing principle, and the second label information being information obtained by compressing the first label data information using a compression sensing principle.

[0068] In some embodiments, the processor for reading the computer program in the memory performs one of the following operations:

[0069] sending target information to at least two network devices through directional beams;

[0070] sending target information to at least two network devices through an omnidirectional antenna.

[0071] In some embodiments, the processor for reading the computer program in the memory further performs the following operation:

[0072] receive, by the transceiver, indication information sent by a network device, the indication information being used to indicate whether the terminal jointly compresses and feeds back first downlink channel state information (CSI) from multiple network devices in a cooperative communication mode;

[0073] In a case where the indication information indicates that the terminal jointly compresses and feeds back the first downlink CSI from the multiple network devices in the cooperative communication mode, the multiple first downlink CSI are compressed to obtain target information.

[0074] In some embodiments, the processor, for reading the computer program in the memory, further performs the following operations:

[0075] sending, by the transceiver, identification information of a network device associated with the target information to at least two network devices.

[0076] The embodiments of the present disclosure further provide a network device, comprising a memory, a transceiver, and a processor:

[0077] The memory is configured to store a computer program; the transceiver is configured to transceive data under the control of the processor; and the processor is configured to read the computer program in the memory and perform the following operations:

[0078] receiving, by the transceiver, target information sent by a terminal, the target information being obtained by compressing multiple first downlink channel state information (CSI), each first downlink CSI corresponding to one network device;

[0079] decompressing the target information to obtain multiple decompressed first downlink CSI.

[0080] In some embodiments, the processor, for reading the computer program in the memory, further performs the following operations:

[0081] inputting the target information into a second target model to obtain multiple decompressed first downlink CSI output by the second target model;

[0082] The second target model is a double-layer decompression model, a first layer decompression model decompresses the target information to obtain fourth downlink CSI, and multiple second layer decompression models decompress the fourth downlink CSI respectively.

[0083] The second target model is trained by a bilateral model and first training data; or the second target model is trained by a unilateral model and second training data, the second training data including: first-layer decompression model training data, first-layer decompression model label data, second-layer decompression model training data, and second-layer decompression model label data, the first-layer decompression model training data being second label data information, the first-layer decompression model label data being first label data information, the second-layer decompression model training data being the first label data information, and the second-layer decompression model label data being third downlink CSI from multiple network devices.

[0084] The first training data includes at least one of: third downlink CSI corresponding to multiple network devices respectively, the number of sending antenna ports of multiple network devices, the number of subbands occupied by signals sent by multiple network devices, and label data information.

[0085] The first label data is information obtained by compressing third downlink CSI of multiple network devices respectively by using a compression sensing principle, and the second label information is information obtained by compressing the first label data information by using a compression sensing principle.

[0086] In some embodiments, the bilateral model includes at least one of:

[0087] A self-encoder model, a convolutional neural network model, a generative adversarial network model, and a Transformer structure model.

[0088] In some embodiments, in the case where the second target model is trained by a bilateral model and first training data, the processor for reading the computer program in the memory further performs at least one of the following operations:

[0089] The first target model is sent to a terminal, the first target model being a model for compressing multiple first downlink CSIs, and the first target model being obtained by training a network device by using first training data;

[0090] The second target model sent by the terminal is received.

[0091] In some embodiments, the processor for reading the computer program in the memory further performs the following operation:

[0092] The transceiver sends indication information to the terminal, the indication information being used to indicate whether the terminal jointly compresses and jointly feeds back first downlink CSI from multiple network devices in a cooperative communication mode.

[0093] In some embodiments, the processor, configured to read the computer program in the memory, is further configured to perform one of the following operations:

[0094] receive, by the transceiver, identification information of a network device associated with the target information sent by the terminal;

[0095] receive, by the transceiver, identification information of a network device providing cooperative communication for the terminal sent by the control center.

[0096] The embodiments of the present disclosure further provide a channel state information transmission apparatus applied to a terminal, comprising:

[0097] a first obtaining unit configured to obtain a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to one network device;

[0098] a second obtaining unit configured to compress the plurality of first downlink CSI to obtain target information;

[0099] a first sending unit configured to send the target information to at least two network devices, the at least two network devices providing coherent joint transmission for the terminal.

[0100] The embodiments of the present disclosure further provide a channel state information obtaining apparatus applied to a network device, comprising:

[0101] a first receiving unit configured to receive target information sent by a terminal, the target information being obtained by compressing a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to one network device;

[0102] a third obtaining unit configured to decompress the target information to obtain a plurality of decompressed first downlink CSI.

[0103] The embodiments of the present disclosure further provide a processor-readable storage medium, which stores a computer program, the computer program being configured to enable the processor to perform the above method.

[0104] The embodiments of the present disclosure further provide a computer program product, which comprises computer instructions, the computer instructions being configured to enable the processor to perform the steps of the above method when executed.

[0105] The present disclosure has the following beneficial effects:

[0106] The above scheme compresses a plurality of first downlink CSI to obtain target information, and sends the target information to at least two network devices, thereby reducing the acquisition delay of downlink CSI of each cooperative network device to the UE and saving signaling overhead. BRIEF DESCRIPTION OF DRAWINGS

[0107] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure or the related art, the following will briefly introduce the drawings needed to be used in the embodiments or related art description. Obviously, the drawings in the following description are only some embodiments of the present disclosure, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0108] FIG. 1 shows a structural diagram of a network system suitable for embodiments of the present disclosure;

[0109] FIG. 2 shows a flow diagram of a channel state information transmission method according to an embodiment of the present disclosure;

[0110] FIG. 3 shows a schematic diagram of a double-layer compression AI model structure deployed in a terminal;

[0111] FIG. 4 shows a schematic diagram of a double-layer decompression AI model structure deployed in each cooperative network device;

[0112] FIG. 5 shows a schematic diagram of CSI joint compression and feedback;

[0113] FIG. 6 shows another schematic diagram of CSI joint compression and feedback;

[0114] FIG. 7 shows a flow diagram of a channel state information acquisition method according to an embodiment of the present disclosure;

[0115] FIG. 8 shows a schematic diagram of a channel state information transmission device according to an embodiment of the present disclosure;

[0116] FIG. 9 shows a structural diagram of a terminal according to an embodiment of the present disclosure;

[0117] FIG. 10 shows a schematic diagram of a channel state information acquisition device according to an embodiment of the present disclosure;

[0118] FIG. 11 shows a structural diagram of a network device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0119] The technical solutions in the embodiments of the present disclosure will be described clearly and completely below with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only some embodiments of the present disclosure, not all embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present disclosure.

[0120] The terms "first", "second", and the like in the description and in the claims of this disclosure are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the use of the terms so construed can be interchanged, such that the embodiments of the disclosure described herein are capable of attaining the same result irrespective of the different expressed order. Moreover, the terms "comprising" and "including" and any variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, system, product, or apparatus that comprises a list of steps or elements not expressly listed is not excluded from use or adoption of such a process, method, system, product, or apparatus.

[0121] The term "and / or", used in the embodiments of the disclosure, describes an associated relationship with associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. The character " / " generally represents an "or" relationship between the associated objects. The term "a plurality of" in the embodiments of the disclosure means two or more, and other quantifiers are similar.

[0122] In the embodiments of the disclosure, the words "exemplary" or "for example" are used to mean serving as an example, instance, or illustration. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of the disclosure should not be construed as being more preferred or advantageous than other embodiments or design schemes. Rather, the use of the words "exemplary" or "for example" is intended to present concepts in a concrete manner.

[0123] The related concepts mentioned in the disclosure are briefly described as follows.

[0124] The current scheme for obtaining downlink CSI of each cooperative base station has the following disadvantages:

[0125] Case 1: The terminal uses a directional antenna or a narrower transmission beam to feed back the estimated downlink CSI to the corresponding base station respectively, and then each cooperative base station obtains the downlink CSI between the other cooperative base stations and the terminal through the backhaul. When the terminal uses a directional antenna or a narrower transmission beam, each downlink CSI is transmitted to each cooperative base station through space division, and the terminal can use the same time-frequency resource to find the downlink CSI. At this time, the time-frequency resource occupied by the feedback downlink CSI is the smallest. However, since each cooperative base station needs to obtain the downlink CSI between the other cooperative base stations and the terminal through the backhaul, it brings a large transmission delay.

[0126] Case 2, the terminal uses an omnidirectional antenna to respectively feed back the estimated downlink CSI to the corresponding base station in a time-division or frequency-division manner, and then each cooperative base station obtains the downlink CSI of other cooperative base stations to the terminal through a backhaul link. When the terminal uses an omnidirectional antenna, each cooperative base station needs to use a time-division or frequency-division manner to respectively send the downlink CSI to the base station, and the time-frequency resources used are N times those of Case 1; but in Case 2, if each base station coordinates the time-frequency resources for receiving the downlink CSI, each cooperative base station can also receive the downlink CSI of other cooperative base stations, at this time, Case 2 evolves into the following Case 4; Case 4 uses time-frequency resources N times those of Case 1 for downlink CSI feedback, but does not bring additional time delay.

