Communication method and apparatus

Abstract

An embodiment of the present application provides a communication method and apparatus. The method includes: an encoder compresses M layers of channel information by using at least two artificial intelligence (AI) models to obtain N pieces of compressed information, where each of the N pieces of compressed information is obtained by compressing channel information of some layers in the M layers of channel information, where N and M are integers greater than 1 and N is less than M; the encoder sends the N pieces of compressed information to a decoder; compared to conventional schemes, the present application, which can feedback channel information of multiple layers by using AI networks, can reduce the overhead caused by feedbacking channel information; further, N is less than M, i.e., joint compression is performed on channel information of at least two layers of the M layers; compared to independent compression of the channel information of each of the M layers, this can further reduce feedback overhead; further, N is greater than 1; compared to one joint compression of the channel information of the M layers, this can reduce computational complexity.

Description

[Technical Field]

[0001] The present application relates to the field of communications technology, and more particularly to communications methods and apparatus. [Background technology]

[0002] This application claims priority to Chinese Patent Application No. 202211038332.X, entitled "COMMUNICATION METHOD AND APPARATUS," filed with the State Intellectual Property Office of China on August 29, 2022, which is incorporated herein by reference in its entirety.

[0003] In order for a network device to support transmitting multiple streams of data in parallel to a terminal device, the network device needs to determine precoding matrices for the multiple streams based on channel information, such as channel state information (CSI), fed back by the terminal device. The number of streams in this specification may also be referred to as the number of ranks or the number of layers. The overhead of feeding back channel information increases with the number of streams. Summary of the Invention

[0004] The present application provides a communication method and apparatus for reducing overhead caused by channel information feedback.

[0005] According to a first aspect, a communication method is provided. This method can be implemented by an encoder, or by a chip or circuit used in the encoder. This is not limited in the present application. For ease of explanation, the following uses an example in which this method is implemented by an encoder for explanation.

[0006] The method may include: an encoder compressing M layers of channel information by using at least two artificial intelligence (AI) models to obtain N pieces of compressed information, where each of the N pieces of compressed information is obtained by compressing channel information of some layers in the M layers of channel information by using at least one of the at least two AI models, where N and M are integers greater than 1 and N is less than M; and the encoder sending the N pieces of compressed information to a decoder.

[0007] Based on the above technical solution, the encoder compresses channel information of M layers by using at least two AI models to obtain N pieces of compressed information and sends the N pieces of compressed information to the decoder, so that the decoder can perform decoding based on the N pieces of compressed information to restore the channel information of the M layers. In this way, channel information of multiple layers can be fed back by using an AI network. Compared with conventional schemes, this can reduce the overhead caused by feeding back channel information. Furthermore, the N pieces of compressed information are obtained based on the channel information of M layers, where N is less than M. In other words, joint compression is performed on the channel information of at least two of the M layers. Compared with separate compression on the channel information of each of the M layers, this can further reduce the feedback overhead by utilizing the relationship between different channel information between layers. Furthermore, N is greater than 1. Compared with one joint compression on the channel information of the M layers, this can reduce the computational complexity.

[0008] Regarding the first aspect, in some implementations of the first aspect, the N pieces of compressed information include at least one first compressed information and at least one second compressed information, where the first compressed information is obtained by performing joint compression on at least two layers of channel information among the M layers of channel information by using a first AI model among the at least two AI models, and the second compressed information is obtained by performing individual compression on one layer of channel information among the M layers of channel information by using a second AI model among the at least two AI models.

[0009] Based on the above technical solution, the compression method of the channel information of M layers includes individual compression and joint compression, so that not only can the overhead caused by the individual compression of the channel information of M layers be reduced, but also the computational complexity caused by one joint compression of the channel information of M layers be reduced.

[0010] Regarding the first aspect, in some implementations of the first aspect, the compression scheme for the channel information of the M layers is determined based on at least one of the following information: a computing resource of an encoder, a computing resource of a decoder, an AI model of the encoder, an AI model of the decoder, a performance of performing joint compression on the channel information of some layers among the channel information of the M layers, a performance of performing individual compression on the channel information of some layers among the channel information of the M layers, and the value of M. The compression scheme for the channel information of the M layers includes joint compression, or the compression scheme for the channel information of the M layers includes joint compression and individual compression.

[0011] Separate compression indicates that separate compression is performed on the channel information of one layer, for example, the channel information of one layer is input independently into the AI ​​model to obtain one compressed piece of information.

[0012] Joint compression refers to joint compression performed on at least two layers of channel information, e.g., at least two layers of channel information are input into an AI model to obtain one compressed piece of information.

[0013] In one example, the compression method for the channel information of the M layers is determined based on the computing resources of the encoder. For example, if the computing resources of the encoder are large, for example, if the available computing power used by the encoder to compress the channel information is greater than or equal to a preset value, the compression method for the channel information of the M layers may include joint compression. In another example, if the available computing power used by the encoder to compress the channel information is less than a preset value, individual compression may be performed on the channel information of each layer.

[0014] In another example, the compression method of the channel information of the M layers is determined based on the computing resources of the decoder. For example, if the computing resources of the decoder are large, for example, if the available computing power used by the decoder to compress the channel information is greater than or equal to a preset value, the compression method of the channel information of the M layers may include joint compression.

[0015] In another example, the compression scheme for the channel information of the M layers is determined based on an AI model of the encoder. For example, if the AI ​​model of the encoder is applicable to joint compression (e.g., an AI model applicable to individual compression has been updated), the compression scheme for the channel information of the M layers may be joint compression. If the AI ​​model of the encoder is applicable to individual compression (e.g., an AI model applicable to joint compression has been updated), the compression scheme for the channel information of the M layers may be individual compression.

[0016] In another example, the compression scheme of the channel information of the M layers is determined based on an AI model of the decoder. For example, if the AI ​​model of the decoder is applicable to joint compression decoding, the compression scheme of the channel information of the M layers may include joint compression. If the AI ​​model of the decoder is applicable to individual compression decoding, the compression scheme of the channel information of the M layers may include individual compression.

[0017] In another example, the compression scheme for the channel information of the M layers is determined based on the performance of performing joint compression on the channel information of some layers among the channel information of the M layers. For example, if the performance of performing joint compression on the channel information of some layers among the channel information of the M layers is greater than or equal to a preset value, the compression scheme for the channel information of the M layers may include joint compression.

[0018] In another example, the compression scheme for the channel information of the M layers is determined based on the performance of performing individual compression on the channel information of some layers in the channel information of the M layers. For example, if the performance of performing individual compression on the channel information of some layers in the channel information of the M layers is greater than or equal to a preset value, the compression scheme for the channel information of the M layers may include individual compression.

[0019] In another example, the compression scheme of the channel information of the M layers is determined based on the value of M. For example, when M is greater than 1, the compression scheme of the channel information of the M layers includes joint compression. In another example, when M is greater than 1 and less than or equal to a first preset value, the compression scheme of the channel information of the M layers is individual compression, or when M is greater than the first preset value, the compression scheme of the channel information of the M layers includes joint compression.

[0020] It may be understood that one or more of the above examples may be used in combination to determine a compression scheme for the channel information of the M layers, and that, for example, the compression scheme is determined based on the computing resources and AI model of the encoder. If the available computing power used by the encoder to compress the channel information is greater than or equal to a preset value and the AI ​​model of the encoder is applicable to joint compression, the compression scheme for the channel information of the M layers may include joint compression.