[0127] Case 3, the terminal uses a directional antenna or a narrower transmission beam to send the downlink CSI from all cooperative base stations to all cooperative base stations. When the terminal uses a directional antenna or a narrower transmission beam, since the downlink CSI of all cooperative base stations is included in each transmission beam, the time-frequency resources occupied are N times those of Case 1, but do not bring additional time delay.

[0128] Case 4, the terminal uses an omnidirectional antenna to send the downlink CSI from all cooperative base stations to all cooperative base stations. The time-frequency resources used for downlink CSI feedback are N times those of Case 1, but do not bring additional time delay.

[0129] Embodiments of the present disclosure are described below with reference to the accompanying drawings. The channel state transmission and acquisition method, device, terminal and network device provided by the embodiments of the present disclosure can be applied in a wireless communication system. The wireless communication system can be a system using a fifth generation (5th Generation, 5G) mobile communication technology (hereinafter referred to as a 5G system), and those skilled in the art can understand that the 5G NR system is only an example and is not limited.

[0130] Referring to FIG. 1, FIG. 1 is a structural diagram of a network system to which embodiments of the present disclosure can be applied. As shown in FIG. 1, the network system includes a user terminal 11 and a base station 12. The user terminal 11 can be a user equipment (UE), such as a mobile phone, a tablet personal computer (Tablet PC), a laptop computer, a personal digital assistant (PDA), a mobile Internet device (MID), or a wearable device. It should be noted that the specific type of the user terminal 11 is not limited in the embodiments of the present disclosure. The base station 12 can be a base station of a 5G and later version (for example, a gNB or a 5G NR NB), or a base station of another communication system, or a node B. It should be noted that the base station 12 is taken as an example of a 5G base station in the embodiments of the present disclosure, but the specific type of the base station 12 is not limited.

[0131] Embodiments of the present disclosure provide a channel state transmission method and device, a terminal and a network device to solve the problem of large delay and signaling overhead in the acquisition of downlink CSI from each cooperative base station to a UE.

[0132] The method and the device are based on the same disclosure concept. Since the principles of solving problems are similar, the implementation of the device and the method can be referred to each other, and the repeated parts will not be described here.

[0133] As shown in FIG. 2, the present disclosure provides a channel state information transmission method, which is executed by a terminal and includes the following steps.

[0134] In step S201, a plurality of first downlink channel state information (CSI) is acquired, and each first downlink CSI corresponds to one network device.

[0135] In step 202, the plurality of first downlink CSI is compressed to obtain target information.

[0136] In step 203, the target information is sent to at least two network devices, and the at least two network devices provide coherent joint transmission for the terminal.

[0137] It should be noted that, in the embodiments of the present disclosure, the plurality of first downlink CSI is compressed to obtain target information, and the target information is sent to at least two network devices, thereby reducing the acquisition delay of downlink CSI from each cooperative network device to a UE and saving signaling overhead.

[0138] It should be noted that the embodiments of the present disclosure are applied to a distributed MIMO, or a cell-free massive MIMO, or a cooperative communication system, and a scenario in which multiple network devices serve one terminal. In some embodiments, the network device in the embodiments of the present disclosure can also be referred to as a cooperative network device. In some embodiments, the network device can be a base station, a TRP, or an AP.

[0139] In some embodiments, each cooperative network device transmits a network device-specific downlink channel state information reference signal (CSI-RS), and a terminal receives the downlink CSI-RS transmitted by each cooperative network device. The terminal estimates the downlink CSI from each cooperative network device to the terminal based on the received downlink CSI-RS transmitted by each cooperative network device.

[0140] In some embodiments, in one implementation, the compression of the plurality of first downlink CSIs and the obtaining of the target information include:

[0141] inputting the plurality of first downlink CSIs into a first target model, and obtaining target information output by the first target model;

[0142] The first target model is a double-layer compression model. A first layer compression model compresses the plurality of first downlink CSIs respectively, and a second layer compression model compresses the plurality of second downlink CSIs output by the first layer compression model. The first target model is obtained by training first training data. The first training data includes at least one of the following: third downlink CSIs corresponding to the plurality of network devices respectively, the number of transmission antenna ports of the plurality of network devices, the number of subbands occupied by the signals transmitted by the plurality of network devices, and label data information.

[0143] The third downlink CSI is a downlink CSI used for model training.

[0144] It should be noted that the third downlink CSI mentioned here refers to a downlink CSI used for model training, and the first downlink CSI refers to a downlink CSI measured and obtained by a terminal in actual application. It is worth noting that both the first downlink CSI and the third downlink CSI are obtained by measuring a CSI-RS transmitted by a network device. The difference between the first downlink CSI and the third downlink CSI is that the two kinds of downlink CSIs are downlink CSIs measured and obtained at different times by different CSI-RSs.

[0145] In some embodiments, the first target model is a model trained by a bilateral model, or the first target model is a model trained by a unilateral model.

[0146] It should be noted that the bilateral model can be understood as joint training of a compression model and a decompression model, that is, the compression model and the decompression model can be determined by training on one side, and the unilateral model refers to that the compression model and the decompression model need to be trained independently.

[0147] In some embodiments, the bilateral model includes at least one of the following:

[0148] a self-encoder model, a convolutional neural network model, a generative adversarial network (GAN) model, and a Transformer structure model.

[0149] In some embodiments, in the case where the first target model is a model trained by a bilateral model, the method further includes at least one of the following:

[0150] A11, sending a second target model to a plurality of network devices, the second target model being a model for decompressing the target information, the second target model being obtained by training the first training data by a terminal;

[0151] In some embodiments, this case can be understood as that the model training is performed on the terminal side, and after the terminal side completes the training, the model for decompression is sent to the network device.

[0152] A12, receiving a first target model sent by a network device;

[0153] In some embodiments, this case can be understood as that the model training is performed on the network device side, and after the network device side completes the training, the model for compression is sent to the terminal.

[0154] A13, obtaining a preconfigured first target model;

[0155] In some embodiments, this case can be understood as that the model training is performed on a device other than the terminal and the network device, for example, the model is trained by a terminal manufacturer, and after the terminal manufacturer trains the first target model and the second target model, the first target model is sent to the terminal, and the second target model is sent to the network device. This case can be understood as that the model is a preconfigured model.

[0156] It should be noted that in the case of the model trained by the first target model being a double-sided model, in some embodiments, the first training data at least includes: third downlink CSI corresponding to the plurality of network devices respectively, the number of transmission antenna ports of the plurality of network devices, the number of subbands occupied by the plurality of network devices to transmit signals, and label data information. The label data information in this case is the third downlink CSI corresponding to the plurality of network devices respectively.

[0157] In some embodiments, in the case of the model trained by the first target model being a single-sided model, the first training data includes: third downlink CSI corresponding to the plurality of network devices respectively, the number of transmission antenna ports of the plurality of network devices, the number of subbands occupied by the plurality of network devices to transmit signals, and label data information. In some embodiments, the label data information includes first label data and second label data, the first label data is information obtained by compressing the third downlink CSI of the plurality of network devices respectively using the principle of compressed sensing, and the second label information is information obtained by compressing the first label data information using the principle of compressed sensing.

[0158] The specific applications of the double-sided model and the single-sided model are described in detail as follows.

[0159] I. The first target model is a model trained by a double-sided model

[0160] In some embodiments, the design principle of the double-sided model is:

[0161] In order to fully utilize the common sparsity characteristics of CSI between different cooperative network devices in distributed MIMO, the embodiments of the present disclosure design a double-layer compressed artificial intelligence (AI) model structure: the first layer is used to compress the downlink CSI from each cooperative network device respectively, and the second layer is used to jointly compress the compressed downlink CSI from each cooperative network device to remove the correlation between the downlink CSI of each cooperative network device, wherein the structure of the double-layer compressed model is shown in FIG. 3, wherein the first layer can include a plurality of AI compression models corresponding to the downlink CSI of each cooperative network device respectively, and the second layer includes an AI compression model for jointly compressing the output results of the plurality of first layer AI models to obtain compressed downlink CSI.

[0162] In some embodiments, the joint compression referred to in the embodiments of the present disclosure means that the downlink CSI from a plurality of cooperative network devices is compressed together, the purpose being to not only remove the correlation within the downlink CSI from each cooperative network device, but also to remove the correlation between the downlink CSI from a plurality of cooperative network devices.

[0163] Corresponding to the double-layer compression model, the double-layer decompression AI model also contains a two-layer structure, the first layer is used to decompress the jointly compressed data, and the second layer is used to decompress the downlink CSI of each cooperative network device respectively, and the specific structure is shown in Figure 4, wherein the first layer includes an AI decompression model, which decompresses the jointly compressed information of each cooperative network device, and the second layer includes multiple AI decompression models, which decompress the compressed downlink CSI of each cooperative network device corresponding to each cooperative network device after decompressing the jointly compressed information of each cooperative network device, to obtain the downlink CSI corresponding to each cooperative network device.