[0021] Based on the above technical solutions, the compression schemes of the channel information of M layers can be determined based on one or more of the above information. In this way, an appropriate compression scheme can be dynamically selected based on the actual communication status, for example, some information of the encoder and / or decoder, the number of layers M, or the performance of individual compression or joint compression.

[0022] Regarding the first aspect, in some implementations of the first aspect, the method further includes: the encoder determines a compression scheme for the channel information of the M layers, where the compression scheme for the channel information of the M layers includes joint compression, or the compression scheme for the channel information of the M layers includes joint compression and individual compression.

[0023] Based on the above technical solution, the encoder may determine a compression scheme for the channel information of M layers, and may further compress the channel information of M layers based on the compression scheme determined by the encoder and of the channel information of the M layers to obtain N compressed information. In this way, the encoder may dynamically select an appropriate compression scheme based on the actual communication status, and the scheme is flexible.

[0024] Regarding the first aspect, in some implementations of the first aspect, the method further includes: the encoder sends a compression scheme of the channel information of the M layers to the decoder.

[0025] Based on the above technical solution, the encoder can send the compression scheme of the channel information of M layers to the decoder, so that the decoder obtains the channel information of M layers through decoding based on the compression scheme of the channel information of M layers and the received N compressed information.

[0026] Regarding the first aspect, in some implementations of the first aspect, the method further includes: an encoder receives a compression scheme for the channel information of M layers, where the compression scheme for the channel information of the M layers includes joint compression, or the compression scheme for the channel information of the M layers includes joint compression and individual compression.

[0027] Based on the above technical solution, the decoder can determine the compression method of the channel information of M layers and send the compression method to the encoder. In this way, the decoder can dynamically select an appropriate compression method based on the actual communication status, and the overhead caused by the encoder determining the compression method can be reduced.

[0028] Regarding the first aspect, in some implementations of the first aspect, the method further includes: an encoder receiving a reference signal from a decoder, and the encoder performing channel measurement based on the reference signal to obtain channel information of the M layers.

[0029] Regarding the first aspect, in some implementations of the first aspect, the method further includes: the encoder determines the value of M based on the result of the channel measurement.

[0030] Regarding the first aspect, in some implementations of the first aspect, the encoder compressing M layers of channel information by using at least two artificial intelligence (AI) models includes: determining a first AI model based on an arrangement manner of channel information in each layer and between layers in the X layers, where the at least two AI models include the first AI model, and the M layers include X layers, where X is an integer greater than 1 and less than M; and performing joint compression on the X layers of channel information by using the first AI model to obtain first compressed information, where the N pieces of compressed information include the first compressed information.

[0031] For example, the arrangement of channel information in each layer and between layers among the X layers includes any one of adjacent arrangement of channel information of different subbands in the same layer and adjacent arrangement of channel information of different layers in the same subband.

[0032] For example, the encoder determining the first AI model based on the arrangement manner of channel information for each layer and between layers in the X layers includes: The encoder calculates the accuracy of feeding back channel information in the joint compression scheme based on the arrangement manner of channel information for each layer and between layers in the X layers, and further determines an AI model to be used for the joint compression, i.e., the first AI model.

[0033] Based on the above technical solution, the encoder may perform joint compression on the channel information of X layers based on the arrangement manner of the channel information of each layer and between layers in the X layers. The decoder may perform decoding based on the compressed information obtained by compressing the channel information of the X layers to obtain the channel information of the X layers, and may correctly parse the output of the decoder based on the arrangement manner of the channel information of each layer and between layers in the X layers.

[0034] Regarding the first aspect, in some implementations of the first aspect, the method further includes: the encoder receives from the decoder an arrangement manner of channel information for each layer and between layers in the X layers, or the encoder sends to the decoder an arrangement manner of channel information for each layer and between layers in the X layers.

[0035] For example, there is a correspondence between the index value and the arrangement manner of channel information for each layer and between layers among the X layers. For example, the encoder receives the arrangement manner of channel information for each layer and between layers among the X layers from the decoder. In a possible implementation, the encoder receives the index value from the decoder, and the index value indicates the arrangement manner of channel information for each layer and between layers among the X layers. In other words, the encoder may determine the arrangement manner corresponding to the index value based on the index value and the correspondence between the index value and the arrangement manner of channel information for each layer and between layers among the X layers.

[0036] Based on the above technical solution, the decoder can determine the arrangement of channel information in each layer and between layers in X layers, and send the arrangement to the encoder. In this way, the encoder can perform encoding based on the arrangement indicated by the decoder, so that the decoder can correctly parse the decoded information.

[0037] Regarding the first aspect, in some implementations of the first aspect, the method further includes: the encoder sends first information to the decoder, where the first information indicates a sequence of channel information of different layers during joint compression performed on the channel information of at least two layers among the channel information of the M layers.

[0038] For example, there is a correspondence between the index value and the sequence of channel information of different layers. In a possible implementation, the first information is an index value, and the index value indicates the sequence of channel information of different layers. In other words, the decoder determines the sequence of channel information of different layers and corresponding to the index value based on the index value and the correspondence between the index value and the sequence of channel information of different layers.

[0039] Based on the above technical solution, the encoder can compress the jointly compressed channel information based on the sequence of the channel information during the joint compression, and the decoder can correctly parse the output of the decoder based on the sequence of the channel information during the joint compression.

[0040] Regarding the first aspect, in some implementations of the first aspect, the encoder is a terminal device and the decoder is a network device.

[0041] Regarding the first aspect, in some implementations of the first aspect, the channel information corresponding to each of the N pieces of compressed information does not overlap.

[0042] According to a second aspect, a communication method is provided. This method can be implemented by a decoder, or by a chip or circuit used in a decoder. This is not limited in the present application. For ease of explanation, the following uses an example in which this method is implemented by a decoder for explanation.

[0043] The method may include: a decoder receiving N pieces of compressed information from an encoder, where the N pieces of compressed information are obtained by compressing channel information of M layers by using at least two artificial intelligence (AI) models, and each of the N pieces of compressed information is obtained by compressing channel information of some layers in the channel information of the M layers by using at least one of the at least two AI models, where N and M are integers greater than 1 and N is less than M. The decoder decodes the N pieces of compressed information to obtain the channel information of the M layers.

[0044] Regarding the second aspect, in some implementations of the second aspect, the N pieces of compressed information include at least one first compressed information and at least one second compressed information, where the first compressed information is obtained by performing joint compression on at least two layers of channel information among the M layers of channel information by using a first AI model among the at least two AI models, and the second compressed information is obtained by performing individual compression on one layer of channel information among the M layers of channel information by using a second AI model among the at least two AI models.

[0045] Regarding the second aspect, in some implementations of the second aspect, the compression scheme for the channel information of the M layers is determined based on at least one of the following information: the computing resources of the encoder, the computing resources of the decoder, the AI ​​model of the encoder, the AI ​​model of the decoder, the performance of performing joint compression on the channel information of some layers among the channel information of the M layers, the performance of performing individual compression on the channel information of some layers among the channel information of the M layers, and the value of M. The compression scheme for the channel information of the M layers includes joint compression, or the compression scheme for the channel information of the M layers includes joint compression and individual compression.

[0046] Regarding the second aspect, in some implementations of the second aspect, the decoder decoding the N pieces of compressed information includes: the decoder decodes the N pieces of compressed information based on a compression scheme of the channel information of the M layers, where the compression scheme of the channel information of the M layers includes joint compression, or the compression scheme of the channel information of the M layers includes joint compression and individual compression.

[0047] Regarding the second aspect, in some implementations of the second aspect, the method further includes: the decoder determines a compression scheme for the channel information of the M layers.

[0048] Regarding the second aspect, in some implementations of the second aspect, the method further includes: the decoder sends a compression scheme of the channel information of the M layers to the encoder.

[0049] Regarding the second aspect, in some implementations of the second aspect, the method further includes: the decoder receives a compression scheme of the channel information of the M layers from the encoder.

[0050] Regarding the second aspect, in some implementations of the second aspect, the decoder decoding the N pieces of compressed information to obtain channel information of M layers includes: the decoder decodes first compressed information based on an arrangement manner of channel information of each layer and between layers in the X layers to obtain channel information of X layers, where the first compressed information is obtained by performing joint compression on the channel information of the X layers by using a first AI model of the at least two AI models, the N pieces of compressed information include the first compressed information, and the M layers include X layers, where X is an integer greater than 1 and less than M.

[0051] For example, the arrangement of channel information in each layer and between layers among the X layers includes any one of adjacent arrangement of channel information of different subbands in the same layer and adjacent arrangement of channel information of different layers in the same subband.

[0052] Based on the above technical solution, the decoder may input the first compressed information into an AI model to obtain channel information of X layers, and parse the output based on the arrangement manner of the channel information of each layer and between layers in the X layers.

[0053] Regarding the second aspect, in some implementations of the second aspect, the method further includes: the decoder sends to the encoder an arrangement manner of channel information for each layer and between layers in the X layers, or the decoder receives from the encoder an arrangement manner of channel information for each layer and between layers in the X layers.

[0054] Regarding the second aspect, in some implementations of the second aspect, the method further includes: a decoder receives first information from an encoder, where the first information indicates a sequence of channel information of different layers during joint compression performed on channel information of at least two layers among the channel information of the M layers.

[0055] Regarding the second aspect, in some implementations of the second aspect, the encoder is a terminal device and the decoder is a network device.

[0056] Regarding the second aspect, in some implementations of the second aspect, the channel information corresponding to each of the N pieces of compressed information does not overlap.

[0057] For the beneficial effects and possible designs of the second embodiment, please refer to the relevant descriptions in the first embodiment, and the details will not be described again here.

[0058] According to a third aspect, a communication device is provided. The device is configured to implement the method provided in any one of the first to second aspects. Specifically, the device may include units and / or modules, such as a processing unit and / or a communication unit, configured to implement the method provided in any one of the above implementations of the first or second aspect.

[0059] In one implementation, the apparatus is a communication device (e.g., an encoder or decoder). When the apparatus is a communication device, the communication unit may be a transceiver or an input / output interface, and the processing unit may be at least one processor. Optionally, the transceiver may be a transceiver circuit. Optionally, the input / output interface may be an input / output circuit.

[0060] In another implementation, the apparatus is a chip, chip system, or circuit used in a communication device. When the apparatus is a chip, chip system, or circuit used in a terminal device, the communication unit may be an input / output interface, an interface circuit, an output circuit, an input circuit, a pin, an associated circuit, etc. on the chip, chip system, or circuit, and the processing unit may be at least one processor, processing circuit, logic circuit, etc.

[0061] According to a fourth aspect, there is provided a communications device, the device including: a memory configured to store a program; and at least one processor configured to execute a computer program or instructions stored in the memory to perform a method provided in any one of the above implementations of the first or second aspect.

[0062] In one implementation, the apparatus is a communications device (eg, an encoder or decoder).

[0063] In another implementation, the apparatus is a chip, chip system, or circuit used in a communications device.

[0064] According to a fifth aspect, the present application provides a processor configured to perform the method provided in the above aspect.

[0065] Operations such as sending and acquiring / receiving related to a processor may be understood as operations such as output and input of a processor, or operations such as sending and receiving performed by a radio frequency circuit and an antenna, unless otherwise specified or if those operations do not contradict the actual function or internal logic of the operations in the relevant description, which is not limited in this application.

[0066] According to a sixth aspect, there is provided a computer-readable storage medium, the computer-readable storage medium storing program code for execution by a device, the program code being used to implement the method provided in any one of the implementations of the first or second aspect.

[0067] According to a seventh aspect, there is provided a computer program product comprising instructions, which when run on a computer, enable the computer to carry out the method of any one of the implementations of the first or second aspect.

[0068] According to an eighth aspect, a chip is provided, the chip including a processor and a communication interface, the processor reads instructions stored in a memory by using the communication interface to perform the method provided in either one of the implementations of the first or second aspect.

[0069] Optionally, in one implementation, the chip further includes a memory, the memory storing a computer program or instructions, and the processor configured to execute the computer program or instructions stored in the memory, such that when the computer program or instructions are executed, the processor is configured to perform the method provided in any one of the implementations of the first or second aspect.

[0070] According to a ninth aspect, there is provided a communication system including an encoder and a decoder.

[0071] In this application, an encoder may also be called a coding device and may have other functions besides encoding, and a decoder may also be called a decoding device and may have other functions besides decoding. [Brief explanation of the drawings]

[0072] [Figure 1] 1 is a diagram of a wireless communication system 100 applicable to an embodiment of the present application. [Figure 2] 2 is a diagram of a communication method 200 according to an embodiment of the present application. [Figure 3] FIG. 1 is a diagram showing a scheme for arranging channel information in each layer and between layers among X layers. [Figure 4] FIG. 10 is another diagram illustrating the arrangement of channel information in each layer and between layers among X layers. [Figure 5] FIG. 1 is an encoding and decoding diagram for compressing M layers of channel information. [Figure 6] FIG. 10 is a diagram of the decoder output. [Figure 7] 7 is a diagram of a communication method 700 according to an embodiment of the present application. [Figure 8] FIG. 10 is a diagram of channel information output by model A. [Figure 9] FIG. 10 is a diagram of channel information output by model B. [Figure 10] 10 is a diagram of a communication method 1000 according to an embodiment of the present application. [Figure 11] 11 is a block diagram of a communication device 1100 according to an embodiment of the present application. [Figure 12] 12 is a diagram of another communication device 1200 according to an embodiment of the present application. [Figure 13] FIG. 13 is a diagram of a chip system 1300 according to an embodiment of the present application. DETAILED DESCRIPTION OF THE INVENTION

[0073] The following describes the technical solutions in the embodiments of the present application with reference to the accompanying drawings.

[0074] The technical solutions provided in this application may be applied to various communication systems, such as 5th generation (5G) systems or new radio (NR) systems, long term evolution (LTE) systems, LTE frequency division duplex (FDD) systems, and LTE time division duplex (TDD) systems. The technical solutions provided in this application may also be applied to future communication systems, such as 6th generation mobile communication systems. The technical solutions provided in this application may also be applied to device-to-device (D2D) communication, vehicle-to-everything (V2X) communication, machine-to-machine (M2M) communication, machine type communication (MTC), and internet of things (IoT) communication systems or other communication systems.