[0164] Specifically, the acquisition process of the double-layer compression / decompression AI model specifically includes the following processes:

[0165] 1. Input of the model

[0166] The downlink CSI estimated by the terminal from each cooperative network device, the number of transmission antenna ports of each cooperative network device, and the number of subbands occupied by the transmission signal.

[0167] 2. Output of the model

[0168] The jointly compressed CSI from each cooperative network device.

[0169] 3. Data collection

[0170] In some embodiments, the data collection process when the model is trained on the network device side includes one of the following methods:

[0171] Method 1 mainly includes:

[0172] The terminal reports the corresponding uncompressed downlink CSI to each cooperative network device independently;

[0173] Each cooperative network device receives the downlink CSI reported by the terminal;

[0174] The network device responsible for training the model obtains the downlink CSI between the terminal and other cooperative network devices at the same time through the backhaul;

[0175] The network device responsible for training the model stores the obtained downlink CSI from all cooperative network devices according to the acquisition time;

[0176] The collected data is the input data of the first layer encoder.

[0177] Method 2 mainly includes:

[0178] The terminal jointly reports the uncompressed downlink CSI to each cooperative network device;

[0179] Each cooperative network device (including the network device responsible for training the model) receives the downlink CSI reported by the terminal from all cooperative network devices;

[0180] The network device responsible for training the model stores the obtained downlink CSI from all cooperative network devices according to the acquisition time.

[0181] In some embodiments, the data collection process when the model is trained on the terminal side is specifically:

[0182] The terminal estimates the downlink CSI from each cooperative network device according to the downlink CSI-RS sent by each cooperative network device;

[0183] The terminal stores the downlink CSI from each cooperative network device according to time;

[0184] The terminal side can also report the stored information to the terminal manufacturer, and the terminal manufacturer performs model training, and after training, deploys the model to the terminal side.

[0185] 4. Model training

[0186] The terminal (or terminal manufacturer) or network device can use the following two model structures for model training using the collected training data and label data information:

[0187] The first one is based on the Auto-encoder model structure

[0188] The input of the nth (1≤n≤N) encoder model in the first layer is the label data information of the nth decoder model in the second layer.

[0189] The first layer encoder is used to compress the downlink CSI from each cooperative network device respectively, and the second layer encoder is used to compress the downlink CSI from all cooperative network devices jointly. The first layer encoder and the second layer encoder can use the same structure of convolutional neural network or self-attention mechanism network, but the dimensions of the input and output are different.

[0190] The first layer decoder is used to jointly decompress the downlink CSI from each cooperative network device, and the second layer decoder is used to decompress the downlink CSI from each cooperative network device respectively. The first layer decoder and the second layer decoder can be constructed using the same structure of convolutional neural network or self-attention mechanism, but the dimensions of the input and output are different.

[0191] The model training adopts a squared generalized cosine similarity (SGCS) as a loss function; based on the principle of back propagation, the gradient is calculated according to the loss between the original downlink CSI data and the reconstructed data output by the decoder, and the weights of the encoder and the decoder are updated until a preset number of iterations is reached, or the SGCS accuracy no longer improves.

[0192] In some embodiments, the encoder corresponds to the first target model described above, and the decoder corresponds to the second target model described above.

[0193] The second, based on the GAN model structure

[0194] The generative adversarial network for compression includes an encoder, a generator and a discriminator, wherein the encoder is responsible for compressing data, the generator is responsible for decompressing and reconstructing data, and the discriminator ensures the quality of the data reconstructed by the generator through adversarial training. In some embodiments, the encoder corresponds to the first target model described above, and the generator corresponds to the second target model described above.

[0195] Encoder network design: the first layer encoder is used to compress the downlink CSI from each cooperative network device respectively, and the second layer encoder is used to jointly compress the downlink CSI from all cooperative network devices. Both the first layer encoder and the second layer encoder can adopt a convolutional neural network or a self-attention network with the same structure, but the dimensions of the input and output are different.

[0196] Generator network: the first layer generator is used to jointly decompress the downlink CSI from each cooperative network device, and the second layer generator is used to decompress the downlink CSI from each cooperative network device respectively. Both the first layer generator and the second layer generator can be constructed by a convolutional neural network or a self-attention mechanism with the same structure, but the dimensions of the input and output are different.

[0197] Discriminator network: both the first layer discriminator and the second layer discriminator can be constructed based on a convolutional neural network or a self-attention mechanism, and are used to distinguish between the data generated by the generator and the real downlink CSI (i.e. the downlink CSI reported from the terminal).

[0198] The model training adopts SGCS as a loss function. During the model training process, an alternating training mode is used, i.e. first fix the discriminator, train the generator; then fix the generator, train the discriminator, until a preset number of iterations is reached, or the SGCS accuracy no longer improves.

[0199] 5. Model distribution

[0200] In one case, the model distribution based on Auto-encoder structure contains the following cases according to the difference of the training position of the model:

[0201] When the model is trained at the terminal, the terminal needs to send the decoder part to each cooperative network device;

[0202] When the model is trained at the terminal manufacturer, the terminal manufacturer needs to deploy the encoder model to the terminal and update it regularly; as for how to deploy the decoder model to the network device, the terminal manufacturer can deploy it to the network device by deploying the decoder to the terminal first, and then sending the decoder part to each cooperative network device; or by pre-configuration or factory setting;

[0203] When the model is trained at the network device side, the network device responsible for training the model needs to send the trained encoder model to the terminal and the decoder model to other cooperative network devices.

[0204] In another case, the model distribution based on GAN structure contains the following cases:

[0205] When the model is trained at the terminal, the terminal needs to send the generator part to each cooperative network device;

[0206] When the model is trained at the terminal manufacturer, the terminal manufacturer needs to deploy the encoder model to the terminal and update it regularly; as for how to deploy the generator model to the network device, the terminal manufacturer can deploy it to the network device by deploying the generator model to the terminal first, and then sending the generator part to each cooperative network device; or by pre-configuration or factory setting;

[0207] When the model is trained at the network device side, the network device responsible for training the model needs to send the trained encoder model to the terminal and the generator model to other cooperative network devices.

[0208] 6. Model inference

[0209] The terminal inputs the estimated downlink CSI from each cooperative network device to the encoder of the trained Auto-encoder model or the encoder of the GAN model, jointly compresses the downlink CSI of each cooperative network device, and feeds back the compressed downlink CSI to each cooperative network device. Each cooperative network device decompresses the received jointly compressed CSI based on the decoder of the Auto-encoder model or the generator of the GAN to obtain the downlink CSI of each cooperative network device.

[0210] II. The first target model is a model trained by a one-sided model

[0211] In some embodiments, the training process of the model for compression (which can also be referred to as a compressor model) is as follows:

[0212] 1. Input of the model for compression

[0213] The terminal estimates the downlink CSI from each cooperative network device, the number of transmission antenna ports of each cooperative network device, and the number of subbands occupied by the transmission signal.

[0214] 2. Output of the model for compression

[0215] The jointly compressed CSI from each cooperative network device.

[0216] 3. Data collection of the model for compression

[0217] Specifically, the model is trained at the terminal side or by the terminal manufacturer, and specifically includes:

[0218] The terminal estimates the downlink CSI from each cooperative network device according to the downlink CSI-RS transmitted by each cooperative network device.

[0219] The terminal stores the downlink CSI from each cooperative network device according to time.

[0220] The terminal obtains label data information, wherein the label data includes: first label data and second label data, the first label data is information obtained by compressing a third downlink CSI of each network device using the principle of compressed sensing, and the second label data is information obtained by compressing the first label data information using the principle of compressed sensing.

[0221] In some embodiments, the terminal side can also report the stored training data and label data information to the terminal manufacturer, and the terminal manufacturer performs model training, and after training, deploys the model to the terminal side.

[0222] 4. Training of the model for compression

[0223] The terminal or the terminal manufacturer uses the collected training data and label data information, and adopts a double-layer compression network structure as shown in FIG. 4, wherein the first layer model for compression (which can also be referred to as a first layer compressor) and the second layer model for compression (which can also be referred to as a second layer compressor) can both adopt a convolutional neural network or a self-attention mechanism network with the same structure, but the dimensions of the input and output are different. Based on the back propagation principle, the gradient is calculated according to the output of the second layer compressor and the loss of the label data, and the weights of the first layer compressor and the second layer compressor are updated until a preset number of iterations is reached, or the SGCS accuracy no longer improves.

[0224] 5、Model inference

[0225] The terminal inputs the estimated downlink CSI from each cooperative network device into the trained compressor model, jointly compresses the downlink CSI of each cooperative network device, and obtains the jointly compressed downlink CSI.

[0226] The training process of the decompressor model corresponding to the compressor model specifically includes:

[0227] 1、Input of decompressor model

[0228] The input data is the data after jointly compressing the downlink CSI of each cooperative network device based on the compression sensing principle, and the compression ratio of compression sensing.