[0075] The terminal device in the embodiments of the present application may include various devices having wireless communication capabilities and be configured to connect to people, objects, machines, etc. The terminal device may be widely used in various scenarios, such as cellular communication, D2D, V2X, peer-to-peer (P2P), M2M, MTC, IoT, virtual reality (VR), augmented reality (AR), industrial control, autonomous driving, telemedicine, smart grid, smart furniture, smart office, smart wearable, smart transportation, smart city drones, robots, remote sensing, passive sensing, positioning, navigation and tracking, and autonomous delivery. The terminal device may be a terminal in any one of the above scenarios, such as an MTC terminal or an IoT terminal.A terminal device may be a user equipment (UE), terminal, fixed device, mobile station device or mobile device, subscriber unit, handheld device, in-vehicle device, wearable device, cellular phone, smartphone, SIP phone, wireless data card, personal digital assistant (PDA), computer, tablet computer, notebook computer, wireless modem, handheld device (handset), laptop computer, computer with wireless transceiver functionality, smartbook, vehicle, satellite, global positioning system (GPS) device, target tracking device, or flying device (e.g., unmanned aerial vehicle). The terminal device may be a mobile phone, a helicopter, a multi-copter, a four-copter, or an airplane, a ship, a remote control device, a smart home device, or an industrial device, or may be a device built into the above device (e.g., a communication module, a modem, or a chip in the above device), or may be another processing device connected to a wireless modem. For ease of explanation, the following uses a terminal or a UE as an example of a terminal device for explanation.

[0076] It should be appreciated that in some scenarios, the UE may also act as a base station, for example, a scheduling entity providing sidelink signals between UEs in scenarios such as V2X, D2D, or P2P.

[0077] In an embodiment of the present application, an apparatus configured to implement the functions of a terminal device may be the terminal device, or may be an apparatus capable of supporting the terminal device to implement the functions, such as a chip system or a chip, and the apparatus may be installed in the terminal device. In an embodiment of the present application, the chip system may include a chip, or may include a chip and another discrete component.

[0078] The network device in the embodiment of the present application may be a device configured to communicate with a terminal device. The network device may alternatively be referred to as an access network device or a radio access network device. For example, the network device may be a base station. The network device in the embodiment of the present application may be a radio access network (RAN) node (or device) that connects the terminal device to a wireless network. The base station may broadly cover or be interchangeable with various names, such as, for example, a NodeB (NodeB), evolved NodeB (eNB), next-generation NodeB (gNB), relay station, access point, transmission reception point (TRP), transmission point (TP), primary station, secondary station, motor slide retainer (MSR) node, home base station, network controller, access node, wireless node, access point (AP), transmitting node, transceiver node, baseband unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), central unit (CU), distributed unit (DU), positioning node, etc. The base station may be a macro base station, a micro base station, a relay node, a donor node, etc., or a combination thereof. The base station may alternatively be a communication module, modem, or chip disposed in the above device or apparatus. The base station may alternatively be a mobile switching center, a device implementing base station functionality in D2D, V2X, and M2M communications, a network-side device in a 6G network, a device implementing base station functionality in future communication systems, etc.The base stations may support networks of the same access technology or different access technologies. The specific technology used for the network devices and the specific device configuration are not limited to the embodiments of the present application.

[0079] The base station may be fixed or mobile. For example, a helicopter or unmanned aerial vehicle may be configured as a mobile base station, and one or more cells may move based on the location of the mobile base station. In another example, a helicopter or unmanned aerial vehicle may be configured as a device for communicating with another base station.

[0080] In an embodiment of the present application, an apparatus configured to implement the functions of a network device may be a terminal device, or may be an apparatus capable of supporting a network device to implement the functions, such as a chip system or a chip, and the apparatus may be installed in the network device. In an embodiment of the present application, the chip system may include a chip, or may include a chip and another discrete component.

[0081] The network devices and terminal devices may be deployed on land, on water, or on airborne aircraft, balloons, and satellites, including indoor or outdoor scenarios and handheld or vehicle-mounted scenarios. The scenarios in which the network devices and terminal devices are located are not limited in the embodiments of the present application.

[0082] First, a network architecture applicable to the embodiments of the present application is briefly described as follows.

[0083] FIG. 1 is a diagram of a wireless communication system 100 applicable to one embodiment of the present application. As shown in FIG. 1, the wireless communication system 100 may include at least one network device, for example, the network device 110 shown in FIG. 1. The wireless communication system 100 may further include at least one terminal device, for example, the terminal device 120 and the terminal device 130 shown in FIG. 1. Multiple antennas may be configured for both the network device and the terminal device, and the network device and the terminal device may communicate with each other by using multi-antenna technology. The terminal devices may also communicate with each other. For example, the terminal devices may communicate with each other directly. In another example, the terminal devices may communicate with each other by using another communication device, for example, a network device or another terminal device.

[0084] When a network device communicates with a terminal device, the network device may manage one or more cells, and a cell may have an integer number of terminal devices. Optionally, network device 110 and terminal device 120 form a single-cell communication system. Without loss of generality, the cell is referred to as cell #1. Network device 110 may be the network device in cell #1, or network device 110 may serve a terminal device (e.g., terminal device 120) in cell #1.

[0085] It should be noted that a cell may be understood as an area within the coverage of a wireless signal of a network device.

[0086] It should be understood that Fig. 1 is merely a simplified diagram of an example for ease of understanding. The wireless communication system 100 may further include another network device or may further include another terminal device not shown in Fig. 1. The embodiments of the present application are applicable to any communication scenario in which a transmitting end device communicates with a receiving end device.

[0087] To facilitate understanding of the embodiments of the present application, the following first provides a brief explanation of terms used in the embodiments of the present application.

[0088] 1. Artificial intelligence (AI) enables machines to learn, accumulate experience, and solve problems such as natural language understanding, image recognition, and chess playing that can be solved by humans through experience.

[0089] 2. Machine learning is an implementation of artificial intelligence. Machine learning is a method that provides machines with the ability to learn to complete functions that cannot be implemented by direct programming. In practice, machine learning is a method of training a model by using data and then using the model to make predictions.

[0090] 3. A neural network is a specific embodiment of a machine learning method. A neural network is a mathematical model that processes information and mimics the behavioral characteristics of animal neural networks. A neural network may include three types of computing layers: an input layer, a hidden layer, and an output layer. Each layer has one or more logical decision-making units, which may be called neurons. Common neural network structures include feedforward neural networks (FNNs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), etc. These network structures are based on neurons. Each neuron may perform a weighted sum operation on its input values ​​and generate an output based on the weighted sum result by using a nonlinear function. The weight values ​​for performing the weighted sum operation by the neurons in a neural network and the nonlinear function may be called the parameters of the neural network. The connectivity between neurons in a neural network may be referred to as the structure of the neural network, and the parameters of all the neurons in the neural network may form the parameters of the neural network.

[0091] 4. A deep neural network is a neural network that has multiple hidden layers.

[0092] 5. Deep learning is machine learning based on deep neural networks.

[0093] 6. A precoding matrix indicator (PMI) may indicate a precoding matrix. The precoding matrix may be determined by a terminal device based on a channel matrix of each frequency band. For example, the channel matrix may be determined by a terminal device using a channel estimation method. A vector in the precoding matrix may be referred to as a precoding vector.

[0094] For example, to obtain a precoding matrix that can adapt to a channel, a network device may perform channel measurements in advance by sending a reference signal to a terminal device, and the terminal device may determine a precoding vector for each frequency band based on the channel matrix for each frequency band.

[0095] Assuming that the channel matrix of the frequency band is H, the terminal device uses the channel matrix H or the covariance matrix HH of the channel matrix to determine the precoding matrix of the frequency band. H We can perform singular value decomposition (SVD) on H. The superscript H denotes the conjugate transpose. For example, H H denotes the conjugate transpose of the matrix (or vector) H. Then, the terminal device may quantize each element in the precoding matrix of each frequency band and feed back the quantized value to the network device by using the PMI, so that the network device restores the precoding matrix based on the PMI.

[0096] The above description of PMI is merely an example and does not constitute any limitation on the present application.

[0097] With the development of wireless communication technology, more and more services are supported, imposing increasingly higher requirements on communication systems with respect to indicators such as system capacity and communication delay. Communication rates can be effectively improved by expanding the bandwidth available to a UE. To support wider available bandwidths, multiple contiguous or non-contiguous frequency bands may be allocated to a UE for communication. To support effective communication on each frequency band, a network device needs to acquire channel state information (CSI) on each frequency band. In a possible manner, a UE performs uplink feedback, and as a result, the network device acquires downlink CSI. Specifically, the network device sends a downlink reference signal to the UE, and the UE receives the downlink reference signal. The UE knows the sending information of the downlink reference signal. Therefore, the UE may estimate (or measure) the downlink channel through which the downlink reference signal passes based on the sending information of the downlink reference signal and the downlink reference signal received by the UE. The UE then obtains a downlink channel matrix based on the measurement, generates CSI, and feeds back the generated CSI to the network device.

[0098] An important part of the CSI feedback is the PMI. Specifically, one or more bit values ​​are used in the CSI to quantize the channel matrix or precoding matrix. For example, the channel matrix or precoding matrix may be quantized by using a "0" bit value and / or a "1" bit value. As described above, the precoding matrix may be a precoding matrix determined by the UE based on the channel matrix of each frequency band.

[0099] In a conventional PMI design (sometimes called codebook design) method, a set of precoding matrices and corresponding numbers are predefined in a protocol. These precoding matrices are sometimes called codewords or precoding information. The channel matrix or precoding matrix can be approximated by using a predefined codeword or a linear combination of multiple predefined codewords. The UE then feeds back the number corresponding to the codeword and at least one weighting coefficient to the network device using the PMI, allowing the network device to reconstruct the channel matrix or precoding matrix. Higher accuracy of the CSI fed back by the UE indicates richer and more accurate channel information reconstructed by the network device based on the CSI. Therefore, the precoding matrix determined by the network device is more accurate, the signal-to-interference-and-noise ratio of the signal received by the UE is higher, and the system capacity is greater. However, as the scale of the antenna array in the system increases, the number of antenna ports that can be supported increases. Since the size of the complete channel matrix is ​​directly proportional to the number of antenna ports, feeding back the complete channel matrix to network devices by using CSI in massive multiple-input multiple-output (Massive MIMO) systems implies huge feedback overhead.

[0100] To support parallel transmission of multi-stream data to a UE, a network device needs to reconstruct a precoding matrix for multi-streaming based on CSI feedback information. The number of streams in this specification may also be referred to as a rank quantity or a layer quantity, and will be collectively referred to as a layer quantity hereinafter. In a conventional multi-layer CSI feedback scheme, for a precoding matrix in each layer, feedback information is generated based on a codebook by using the above method, so that the network device can reconstruct the precoding matrix in the layer. Therefore, the CSI feedback overhead increases as the number of layers increases.

[0101] Machine learning, for example, deep learning, has strong nonlinear feature extraction capabilities and can extract correlations between different layers. Therefore, compared with conventional solutions, machine learning can include more channel information in the same scale of feedback. This reduces the information loss of CSI compressed feedback and improves the accuracy of channel reconstruction at the network device side. Furthermore, compared with conventional solutions, a smaller amount of feedback can be used to represent the same channel information, further reducing feedback overhead.

[0102] In this regard, the present application provides a scheme in which an AI network is used to feedback multi-layer channel information, so that the feedback overhead can be reduced and the computational complexity of the AI ​​network can also be reduced.

[0103] It can be understood that the principle of feeding back channel information by using an AI network may be as follows: the input of the AI ​​model deployed in the encoder is a group of channel information, the output of the AI ​​model deployed in the encoder is compressed information, the input of the AI ​​model deployed in the decoder is compressed information, and the output of the AI ​​model deployed in the decoder is a group of channel information. An example of a terminal device feeding back channel information to a network device is used for explanation. For example, the terminal device may use a training set and input the channel data in the training set to the encoder to obtain an output. The error caused by compressing the channel data may be used as a loss function, and the loss function is a function of the weights of the AI ​​model of the encoder. If the loss function is smaller than a threshold, training is stopped. Otherwise, the weights of the encoder are updated to reduce the loss function.

[0104] It may be further understood that the AI ​​model may be implemented using hardware circuitry, software, or a combination of software and hardware. This is not limited to this. Non-limiting examples of software include program code, programs, subprograms, instructions, instruction sets, code, code segments, software modules, applications, software applications, etc.

[0105] It should be noted that in this application, "indication" may include direct indication, indirect indication, explicit indication, and implicit indication. When indication information is described as indicating A, it may be understood that the indication information conveys A, directly indicates A, or indirectly indicates A.

[0106] In this application, information indicated by the indication information is referred to as information to be indicated. In a specific implementation, the information to be indicated may be indicated in multiple ways, including, but not limited to, the following ways: The information to be indicated may be directly indicated, for example, the information to be indicated or an index of the information to be indicated. Alternatively, the information to be indicated may be indirectly indicated by indicating other information, and there may be an association relationship between the other information and the information to be indicated. Alternatively, only a portion of the information to be indicated may be indicated, and other portions of the information to be indicated may be known or agreed upon in advance. For example, specific information may alternatively be indicated by using a pre-agreed (e.g., protocol-defined) arrangement sequence of multiple pieces of information to reduce indication overhead to some extent. Furthermore, the information to be indicated may be sent as a whole or may be divided into multiple sub-information pieces to be sent separately. Furthermore, the sending cycles and / or sending opportunities of these sub-information pieces may be the same or different.

[0107] The following describes in detail a communication method provided in an embodiment of the present application with reference to the accompanying drawings. The embodiment provided in the present application may be applied to, but is not limited to, the network architecture shown in Figure 1.

[0108] In the following embodiments, for example, the encoder is a terminal device and the decoder is a network device, or the encoder is a terminal device and the decoder is another terminal device, or the encoder is a network device and the decoder is another network device, or the encoder is a network device and the decoder is a terminal device. This is not limited. It may be understood that the following encoders may be replaced with encoding devices, and the decoders may be replaced with decoding devices. Furthermore, it may be further understood that the following encoders may include circuits other than the AI ​​model, such as processing circuits, memory circuits, and transceiver circuits.

[0109] In the following embodiments, multiple AI models may be deployed in an encoder, such that the encoder performs encoding based on the deployed AI models, e.g., compressing channel information. Multiple AI models may be deployed in a decoder, such that the decoder performs decoding based on the deployed AI models, e.g., decompressing compressed information to obtain channel information. For simplicity, hereinafter, an AI model deployed by an encoder is referred to as an encoder AI model, and an AI model deployed by a decoder is referred to as a decoder AI model.

[0110] 2 is a diagram of a communication method 200 according to an embodiment of the present application. The method 200 may include the following steps.

[0111] 210: An encoder compresses M layers of channel information by using at least two AI models to obtain N pieces of compressed information, where N and M are integers greater than 1 and N is less than M.