[0229] 2、Output of decompressor model

[0230] The downlink CSI from each cooperative network device;

[0231] 3、Data collection of decompressor model

[0232] In some embodiments, the model is trained at the network device side;

[0233] In some embodiments, the training data acquisition method includes:

[0234] The terminal reports the data after jointly compressing the downlink CSI of each cooperative network device by compression sensing;

[0235] Each cooperative network device obtains the downlink CSI of other cooperative network devices through Backhaul, and obtains the data after jointly compressing the downlink CSI of each cooperative network device based on compression sensing and using the same compression matrix as the terminal side.

[0236] In some embodiments, the label data acquisition method includes:

[0237] The base station obtains the downlink CSI of other cooperative network devices through Backhaul;

[0238] The terminal reports the downlink CSI of each cooperative network device.

[0239] 4、Decompressor model training

[0240] The network device utilizes the collected training data and label data information, wherein the first layer decompressor and the second layer decompressor can both adopt a convolutional neural network or a self-attention mechanism network with the same structure, but the dimensions of the input and output are different, based on the back propagation principle, the gradient is calculated according to the loss of the output of the second layer decompressor and the label data, and the weights of the first layer decompressor and the second layer decompressor are updated until a preset number of iterations is reached, or the SGCS accuracy no longer improves.

[0241] 5、Model inference

[0242] The network device inputs the jointly compressed downlink CSI from each cooperative network device to the trained compressor model, decompresses, and obtains the downlink CSI of each cooperative network device.

[0243] It should be noted that after obtaining the jointly compressed target information based on the first target model, in some embodiments, in one implementation, the method further comprises the following one:

[0244] A11、sending the target information to at least two network devices through directional beams;

[0245] In some embodiments, when the terminal works in a frequency range 2 (Frequency Range 2, FR2) frequency band, due to the increase of path loss, the terminal needs to use a narrower transmission beam. The terminal uses the same time-frequency resource to send the target information to each cooperative network device through space division, as shown in FIG. 5, three base stations (BS) serve the terminal, and the base stations respectively issue downlink CSI-RS. The terminal measures the downlink CSI corresponding to the three base stations, jointly compresses the downlink CSI corresponding to the three base stations to obtain the target information, and sends the target information to the three base stations through directional beams.

[0246] A12、sending the target information to at least two network devices through an omnidirectional antenna;

[0247] In some embodiments, when the terminal works in a frequency range 1 (Frequency Range 1, FR1) frequency band, the terminal does not need to use a narrower transmission beam. The terminal sends the target information to each cooperative network device on a time-frequency resource negotiated by each cooperative network device, as shown in FIG. 6, three base stations (BS) serve the terminal, and the base stations respectively issue downlink CSI-RS. The terminal measures the downlink CSI corresponding to the three base stations, jointly compresses the downlink CSI corresponding to the three base stations to obtain the target information, and sends the target information to the three base stations through an omnidirectional antenna.

[0248] In some embodiments, in one implementation, the method further comprises:

[0249] receive indication information sent by the network device, the indication information being used to indicate whether the terminal jointly compresses and jointly feeds back the first downlink CSI from the multiple network devices in the cooperative communication mode;

[0250] The method further includes compressing the multiple first downlink CSI to obtain target information.

[0251] In the case where the indication information indicates that the terminal jointly compresses and jointly feeds back the first downlink CSI from the multiple network devices in the cooperative communication mode, the method further includes compressing the multiple first downlink CSI to obtain target information.

[0252] In some embodiments, the network device can send the indication information to the terminal through a physical downlink control channel (PDCCH); for example, the network device transmits a control bit, such as a JointCSIFeedback indication bit, on the PDCCH, and the value of the control bit is 0 or 1, to inform the terminal whether to jointly compress and jointly feed back the first downlink CSI from the multiple network devices in the cooperative communication mode. In some embodiments, an alternative expression is that the network device informs the terminal whether the terminal feeds back the downlink CSI from each cooperative network device to the terminal or the terminal feeds back the downlink CSI after joint compression of the cooperative network devices in the cooperative communication mode.

[0253] It should be noted that, in order to facilitate the network device to distinguish the downlink CSI, the network device needs to know which downlink CSI of the cooperative network devices is contained in the jointly compressed downlink CSI received by the network device. In some embodiments, in one implementation, the method further includes:

[0254] The terminal sends the identification information of the network device associated with the target information to at least two network devices.

[0255] That is, in this implementation, the terminal informs the network device of the identification information of the network device.

[0256] In some embodiments, in another implementation, a control center connected to each cooperative network device informs each cooperative network device of the identification information of the cooperative network device of the terminal, that is, the identification information of the cooperative network device is informed to each network device by the control center.

[0257] It should be noted that at least one embodiment of the present disclosure mines the common sparsity characteristics of CSI between different cooperative network devices under distributed MIMO, wherein the AI model used for compression contains a two-layer structure, the first layer is used for respectively compressing the downlink CSI from the cooperative network devices, and the second layer is used for jointly compressing the compressed downlink CSI from each cooperative network device to remove the correlation between the downlink CSI of each cooperative network device, then the terminal feeds back the compressed CSI to each cooperative network device, and each cooperative network device decompresses the received compressed downlink CSI, wherein the model used for decompression also contains a two-layer structure, the first layer is used for decompressing the jointly compressed data, and the second layer is used for respectively decompressing the downlink CSI of each cooperative base station, so that each network device can simultaneously obtain the CSI between the network device itself and the terminal, and also obtain the CSI between other cooperative network devices and the terminal. Embodiments of the present disclosure can realize joint CSI feedback and compression under cooperative network devices with lower feedback overhead, and without increasing the time delay of CSI feedback.

[0258] The technical solutions provided by the embodiments of the present disclosure can be applied to various systems, especially 5G systems. For example, the applicable systems can be global system of mobile communication (GSM) systems, code division multiple access (CDMA) systems, wideband code division multiple access (WCDMA) general packet radio service (GPRS) systems, long term evolution (LTE) systems, LTE frequency division duplex (FDD) systems, LTE time division duplex (TDD) systems, long term evolution advanced (LTE-A) systems, universal mobile systems (UMTS), worldwide interoperability for microwave access (WiMAX) systems, 5th generation mobile communication technology (5G) new radio (NR) systems, and the like. Among these various systems, there are terminals (which can also be referred to as terminal devices) and network devices. The system can also include a core network part, such as an evolved packet system (EPS), a 5G system (5GS), and the like.

[0259] The terminal device to which the embodiments of the present disclosure relate can also be referred to as a device providing voice and / or data connectivity to a user, a handheld device having wireless connection function, or other processing devices connected to a wireless modem, etc. In different systems, the name of the terminal device can also be different, for example, in the 5G system, the terminal device can be referred to as a user equipment (UE). The wireless terminal device can communicate with one or more core networks (CN) via a radio access network (RAN), and the wireless terminal device can be a mobile terminal device, such as a mobile phone (or called "cellular" phone) and a computer with a mobile terminal device, for example, it can be a portable, pocket, handheld, computer built-in or vehicle-mounted mobile device, which exchanges language and / or data with the radio access network. For example, personal communication service (PCS) phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), etc. The wireless terminal device can also be referred to as a system, a subscriber unit, a subscriber station, a mobile station, a mobile, a remote station, an access point, a remote terminal, an access terminal, a user terminal, a user agent, a user device, which is not limited in the embodiments of the present disclosure.

[0260] The network device related to the embodiments of the present disclosure can be a base station, which can include a plurality of cells serving terminals. According to different application scenarios, the base station can also be referred to as an access point, or can be a device in an access network that communicates with wireless terminal devices through one or more sectors over an air interface, or other names. The network device can be used to exchange received air frames and Internet Protocol (IP) packets as a router between the wireless terminal device and the rest of the access network, which can include an Internet Protocol (IP) communication network. The network device can also coordinate the management of the properties of the air interface. For example, the network device related to the embodiments of the present disclosure can be a network device (Base Transceiver Station, BTS) in the Global System for Mobile Communications (GSM) or Code Division Multiple Access (CDMA), and can also be a network device (NodeB) in Wide-band Code Division Multiple Access (WCDMA), and can also be an evolved network device (evolutional Node B, eNB or e-NodeB) in a long term evolution (LTE) system, a 5G base station (gNB) in a next generation system (5G network architecture), and can also be a Home evolved Node B (HeNB), a relay terminal node, a femto, a pico, etc., which are not limited in the embodiments of the present disclosure. In some network structures, the network device can include a centralized unit (CU) node and a distributed unit (DU) node, and the centralized unit and the distributed unit can also be arranged geographically apart.

[0261] The network device and the terminal device can each use one or more antennas for multi-input multi-output (MIMO) transmission, which can be single-user MIMO (SU-MIMO) or multi-user MIMO (MU-MIMO). According to the shape and number of root antenna combinations, MIMO transmission can be two-dimensional MIMO (2D-MIMO), three-dimensional MIMO (3D-MIMO), full-dimensional MIMO (FD-MIMO), or massive-MIMO, and can also be diversity transmission or precoding transmission or beamforming transmission, etc.