[0112] In a possible implementation, the encoder receives a reference signal from the decoder, and the encoder performs channel measurements based on the reference signal to obtain channel information for the M layers.

[0113] Each of the N pieces of compressed information is obtained by compressing channel information of some layers in the channel information of the M layers by using at least one of the at least two AI models. In other words, each of the N pieces of compressed information is obtained by compressing channel information of some layers in the channel information of the M layers by using an AI model of the at least two AI models.

[0114] For example, the channel information for M layers may be divided into N groups of channel information, and each group of channel information may be compressed using an AI model to obtain N compressed pieces of information, each group of channel information including channel information for at least one layer, and the N groups of channel information may be non-overlapping.

[0115] For example, the compression scheme of each group of channel information may be individual compression or joint compression. In the embodiments of the present application, the compression scheme is mentioned multiple times, and includes individual compression and joint compression.

[0116] For example, individual compression indicates that the channel information of one layer is individually compressed, i.e., the channel information of the layer is input into the AI ​​model to obtain one compressed piece of information.

[0117] Joint compression indicates that joint compression is performed on at least two layers of channel information (e.g., two layers of channel information, three layers of channel information, or four layers of channel information). For example, at least two layers of channel information are input into an AI model to obtain one compressed piece of information.

[0118] Optionally, the compression scheme of the channel information of the M layers is joint compression, or the compression scheme of the channel information of the M layers includes individual compression and joint compression.

[0119] If possible, the compression scheme for the M layers of channel information includes individual compression and joint compression. Based on this case, the N pieces of compressed information include at least one first compressed information and at least one second compressed information. The first compressed information is obtained by performing joint compression on at least two layers of channel information among the M layers of channel information by using a first AI model among the at least two AI models, and the second compressed information is obtained by performing individual compression on channel information of a single layer among the M layers of channel information by using a second AI model among the at least two AI models. In other words, the M layers of channel information are divided into N groups of channel information, and some of the groups of channel information among the N groups of channel information are compressed using the joint compression scheme, and other parts of the groups of channel information among the N groups of channel information are compressed using the individual compression scheme. The first AI model may represent an AI model used for joint compression, and the second AI model may represent an AI model used for individual compression.

[0120] In another possible case, the compression method for the channel information of the M layers is joint compression. In other words, the channel information of the M layers is divided into N groups of channel information, and each group of channel information is compressed using a joint compression method. In this case, the N compressed pieces of information include at least two first compressed pieces of information, and the first compressed pieces of information are obtained by performing joint compression on the channel information of at least two layers in the channel information of the M layers by using a first AI model in the at least two AI models.

[0121] For example, the compression scheme may be expressed in an array format. For example, the channel information of M layers may be divided into N groups of channel information, and the compression scheme of the channel information of M layers may indicate the N groups and the elements of each group in a numerical format. For example, the compression scheme of the channel information of M layers may be expressed as [Layer 1, Layer 2] and [Layer 3], which indicates that the channel information of M layers is divided into two groups, and joint compression is performed on the channel information of the first layer and the second layer in the first set, and individual compression is performed on the channel information of the third layer in the second set. In another example, the compression scheme of the channel information of M layers may be expressed as [Layer 1, Layer 2] and [Layer 3, Layer 4], which indicates that the channel information of M layers is divided into two groups, and joint compression is performed on the channel information of the first layer and the second layer in the first set, and joint compression is performed on the channel information of the third layer and the fourth layer in the second set. In another example, the compression scheme of the channel information of M layers is represented as [Layer 1], [Layer 2, Layer 3, Layer 4], and [Layer 5], which indicates that the channel information of the M layers is divided into three groups, and individual compression is performed on the channel information of the first layer in the first set, joint compression is performed on the channel information of the second layer, the third layer, and the fourth layer in the second set, and individual compression is performed on the channel information of the fifth layer in the third set.

[0122] Optionally, the AI ​​models corresponding to the N compressed information are partially the same, completely the same, or completely different. In other words, in the embodiment of the present application, the AI ​​models corresponding to all groups of channel information may be partially the same, completely the same, or completely different.

[0123] When possible, different compression methods correspond to different AI models. For example, individual compression corresponds to a second AI model, and joint compression corresponds to a first AI model. If the encoder performs individual compression on one layer of channel information, the second AI model is used. If the encoder performs joint compression on at least two layers of channel information, the first AI model is used.

[0124] In another possible case, each group of channel information corresponds to a different AI model. For example, assume that M layers of channel information are divided into three groups, denoted as a first group of channel information, a second group of channel information, and a third group of channel information. The encoder compresses the first group of channel information by using the first AI model, the encoder compresses the second group of channel information by using the second AI model, and the encoder compresses the third group of channel information by using the third AI model.

[0125] In an embodiment of the present application, the structure of the AI ​​model is not limited. For example, the structure of the AI ​​model may be, for example, any one of the following structures: an AI model based on a fully connected layer, a convolutional neural network, or a transformer structure. In an embodiment of the present application, in the case of an AI model deployed in an encoder, the input of the AI ​​model is a group of channel information, and the output of the AI ​​model is compressed information, the dimensionality of the compressed information is smaller than the dimensionality of the input channel information. In the case of an AI model deployed in a decoder, the input of the AI ​​model is compressed information, and the output of the AI ​​model is a group of channel information, the dimensionality of the compressed information is smaller than the dimensionality of the input channel information.

[0126] Furthermore, the method for determining the AI ​​models used by the encoder and decoder, i.e., the AI ​​models used by the encoder to compress the M layers of channel information and the AI ​​models used by the decoder to decode the N pieces of compressed information, is not limited. In a possible implementation, the method is predefined, for example, predefined in a standard or agreed upon in advance. In another possible implementation, the encoder provides the decoder with information regarding the AI ​​models used to compress the M layers of channel information (e.g., the AI ​​models used by the encoder to compress the M layers of channel information or identifiers of the AI ​​models used by the encoder to compress the M layers of channel information). The decoder determines the AI ​​models to use to decode the N pieces of compressed information based on the information regarding the AI ​​models and provided by the encoder. In another possible implementation, the encoder or decoder determines the AI ​​models to use based on the compression scheme. For example, the encoder may provide the decoder with the compression scheme, and the decoder may determine the AI ​​models to use to decode the N pieces of compressed information based on the compression scheme. In another example, the decoder may provide the compression scheme to the encoder, and the encoder may determine the AI ​​model to be used to compress the M layers of channel information based on the compression scheme.

[0127] 220: The encoder sends the N compressed information to the decoder.

[0128] Correspondingly, the decoder receives N compressed pieces of information.

[0129] Optionally, the method 200 further comprises step 230 .

[0130] 230: A decoder decodes the N compressed information to obtain the channel information of the M layers.

[0131] According to an embodiment of the present application, an encoder compresses M layers of channel information by using at least two AI models to obtain N pieces of compressed information and sends the N pieces of compressed information to a decoder, so that the decoder can perform decoding based on the N pieces of compressed information to restore the M layers of channel information. In this way, channel information of multiple layers can be fed back using an AI network. Compared with conventional schemes, this can reduce the overhead caused by feeding back channel information. Furthermore, the N pieces of compressed information are obtained based on the M layers of channel information, where N is less than M. In other words, joint compression is performed on the channel information of at least two layers among the M layers of channel information. Compared with individual compression of the channel information of each layer, this can further reduce the feedback overhead by utilizing the relationship between different channel information between layers. Furthermore, N is greater than 1. Compared with one joint compression of the M layers of channel information, this can reduce the computational complexity.