[0262] As shown in FIG. 7, the embodiment of the present disclosure provides a channel state information acquisition method, executed by a network device, comprising:

[0263] Step S701, receiving target information sent by a terminal, the target information being compressed by a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to one network device;

[0264] Step S702, decompressing the target information to obtain a plurality of decompressed first downlink CSIs.

[0265] In some embodiments, the decompressing the target information to obtain a plurality of decompressed first downlink CSIs comprises:

[0266] inputting the target information into a second target model to obtain a plurality of decompressed first downlink CSIs output by the second target model;

[0267] The second target model is a double-layer decompression model, a first layer decompression model decompresses the target information to obtain fourth downlink CSIs, and a second layer of a plurality of decompression models respectively decompresses the fourth downlink CSIs.

[0268] The second target model is trained by a bilateral model and first training data; or the second target model is trained by a unilateral model and second training data, the second training data comprising: training data of a first layer decompression model, label data of the first layer decompression model, training data of a second layer decompression model, and label data of the second layer decompression model, the training data of the first layer decompression model being second label data information, the label data of the first layer decompression model being first label data information, the training data of the second layer decompression model being the first label data information, and the label data of the second layer decompression model being third downlink CSI from multiple network devices.

[0269] The first training data comprises at least one of: third downlink CSI corresponding to multiple network devices respectively, the number of sending antenna ports of multiple network devices, the number of subbands occupied by signals sent by multiple network devices, and label data information.

[0270] The first label data is information obtained by compressing third downlink CSI of multiple network devices respectively by using a compression sensing principle, and the second label information is information obtained by compressing the first label data information by using the compression sensing principle.

[0271] In some embodiments, the bilateral model comprises at least one of:

[0272] A self-encoder model, a convolutional neural network model, a generative adversarial network model, and a Transformer structure model.

[0273] In some embodiments, in the case where the second target model is trained by a bilateral model and first training data, the method further comprises at least one of:

[0274] The first target model is sent to a terminal, the first target model being a model for compressing multiple first downlink CSI, and the first target model being obtained by training a network device by using first training data;

[0275] The second target model sent by the terminal is received.

[0276] In some embodiments, the method further comprises:

[0277] Indication information is sent to the terminal, the indication information being used to indicate whether the terminal jointly compresses and jointly feeds back first downlink CSI from multiple network devices in a cooperative communication mode.

[0278] In some embodiments, the method further comprises one of:

[0279] receive the identification information of the network device corresponding to the target information sent by the terminal;

[0280] receive the identification information of the network device providing cooperative communication for the terminal sent by the control center.

[0281] It should be noted that all the implementation manners in the above embodiments are applicable to the embodiments of the channel state information acquisition method applied to the network device side, and can achieve the same technical effects, which will not be described here.

[0282] As shown in FIG. 8, the disclosure embodiment provides a channel state information transmission device 800 applied to a terminal, comprising:

[0283] A first acquisition unit 801 is configured to acquire a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to one network device;

[0284] A second acquisition unit 802 is configured to compress the plurality of first downlink CSIs to acquire target information.

[0285] A first sending unit 803 is configured to send the target information to at least two network devices, the at least two network devices providing coherent joint transmission for the terminal.

[0286] In some embodiments, the second acquisition unit 802 is configured to:

[0287] input the plurality of first downlink CSIs into a first target model to acquire target information output by the first target model;

[0288] The first target model is a double-layer compression model, a first layer compression model compresses a plurality of first downlink CSIs respectively, and a second layer compression model compresses a plurality of second downlink CSIs output by the first layer compression model; the first target model is acquired by training first training data, the first training data including at least one of the following: third downlink CSIs respectively corresponding to a plurality of network devices, a number of sending antenna ports of the plurality of network devices, a number of subbands occupied by signals sent by the plurality of network devices, and label data information.

[0289] The third downlink CSI is a downlink CSI used for model training.

[0290] In some embodiments, the first target model is a model obtained by training a bilateral model, or the first target model is a model obtained by training a unilateral model.

[0291] In some embodiments, the bilateral model includes at least one of the following:

[0292] The self-encoder model, the convolutional neural network model, the generative adversarial network model, and the Transformer structure model.

[0293] In some embodiments, in the case that the first target model is a model trained by a bilateral model, the apparatus further comprises at least one of the following:

[0294] a second sending unit configured to send a second target model to the plurality of network devices, the second target model being a model for decompressing the target information, the second target model being obtained by training the first training data by the terminal;

[0295] a second receiving unit configured to receive a first target model sent by the network device;

[0296] a fourth obtaining unit configured to obtain a preconfigured first target model

[0297] In some embodiments, in the case that the first target model is a model trained by a bilateral model, the label data information included in the first training data is third downlink CSI corresponding to the plurality of network devices respectively.

[0298] In some embodiments, in the case that the first target model is a model trained by a unilateral model, the label data information included in the first training data comprises first label data and second label data, the first label data being information obtained by compressing third downlink CSI of the plurality of network devices respectively by using the compression sensing principle, and the second label information being information obtained by compressing the first label data by using the compression sensing principle.

[0299] In some embodiments, the first sending unit 803 is configured to implement one of the following:

[0300] sending the target information to the at least two network devices by directional beams;

[0301] sending the target information to the at least two network devices by an omnidirectional antenna.

[0302] In some embodiments, the apparatus further comprises:

[0303] a third receiving unit configured to receive indication information sent by the network device, the indication information being used to indicate whether the terminal jointly compresses and jointly feeds back the first downlink CSI from the plurality of network devices in the cooperative communication mode;

[0304] The second obtaining unit 802 is configured to:

[0305] In a case where the indication information indicates that the terminal jointly compresses and jointly feeds back the first downlink CSI from the multiple network devices in the cooperative communication mode, the multiple first downlink CSIs are compressed to obtain target information.

[0306] In some embodiments, the apparatus further includes:

[0307] A third sending unit is configured to send, to the at least two network devices, identification information of the network device associated with the target information.

[0308] It should be noted that the apparatus embodiment is one-to-one corresponding to the above-mentioned method embodiment, and all implementation manners in the above-mentioned method embodiment are applicable to the apparatus embodiment, and the same technical effects can be achieved.

[0309] It should be noted that the division of units in the embodiments of the present disclosure is illustrative, and is merely a logical function division. In actual implementation, another division manner can be used. In addition, each functional unit in each embodiment of the present disclosure can be integrated in one processing unit, or each unit can exist physically independently, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0310] When the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, the integrated unit can be stored in a processor-readable storage medium. Based on this understanding, the technical solutions of the present disclosure, essentially or the part that contributes to the related art, or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform all or part of the steps of the methods described in the various embodiments of the present disclosure. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various other media that can store program codes.

[0311] As shown in FIG. 9, the embodiments of the present disclosure further provide a terminal, including a processor 900, a transceiver 910, a memory 920, and a program stored in the memory 920 and executable on the processor 900; wherein the transceiver 910 is connected with the processor 900 and the memory 920 through a bus interface, wherein the processor 900 is configured to read the program in the memory and perform the following processes:

[0312] Obtain a plurality of first downlink channel state information (CSI), each of the plurality of first downlink CSI corresponding to one network device respectively;

[0313] Compress the plurality of first downlink CSI to obtain target information;

[0314] Send the target information to at least two network devices through the transceiver, the at least two network devices providing coherent joint transmission for the terminal.

[0315] The transceiver 910 is configured to receive and send data under the control of the processor 900.

[0316] In FIG. 9, the bus architecture can include any number of interconnected buses and bridges, which link various circuits, including the processor(s) 900 and the memory 920, which is represented by one or more processors. The bus architecture can also link various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and thus will not be described further herein. The bus interface provides an interface. The transceiver 910 can be a plurality of elements, including a transmitter and a receiver, which provide a means for communicating with various other apparatuses over a transmission medium, including wireless channels, wired channels, optical cables, and the like. The user interface 930 can also be an interface that can be externally or internally connected to the required device, including but not limited to a keypad, a display, a speaker, a microphone, a joystick, and the like, for different user devices.

[0317] The processor 900 is responsible for managing the bus architecture and general processing, and the memory 920 can store data used by the processor 900 when performing operations.

[0318] Optionally, the processor 900 can be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a complex programmable logic device (CPLD), and the processor can also adopt a multi-core architecture.

[0319] The processor is configured to execute any of the methods provided by the embodiments of the present disclosure according to the executable instructions obtained from the memory. The processor and the memory can also be physically arranged separately.

[0320] In some embodiments, the processor, for reading the computer program in the memory and performing the following operations:

[0321] inputting the plurality of first downlink CSI into a first target model to obtain target information output by the first target model;

[0322] The first target model is a double-layer compression model, the first layer compression model compresses a plurality of first downlink CSI respectively, and the second layer compression model compresses a plurality of second downlink CSI output by the first layer compression model; the first target model is obtained by training the first training data, and the first training data includes at least one of the following: third downlink CSI corresponding to a plurality of network devices respectively, the number of sending antenna ports of a plurality of network devices, the number of subbands occupied by a plurality of network devices to send signals, and label data information;

[0323] The third downlink CSI is a downlink CSI used for model training.

[0324] In some embodiments, the first target model is a model obtained by training a double-sided model, or the first target model is a model obtained by training a single-sided model.

[0325] In some embodiments, the double-sided model includes at least one of the following:

[0326] autoencoder model, convolutional neural network model, generative adversarial network model, and Transformer structure model.

[0327] In some embodiments, in the case where the first target model is a model obtained by training a double-sided model, the processor, for reading the computer program in the memory, further performs at least one of the following operations:

[0328] sending a second target model to a plurality of network devices, the second target model being a model for decompressing the target information, the second target model being obtained by training the first training data by a terminal;

[0329] receiving a first target model sent by a network device;

[0330] obtaining a preconfigured first target model.

[0331] In some embodiments, in the case where the first target model is a model obtained by training a double-sided model, the label data information included in the first training data is third downlink CSI corresponding to a plurality of network devices respectively.

[0332] In some embodiments, in a case where the first target model is a model trained by a one-sided model, the label data information included in the first training data comprises first label data and second label data, the first label data is information obtained by compressing third downlink CSI of a plurality of network devices respectively by using a compression sensing principle, and the second label information is information obtained by compressing the first label data information by using a compression sensing principle.

[0333] In some embodiments, the processor is configured to read the computer program in the memory and perform one of the following operations:

[0334] sending the target information to the at least two network devices through a directional beam;

[0335] sending the target information to the at least two network devices through an omnidirectional antenna.

[0336] In some embodiments, the processor is configured to read the computer program in the memory and perform the following operation:

[0337] receiving, by the transceiver, indication information sent by a network device, the indication information being used to indicate whether the terminal jointly compresses and jointly feeds back first downlink CSI from a plurality of network devices in a cooperative communication mode;

[0338] In a case where the indication information indicates that the terminal jointly compresses and jointly feeds back the first downlink CSI from the plurality of network devices in the cooperative communication mode, compressing the plurality of first downlink CSI to obtain target information.

[0339] In some embodiments, the processor is configured to read the computer program in the memory and perform the following operation:

[0340] sending, by the transceiver, identification information of a network device associated with the target information to the at least two network devices.

[0341] It should be noted that the above terminal provided by the embodiments of the present disclosure can realize all the method steps realized by the above method embodiments and achieve the same technical effects. Therefore, the same parts and beneficial effects of the method embodiments will not be described in detail.

[0342] The embodiments of the present disclosure further provide a computer readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the channel state information transmission method applied to a terminal when executed by a processor. The processor readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to a magnetic storage (such as a floppy disk, a hard disk, a magnetic tape, a magneto optical disk (MO), etc.), an optical storage (such as a compact disc (CD), a digital video disc (DVD), a Blu-ray disc (BD), a high-definition versatile disc (HVD), etc.), and a semiconductor memory (such as a ROM, an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a non-volatile memory (NAND FLASH), a solid state disk (SSD), etc.), etc.

[0343] As shown in FIG. 10, the embodiments of the present disclosure provide a channel state information acquisition apparatus 1000 applied to a network device, comprising:

[0344] A first receiving unit 1001 is configured to receive target information sent by a terminal, wherein the target information is compressed by a plurality of first downlink channel state information (CSI), and each first downlink CSI corresponds to one network device;

[0345] A third obtaining unit 1002 is configured to decompress the target information to obtain a plurality of decompressed first downlink CSIs.

[0346] In some embodiments, the third obtaining unit 1002 is configured to:

[0347] input the target information into a second target model to obtain a plurality of decompressed first downlink CSIs output by the second target model;

[0348] In some embodiments, the second target model is a double-layer decompression model, a first layer decompression model decompresses the target information to obtain fourth downlink CSIs, and a plurality of second layer decompression models decompress the fourth downlink CSIs respectively.

[0349] The second target model is trained by a bilateral model and first training data; or the second target model is trained by a unilateral model and second training data, the second training data including: first-layer decompression model training data, first-layer decompression model label data, second-layer decompression model training data, and second-layer decompression model label data, the first-layer decompression model training data being second label data information, the first-layer decompression model label data being first label data information, the second-layer decompression model training data being the first label data information, and the second-layer decompression model label data being third downlink CSI from multiple network devices.

[0350] The first training data includes at least one of: third downlink CSI corresponding to multiple network devices respectively, the number of sending antenna ports of multiple network devices, the number of subbands occupied by signals sent by multiple network devices, and label data information.

[0351] The first label data is information obtained by compressing third downlink CSI of multiple network devices respectively by using a compression sensing principle, and the second label information is information obtained by compressing the first label data information by using a compression sensing principle.

[0352] In some embodiments, the bilateral model includes at least one of:

[0353] A self-encoder model, a convolutional neural network model, a generative adversarial network model, and a Transformer structure model.

[0354] In some embodiments, in the case where the second target model is trained by a bilateral model and first training data, the apparatus further includes at least one of:

[0355] A fourth sending unit configured to send a first target model to a terminal, the first target model being a model for compressing multiple first downlink CSIs, and the first target model being obtained by training a network device by using first training data.

[0356] A fourth receiving unit configured to receive a second target model sent by the terminal.

[0357] In some embodiments, the apparatus further includes:

[0358] A fifth sending unit configured to send indication information to the terminal, the indication information being used to indicate whether the terminal jointly compresses and jointly feeds back first downlink CSIs from multiple network devices in a cooperative communication mode.

[0359] In some embodiments, the apparatus further includes one of:

[0360] a fifth receiving unit, configured to receive identification information of a network device associated with the target information sent by the terminal;

[0361] a sixth receiving unit, configured to receive identification information of a network device providing cooperative communication for the terminal sent by the control center.

[0362] It should be noted that the apparatus embodiment is one-to-one corresponding to the above-mentioned method embodiment, and all implementation manners in the above-mentioned method embodiment are applicable to the apparatus embodiment, and the same technical effects can be achieved.

[0363] It should be noted that the division of units in the embodiments of the present disclosure is illustrative, and is only a logical function division. In actual implementation, another division mode can be used. In addition, each functional unit in each embodiment of the present disclosure can be integrated in one processing unit, or each unit can exist physically independently, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0364] When the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a processor-readable storage medium. Based on this understanding, the technical solutions of the present disclosure, essentially or the part that contributes to the related art, or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in the various embodiments of the present disclosure. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0365] As shown in FIG. 11, the embodiments of the present disclosure further provide a network device, including a processor 1100, a transceiver 1110, a memory 1120, and a program stored in the memory 1120 and executable on the processor 1100; wherein the transceiver 1110 is connected with the processor 1100 and the memory 1120 through a bus interface, wherein the processor 1100 is used to read the program in the memory and execute the following processes: wherein the processor is used to read the computer program in the memory to execute the following operations:

[0366] The transceiver receives target information sent by the terminal, the target information is obtained by compressing a plurality of first downlink channel state information (CSI), and each first downlink CSI corresponds to one network device;

[0367] The target information is decompressed to obtain a plurality of decompressed first downlink CSIs.

[0368] The transceiver 1110 is configured to receive and send data under the control of the processor 1100.

[0369] In FIG. 11, the bus architecture can include any number of interconnected buses and bridges, which link together various circuits, including the processor(s) 1100 and the memory represented by the memory 1120. The bus architecture can also link various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and thus, not further described herein. The bus interface provides an interface. The transceiver 1110 can be multiple elements, i.e., including a transmitter and a receiver, which provide a means for communicating with various other apparatuses over a transmission medium, including wireless channels, wired channels, optical cables, and the like. The user interface 1130 can also be an interface capable of connecting to external and internal devices as needed, including but not limited to a keypad, a display, a speaker, a microphone, a joystick, and the like.

[0370] The processor 1100 is responsible for managing the bus architecture and general processing, and the memory 1120 can store data used by the processor 1100 when performing operations.

[0371] Optionally, the processor 1100 can be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a complex programmable logic device (CPLD), and the processor can also adopt a multi-core architecture.

[0372] The processor is configured to execute any of the methods provided by the embodiments of the disclosure according to the executable instructions obtained from the memory. The processor and the memory can also be physically arranged separately.

[0373] In some embodiments, the processor is configured to read the computer program in the memory and perform the following operations:

[0374] input the target information into a second target model, and obtain a plurality of decompressed first downlink CSIs output by the second target model;

[0375] The second target model is a double-layer decompression model, a first layer decompression model is configured to decompress the target information to obtain fourth downlink CSIs, and a second layer includes a plurality of decompression models configured to decompress the fourth downlink CSIs respectively.

[0376] The second target model is trained by using a bilateral model and first training data, or the second target model is trained by using a unilateral model and second training data. The second training data includes training data of the first layer decompression model, label data of the first layer decompression model, training data of the second layer decompression model, and label data of the second layer decompression model. The training data of the first layer decompression model is second label data information, the label data of the first layer decompression model is first label data information, the training data of the second layer decompression model is the first label data information, and the label data of the second layer decompression model is third downlink CSIs from a plurality of network devices.