[0132] Optionally, the method 200 further includes: the encoder and the decoder know the compression schemes of the channel information of the M layers. In this way, the encoder may compress the channel information of the M layers based on the compression schemes of the channel information of the M layers to obtain N pieces of compressed information. The decoder may decode the N pieces of compressed information based on the compression schemes of the channel information of the M layers to obtain the channel information of the M layers. The following describes some possible solutions.

[0133] Solution 1: The encoder determines the compression method for the channel information of M layers.

[0134] Based on this solution, the encoder may determine a compression scheme for the channel information of the M layers, and may further compress the channel information of the M layers based on the compression scheme determined by the encoder and of the channel information of the M layers to obtain N pieces of compressed information, and send the N pieces of compressed information to the decoder. According to this solution, the encoder may dynamically select an appropriate compression scheme based on an actual communication status, and the scheme is flexible.

[0135] Optionally, in Solution 1, the encoder sends the compression scheme of the channel information of the M layers to the decoder, so that the decoder obtains the channel information of the M layers through decoding based on the compression scheme of the channel information of the M layers and the received N compressed information.

[0136] The procedure applicable to Solution 1 is described in detail below with reference to method 700.

[0137] Solution 2: The decoder determines the compression method for the channel information of the M layers.

[0138] According to this solution, the decoder can determine the compression method of the channel information of the M layers, and further obtain the channel information of the M layers through decoding based on the compression method of the channel information of the M layers determined by the decoder and the received N compressed information. According to this solution, the decoder can dynamically select an appropriate compression method based on the actual communication status, and this method is flexible.

[0139] Optionally, in Solution 2, the decoder sends a compression scheme of the M layers of channel information to the encoder. In this way, the encoder may compress the M layers of channel information based on the compression scheme of the M layers of channel information and indicated by the network device to obtain N pieces of compressed information, and send the N pieces of compressed information to the decoder.

[0140] The procedures applicable to Solution 2 are described in detail below with reference to Method 1000.

[0141] Solution 3: A separate device determines the compression method for the M layers of channel information.

[0142] Based on this solution, another device may determine a compression scheme for the channel information of M layers. According to this solution, another device may determine a compression scheme for the channel information of M layers, which reduces the overhead caused by determining the compression scheme for the channel information of M layers by the encoder and the decoder.

[0143] In Solution 3, another device may send the compression schemes of the channel information for the M layers to the decoder and / or the encoder. For example, the another device sends the compression schemes of the channel information for the M layers to the decoder and the encoder separately. In another example, the another device sends the compression schemes of the channel information for the M layers to the decoder. After receiving the compression schemes of the channel information for the M layers, the decoder sends the compression schemes of the channel information for the M layers to the encoder. In another example, the another device sends the compression schemes of the channel information for the M layers to the encoder. After receiving the compression schemes of the channel information for the M layers, the encoder sends the compression schemes of the channel information for the M layers to the decoder.

[0144] Solution 4: Pre-definition For example, the compression method of the channel information of M layers is pre-defined in the standard.

[0145] Based on this solution, the encoder and decoder may determine the compression scheme of the channel information of the M layers based on the predefinition. According to this solution, the encoder and decoder may directly determine the compression scheme of the channel information of the M layers based on the predefinition. This reduces the signaling overhead caused by informing the encoder and decoder of the compression scheme of the channel information of the M layers.

[0146] Optionally, the compression method for the channel information of the M layers is determined based on at least one of the capability information of the encoder, the capability information of the decoder, the performance of performing joint compression on the channel information of some layers among the channel information of the M layers, the performance of performing individual compression on the channel information of some layers among the channel information of the M layers, and the value of M.

[0147] In one example, the encoder capability information may include at least one of an encoder computing resource and an encoder AI model. In one example, the decoder capability information may include at least one of a decoder computing resource and a decoder AI model. It can be understood that the term "capability information" in this specification is used for brevity only, and the term "capability information" does not limit the scope of protection of the embodiments of the present application.

[0148] When possible, the above solution 1 is used as an example. The encoder determines a compression method for the channel information of M layers. In this case, the encoder may determine the compression method for the channel information of M layers based on the encoder capability information and / or the decoder capability information. In one example, the decoder sends the decoder capability information to the encoder. In another example, the encoder may estimate the decoder capability information based on a historical communication status between the encoder and the decoder, for example, estimating the decoder's computing resources and / or the decoder's AI model.

[0149] In another possible case, the above solution 2 is used as an example. The decoder determines the compression method for the channel information of the M layers. In this case, the decoder may determine the compression method for the channel information of the M layers based on the encoder capability information and / or the decoder capability information. In one example, the encoder sends the encoder capability information to the decoder. In another example, the decoder may estimate the encoder capability information based on the historical communication status between the decoder and the encoder, for example, to estimate the encoder's computing resources and / or the encoder's AI model.

[0150] The following describes some examples of determining the compression method of M layers of channel information. It can be understood that the following device for determining the compression method of M layers of channel information can be an encoder (i.e., the above solution 1), a decoder (i.e., the above solution 2), or another device (i.e., the above solution 3). This is not limited.

[0151] Example 1: The compression method for the channel information of M layers is determined based on the computing resources of the encoder.

[0152] The encoder's computing resources may represent the available computing power of the encoder for compressing the channel information. In one example, the encoder's computing resources may be determined by the encoder based on the configuration status of the encoder's computing resources (e.g., software resources or hardware resources), or the encoder's computing resources may be determined by the encoder based on the available computing resources of the encoder. The encoder's available computing resources may be determined by the encoder based on the available computing resources of the encoder when the encoder compresses the channel information.

[0153] For example, when the computing resources of the encoder are large, e.g., when the available computing power used by the encoder to compress the channel information is greater than or equal to a preset value, the compression scheme for the channel information of the M layers may include joint compression. For example, the channel information of the M layers is divided into N groups of channel information, and joint compression is performed individually on some of the groups of channel information, and individual compression is performed on other groups of channel information. In another example, when the available computing power used by the encoder to compress the channel information is less than a preset value, individual compression may be performed on the channel information of each layer.

[0154] Example 2: The compression method for the channel information of M layers is determined based on the computing resources of the decoder.

[0155] For the computing resources of the decoder, please refer to the above description of the computing resources of the encoder, and the details will not be described again here.

[0156] For example, when the computing resource of the decoder is large, e.g., when the available computing power used by the decoder to compress the channel information is greater than or equal to a preset value, the compression scheme for the channel information of the M layers may include joint compression, e.g., the channel information of the M layers is evenly divided into N groups of channel information, and joint compression is performed on each group of channel information.

[0157] Example 3: The compression method for the channel information of M layers is determined based on the AI ​​model of the encoder.

[0158] For example, if the AI ​​model of the encoder is applicable to joint compression (e.g., an AI model applicable to individual compression has been updated), the compression scheme of the channel information of the M layers may be joint compression. If the AI ​​model of the encoder is applicable to individual compression (e.g., an AI model applicable to joint compression has been updated), the compression scheme of the channel information of the M layers may be individual compression.

[0159] Example 4: The compression method for the channel information of M layers is determined based on the AI ​​model of the decoder.