[0377] The first training data includes at least one of third downlink CSIs corresponding to the plurality of network devices respectively, a number of transmitting antenna ports of the plurality of network devices, a number of subbands occupied by signals transmitted by the plurality of network devices, and label data information.

[0378] The first label data is information obtained by compressing the third downlink CSIs of the plurality of network devices respectively by using a compression sensing principle, and the second label information is information obtained by compressing the first label data information by using the compression sensing principle.

[0379] In some embodiments, the bilateral model includes at least one of:

[0380] a self-encoder model, a convolutional neural network model, a generative adversarial network model, and a Transformer structure model.

[0381] In some embodiments, when the second target model is trained by using the bilateral model and the first training data, the computer program stored in the memory and read by the processor further performs at least one of the following operations:

[0382] The first target model is a model used for compressing the plurality of first downlink CSIs, and the first target model is obtained by training using first training data.

[0383] The second target model is received from the terminal.

[0384] In some embodiments, the processor, configured to read the computer program in the memory, is further configured to perform one of the following operations:

[0385] transmit, by the transceiver, indication information to the terminal, the indication information being used to indicate whether the terminal jointly compresses and jointly feeds back the first downlink CSI from the plurality of network devices in the cooperative communication mode.

[0386] In some embodiments, the processor, configured to read the computer program in the memory, is further configured to perform one of the following operations:

[0387] receive, by the transceiver, the identification information of the network device associated with the target information sent by the terminal;

[0388] receive, by the transceiver, the identification information of the network device providing cooperative communication for the terminal sent by the control center.

[0389] It should be noted that the above network device provided by the embodiments of the present disclosure can realize all the method steps realized by the method embodiments and achieve the same technical effects. Therefore, the same parts and beneficial effects of the method embodiments will not be described in detail.

[0390] The embodiments of the present disclosure further provide a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of the channel state information acquisition method applied to a network device. The processor readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to a magnetic storage (such as a floppy disk, a hard disk, a magnetic tape, a magneto-optical disk (MO), etc.), an optical storage (such as a CD, a DVD, a BD, a HVD, etc.), and a semiconductor memory (such as a ROM, an EPROM, an EEPROM, a non-volatile memory (NAND FLASH), a solid state disk (SSD), etc.).

[0391] The embodiments of the present disclosure further provide a computer program product including computer instructions, which, when executed by a processor, implement each process in the above method embodiments and achieve the same technical effects. To avoid repetition, details will not be described here.

[0392] Those skilled in the art should understand that the embodiments of the present disclosure can be provided as a method, a system, or a computer program product. Therefore, the present disclosure can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present disclosure can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage and optical storage, etc.) containing computer-usable program code.

[0393] The disclosed embodiments are described with reference to the drawings being flowchart illustrations and / or block diagrams of methods, apparatus (systems) and computer program products according to the disclosed embodiments. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer executable instructions. The computer executable instructions can be provided to a processor of a general purpose computer, special purpose computer, an embedded processor or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0394] These computer executable instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart illustrations and / or block diagrams block or blocks.

[0395] These computer executable instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0396] Further, it is noted that in the disclosed embodiments, it is apparent that the components or steps can be decomposed and / or recombined. These decompositions and / or recombination should be considered as equivalent solutions of the disclosed embodiments. Also, the steps of performing the above series of processes can naturally be executed in time series according to the described order, but do not necessarily have to be executed in time series, and some steps can be executed in parallel or independently of each other. It is understood by those of ordinary skill in the art that all or any steps or components of the disclosed embodiments can be implemented in hardware, firmware, software, or a combination thereof, in any computing device (including processors, storage media, etc.) or network of computing devices, using their basic programming skills upon reading the description of the disclosed embodiments.

[0397] It should be noted that the division of the above modules is only a logical functional division, and all or part of them can be integrated into a physical entity or physically separated in actual implementation. These modules can all be implemented in the form of software called by a processing element; all can be implemented in the form of hardware; or some modules can be implemented in the form of software called by a processing element, and some modules can be implemented in the form of hardware. For example, a certain module can be a separately established processing element, or can be integrated into a certain chip of the above device, in addition, it can also be stored in the form of program code in the memory of the above device, and the function of the above determination module is called and executed by a certain processing element of the above device. The implementation of other modules is similar. In addition, all or part of these modules can be integrated together or independently implemented. The processing element described herein can be an integrated circuit with signal processing capability. In the implementation process, each step of the above method or each module can be completed by the integrated logic circuit of the hardware in the processor element or the instruction in the form of software.

[0398] For example, each module, unit, sub-unit or sub-module can be one or more integrated circuits configured to implement the above method, such as one or more application specific integrated circuits (ASICs), or one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs), etc. For another example, when a certain module above is implemented in the form of program code called by a processing element, the processing element can be a general purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules can be integrated together to implement in the form of system on a chip (SOC).

[0399] The terminology used in the description and the claims of the present disclosure is intended to be interpreted in only its broadest reasonable manner, even though it is used in conjunction with a general symbolic representation of the concepts. The terms "comprises," "comprising," "includes," "including" and "contains," "containing," are intended to be open-ended, meaning that they include at least the elements specified after such terms, but do not exclude other elements. The terms "first," "second," and the like, do not denote any ordinal, sequential, or high-low relationship, but are used to distinguish one from another. The terms "and / or," and / or "and / or" are intended to cover all possible combinations of the elements, including individual elements, combinations, and / or sub-combinations of the elements, and can be used interchangeably with the term "and / or". The term "or" is intended to mean an inclusive "or" rather than an exclusive "or". That is, unless specified otherwise, or clear from the context, the phrase "X employs A or B" is intended to mean that the alternative of "A or B" is the

[0400] It will be apparent to those skilled in the art that various modifications and variations can be made to the present disclosure without departing from the spirit or scope of the disclosure. Thus, it is intended that the present disclosure cover the modifications and variations of this disclosure provided such modifications and variations come within the scope of the application and the equivalents thereof.

Claims

1. A channel state information transmission method applied to a terminal, comprising: obtaining a plurality of first downlink channel state information (CSI), each of the plurality of first downlink CSI corresponding to a network device; compressing the plurality of first downlink CSI to obtain target information; sending the target information to at least two network devices, the at least two network devices providing coherent joint transmission for the terminal.

2. The method of claim 1, wherein, The step of compressing the plurality of first downlink CSI to obtain target information comprises: inputting the plurality of first downlink CSI into a first target model to obtain target information output by the first target model; wherein the first target model is a double-layer compression model, a first layer compression model compresses the plurality of first downlink CSI respectively, and a second layer compression model compresses a plurality of second downlink CSI output by the first layer compression model; the first target model is obtained by training first training data, the first training data comprising at least one of the following: third downlink CSI corresponding to the plurality of network devices respectively, the number of transmit antenna ports of the plurality of network devices, the number of subbands occupied by the plurality of network devices for signal transmission, and label data information; The third downlink CSI is a downlink CSI used for model training.

3. The method of claim 2, wherein, The first target model is a model obtained by training a double-sided model, or the first target model is a model obtained by training a single-sided model.

4. The method of claim 3, wherein, The double-sided model comprises at least one of the following: autoencoder model, convolutional neural network model, generative adversarial network model, and Transformer structure model.

5. The method of claim 3 or 4, wherein, In the case where the first target model is a model obtained by training a double-sided model, the method further comprises at least one of the following: sending a second target model to the plurality of network devices, the second target model being a model for decompressing the target information, the second target model being obtained by training the first training data by the terminal; receiving a first target model sent by a network device; obtaining a preconfigured first target model.

6. The method of claim 3, wherein, In the case where the first target model is a model obtained by training a double-sided model, the label data information included in the first training data is third downlink CSI corresponding to the plurality of network devices respectively.

7. The method of claim 3, wherein, In the case where the first target model is a model obtained by training a single-sided model, the label data information included in the first training data comprises first label data and second label data, the first label data being information obtained by compressing the third downlink CSI of the plurality of network devices respectively using the compression sensing principle, and the second label information being information obtained by compressing the first label data information using the compression sensing principle.

8. The method of claim 1, wherein, The step of sending the target information to at least two network devices comprises one of the following: sending the target information to the at least two network devices through directional beams; sending the target information to the at least two network devices through an omnidirectional antenna.

9. The method of claim 1, wherein, Further comprising: receive indication information sent by a network device, the indication information being used to indicate whether the terminal jointly compresses and feeds back first downlink CSI from multiple network devices in a cooperative communication mode; the target information is obtained by compressing the multiple first downlink CSI. In a case where the indication information indicates that the terminal jointly compresses and feeds back the first downlink CSI from the multiple network devices in the cooperative communication mode, the target information is obtained by compressing the multiple first downlink CSI.

10. The method of claim 1, wherein, Further comprising: sending, to at least two network devices, identification information of a network device associated with the target information.

11. A channel state information acquisition method applied to a network device, comprising: receiving target information sent by a terminal, the target information being obtained by compressing multiple first downlink channel state information (CSI), each first downlink CSI corresponding to one network device; decompressing the target information to obtain multiple decompressed first downlink CSI.

12. The method of claim 11, wherein, The decompression of the target information to obtain multiple decompressed first downlink CSI comprises: inputting the target information into a second target model to obtain multiple decompressed first downlink CSI output by the second target model; wherein the second target model is a double-layer decompression model, a first-layer decompression model decompresses the target information to obtain fourth downlink CSI, and multiple second-layer decompression models decompress the fourth downlink CSI respectively; The second target model is trained by a bilateral model and first training data, or the second target model is trained by a unilateral model and second training data, the second training data comprising: training data of the first-layer decompression model, label data of the first-layer decompression model, training data of the second-layer decompression model, and label data of the second-layer decompression model, the training data of the first-layer decompression model being second label data information, the label data of the first-layer decompression model being first label data information, the training data of the second-layer decompression model being the first label data information, and the label data of the second-layer decompression model being third downlink CSI from multiple network devices; The first training data comprises at least one of: third downlink CSI corresponding to multiple network devices respectively, the number of transmission antenna ports of multiple network devices, the number of subbands occupied by signals transmitted by multiple network devices, and label data information; The first label data is information obtained by compressing the third downlink CSI of multiple network devices respectively using the principle of compressed sensing, and the second label information is information obtained by compressing the first label data information using the principle of compressed sensing.

13. The method of claim 12, wherein, The bilateral model comprises at least one of: autoencoder model, convolutional neural network model, generative adversarial network model, and Transformer structure model.

14. The method of claim 12 or 13, wherein, In a case where the second target model is trained by a bilateral model and first training data, the method further comprises at least one of: The first target model is sent to the terminal, and the first target model is a model used for performing a plurality of first downlink CSI compression, and the first target model is obtained by training the network device using first training data. The second target model sent by the terminal is received.

15. The method of claim 11, wherein, Further comprising: The indication information is sent to the terminal, and the indication information is used to indicate whether the terminal jointly compresses and jointly feeds back the first downlink CSI from a plurality of network devices in a cooperative communication mode.

16. The method of claim 11, wherein, Further comprising one of the following: The identification information of the network device associated with the target information sent by the terminal is received; The identification information of the network device providing cooperative communication for the terminal sent by the control center is received.

17. A terminal comprising a memory, a transceiver, and a processor: The memory is used to store a computer program; the transceiver is used to transceive data under the control of the processor; the processor is used to read the computer program in the memory and perform the following operations: Obtain a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to a network device; Compress the plurality of first downlink CSI to obtain target information; The target information is sent to at least two network devices through the transceiver, and the at least two network devices provide coherent joint transmission for the terminal.

18. The terminal of claim 17, wherein, The processor is used to read the computer program in the memory and perform the following operations: The plurality of first downlink CSI is input into a first target model to obtain target information output by the first target model; The first target model is a double-layer compression model, a first layer compression model compresses a plurality of first downlink CSI respectively, and a second layer compression model compresses a plurality of second downlink CSI output by the first layer compression model; the first target model is obtained by training the first training data, and the first training data includes at least one of the following: a third downlink CSI corresponding to a plurality of network devices respectively, a number of transmission antenna ports of a plurality of network devices, a number of subbands occupied by a plurality of network devices to transmit signals, and label data information; The third downlink CSI is a downlink CSI used for model training.

19. The terminal of claim 18, wherein, The first target model is a model obtained by training a double-sided model, or the first target model is a model obtained by training a single-sided model.

20. The terminal of claim 19, wherein, The double-sided model includes at least one of the following: An autoencoder model, a convolutional neural network model, a generative adversarial network model, and a Transformer structure model.

21. The terminal according to claim 19 or 20, wherein In the case where the first target model is a model obtained by training a double-sided model, the processor is used to read the computer program in the memory and perform at least one of the following operations: The second target model is sent to a plurality of network devices, and the second target model is a model used for decompressing the target information, and the second target model is obtained by training the terminal using first training data; The first target model sent by the network device is received; A preconfigured first target model is obtained.

22. The terminal of claim 19, wherein, In a case where the first target model is a model trained by a bilateral model, the label data information included in the first training data is third downlink CSI corresponding to the plurality of network devices respectively.

23. The terminal of claim 19, wherein, In a case where the first target model is a model trained by a unilateral model, the label data information included in the first training data comprises first label data and second label data, the first label data is information obtained by compressing the third downlink CSI of the plurality of network devices respectively by using a compression sensing principle, and the second label information is information obtained by compressing the first label data information by using the compression sensing principle.

24. The terminal of claim 17, wherein, The processor is configured to read the computer program in the memory and perform one of the following operations: sending target information to at least two network devices through a directional beam; sending target information to at least two network devices through an omnidirectional antenna.

25. The terminal of claim 17, wherein, The processor is configured to read the computer program in the memory and perform the following operation: receiving, by the transceiver, indication information sent by a network device, the indication information being used to indicate whether the terminal jointly compresses and feeds back first downlink CSI from a plurality of network devices in a cooperative communication mode; in a case where the indication information indicates that the terminal jointly compresses and feeds back the first downlink CSI from the plurality of network devices in the cooperative communication mode, compressing the plurality of first downlink CSI to obtain target information.

26. The terminal of claim 17, wherein, The processor is configured to read the computer program in the memory and perform the following operation: sending, by the transceiver, identification information of a network device associated with the target information to at least two network devices.

27. A network device comprising a memory, a transceiver, and a processor: the memory is configured to store a computer program; the transceiver is configured to transceive data under control of the processor; and the processor is configured to read the computer program in the memory and perform the following operations: receiving, by the transceiver, target information sent by a terminal, the target information being obtained by compressing a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to one network device; decompressing the target information to obtain a plurality of decompressed first downlink CSI.

28. The network device of claim 27, wherein, The processor is configured to read the computer program in the memory and perform the following operation: inputting the target information into a second target model to obtain a plurality of decompressed first downlink CSI output by the second target model; wherein the second target model is a double-layer decompression model, a first layer decompression model decompresses the target information to obtain fourth downlink CSI, and a plurality of second layer decompression models decompress the fourth downlink CSI respectively. The second target model is trained by a bilateral model and first training data; or the second target model is trained by a unilateral model and second training data, the second training data comprising: first layer decompression model training data, first layer decompression model label data, second layer decompression model training data, and second layer decompression model label data, the first layer decompression model training data being second label data information, the first layer decompression model label data being first label data information, the second layer decompression model training data being the first label data information, and the second layer decompression model label data being third downlink CSI from multiple network devices; The first training data comprises at least one of: third downlink CSI corresponding to multiple network devices respectively, the number of sending antenna ports of multiple network devices, the number of subbands occupied by signals sent by multiple network devices, and label data information; The first label data is information obtained by compressing third downlink CSI of multiple network devices respectively by using a compression sensing principle, and the second label information is information obtained by compressing the first label data information by using a compression sensing principle.

29. The network device of claim 28, wherein, The bilateral model comprises at least one of: a self-encoder model, a convolutional neural network model, a generative adversarial network model, and a Transformer structure model.

30. The network device of claim 28 or 29, wherein, In the case where the second target model is trained by a bilateral model and first training data, the processor, for reading the computer program in the memory, further performs at least one of the following operations: sending a first target model to a terminal, the first target model being a model for compressing multiple first downlink CSIs, and the first target model being obtained by training a network device by using first training data; receiving a second target model sent by a terminal.

31. The network device of claim 27, wherein, The processor, for reading the computer program in the memory, further performs the following operation: sending indication information to the terminal through a transceiver, the indication information being used to indicate whether the terminal jointly compresses and feeds back first downlink CSIs from multiple network devices in a cooperative communication mode.

32. The network device of claim 27, wherein, The processor, for reading the computer program in the memory, further performs one of the following operations: receiving, through a transceiver, identification information of a network device associated with the target information sent by a terminal; receiving, through a transceiver, identification information of a network device providing cooperative communication for the terminal sent by a control center.

33. A channel state information transmission apparatus applied to a terminal, comprising: a first obtaining unit configured to obtain multiple first downlink channel state information (CSI), each first downlink CSI corresponding to one network device; a second obtaining unit configured to compress the multiple first downlink CSIs to obtain target information; a first sending unit configured to send the target information to at least two network devices, the at least two network devices providing coherent joint transmission for the terminal.

34. A channel state information obtaining apparatus applied to a network device, comprising: A first receiving unit, configured to receive target information sent by a terminal, the target information being obtained by compressing a plurality of first downlink channel state information (CSI), each first downlink CSI corresponding to one network device; A third obtaining unit, configured to decompress the target information to obtain a plurality of decompressed first downlink CSIs. 35.A processor-readable storage medium, the processor-readable storage medium storing a computer program, the computer program being configured to cause the processor to perform the method of any one of claims 1 to 16.

Images