[0160] For example, if the AI ​​model of the decoder is applicable to joint compression decoding, the compression scheme of the channel information of the M layers may include joint compression. If the AI ​​model of the decoder is applicable to individual compression decoding, the compression scheme of the channel information of the M layers may include individual compression.

[0161] Example 5: A compression method for channel information of M layers is determined based on the performance of performing joint compression on some channel information of some layers among the channel information of M layers.

[0162] For example, if the performance of performing joint compression on some layers of channel information among the M layers of channel information is greater than or equal to a preset value, the compression scheme for the M layers of channel information may include joint compression. In another example, the M layers of channel information are divided into N groups of channel information. The encoder individually performs joint encoding on the channel information obtained when N is 1, 2, ..., or M-1, and determines the performance of the joint compression. The encoder determines the value of N when the performance is optimal, and performs joint compression on the N groups of channel information.

[0163] Example 6: The compression method for the channel information of M layers is determined based on the performance of performing individual compression on the channel information of some layers among the channel information of M layers.

[0164] For example, if the performance of performing individual compression on the channel information of some layers among the channel information of the M layers is greater than or equal to a preset value, the compression scheme for the channel information of the M layers may include individual compression. In another example, the encoder performs individual encoding on the channel information of one layer, the channel information of two layers, ..., and the channel information of the M layers separately, and determines the performance of the individual compression. The encoder determines the value of M (e.g., M=M1) when the performance is optimal, performs individual compression on the channel information of the M1 layers, and performs joint compression on the remaining channel information.

[0165] Example 7: The compression method for the channel information of M layers is determined based on the value of M.

[0166] For example, when M is greater than 1, the compression method for the channel information of M layers includes joint compression, and which layers of channel information are jointly compressed and which layers of channel information are individually compressed may be randomly selected.

[0167] In another example, when M is greater than 1 and less than or equal to a first preset value, the compression scheme of the channel information of the M layers is individual compression, or when M is greater than a first preset value, the compression scheme of the channel information of the M layers includes joint compression, and which layer's channel information is subjected to joint compression and which layer's channel information is subjected to individual compression may be selected randomly.

[0168] The above describes that the compression method is determined based on any one of the above information. It can be understood that the above information can also be used in combination. The following briefly lists an example in which the above information is used in combination. Other examples in which the above information is used in combination will not be described again in this specification.

[0169] Example 8: The compression method of the channel information of M layers is determined based on the computing resources of the decoder and the AI ​​model of the encoder.

[0170] For example, if the computing resources of the decoder are large, e.g., the available computing power used by the decoder to compress the channel information is greater than or equal to a preset value, and the AI ​​model of the encoder is applicable to joint compression, the compression scheme for the channel information of the M layers may include joint compression.

[0171] Optionally, method 200 further includes: the encoder and decoder know the arrangement of channel information for each layer and between layers in the X layers, where X is an integer greater than 1 and less than M. In this manner, the encoder may compress the channel information of the X layers based on the arrangement of the channel information for each layer and between layers in the X layers to obtain one piece of compressed information (e.g., denoted as first compressed information). The decoder may decode the first compressed information to obtain the channel information for the X layers and correctly parse the output of the decoder based on the arrangement of the channel information for each layer and between layers in the X layers. For example, the decoder may input the first compressed information to an AI model to obtain the channel information for the X layers and parse the information output from the AI ​​model based on the arrangement of the channel information for each layer and between layers in the X layers.

[0172] Generally, the output of the decoder is N sub×M columns, each of which has N t is a complex vector with length N, which represents a channel eigenvector. The channel eigenvector may be used for precoding after being fed back to the decoder. The channel eigenvector may also be called a precoding vector, and multiple precoding vectors may form a precoding matrix. t represents the number of transmit antenna ports of the decoder, and N sub represents the number of subbands, and M represents the number of layers.

[0173] In order to correctly parse the decoder output, it is necessary to clarify the physical meaning of each element in the decoder output, i.e., the antenna port corresponding to that element, the subband corresponding to that element, and the layer corresponding to that element, i.e., the arrangement scheme of channel information for each layer and between layers among the M layers.

[0174] The arrangement of channel information for each layer and between layers among the X layers indicates the physical meaning of each element output by the decoder after the decoder decodes the compressed information, for example, the antenna port corresponding to the element, the subband corresponding to the element, and the layer corresponding to the element. It is assumed that the channel information for the M layers is divided into N groups of channel information, and joint compression is used for N1 groups of channel information among the N groups of channel information. For each group of channel information among the N1 groups of channel information, the arrangement of channel information for each layer and between layers may be the same or different. This is not limited. The following describes a related solution using the arrangement of channel information for each layer and between layers among the X layers.

[0175] In a possible arrangement scheme, the arrangement scheme of the channel information of each layer and between layers in the X layers is a layer-before-subband arrangement scheme, that is, the channel information of different layers in the same subband is arranged adjacently.

[0176] For example, the encoder may perform joint compression on the channel information of X layers based on a constellation scheme to obtain one compressed information (e.g., denoted as first compressed information), i.e., of the channel information of X layers, and the constellation scheme input by the encoder is a layer-before-subband constellation scheme. Correspondingly, the decoder may decode the first compressed information based on the constellation scheme, i.e., of the channel information of X layers, and the constellation scheme output by the decoder is a layer-before-subband constellation scheme.

[0177] 3 is a diagram of an arrangement scheme of channel information of each layer and between layers in X layers. As shown in FIG. 3, the X layers are three layers, respectively denoted as layer x, layer y, and layer z, and it is assumed that the channel information of the X layers corresponds to three subbands, respectively denoted as subband 1, subband 2, and subband 3. As shown in FIG. 3, the arrangement scheme output by the decoder is a layer-before-subband arrangement scheme. Specifically, first, eigenvectors in different layers in the same subband are arranged adjacently to form an eigenvector group of the subband, and then the eigenvector groups of different subbands are arranged sequentially.

[0178] In another possible arrangement manner, the arrangement manner of the channel information of each layer and between layers in the X layers is a subband-before-layer arrangement manner, that is, the channel information of different subbands in the same layer is arranged adjacently.

[0179] For example, the encoder may perform joint compression on the channel information of X layers based on a constellation scheme to obtain one compressed information (e.g., denoted as first compressed information), i.e., of the channel information of X layers, and the constellation scheme input by the encoder is the subband-before-layer constellation scheme. Correspondingly, the decoder may decode the first compressed information based on the constellation scheme, i.e., of the channel information of X layers, and the constellation scheme output by the decoder is the subband-before-layer constellation scheme.

[0180] 4 is another diagram of an arrangement scheme of channel information of each layer and between layers in X layers. As shown in FIG. 4, the X layers are three layers, respectively denoted as layer x, layer y, and layer z, and it is assumed that the channel information of the X layers corresponds to three subbands, respectively denoted as subband 1, subband 2, and subband 3. As shown in FIG. 4, the arrangement scheme output by the decoder is a subband-before-layer arrangement scheme. Specifically, first, the eigenvectors of different subbands in the same layer are arranged adjacently to form an eigenvector group of a layer, and then the eigenvector groups of different layers are arranged sequentially.

[0181] The above two arrangement schemes are examples for explanation, and the embodiments of the present application are not limited thereto. For example, the arrangement schemes of antenna ports in each layer and between layers can be used alternatively.

[0182] The encoder and decoder may know how to arrange the channel information for each layer and between layers in the X layers in the following manner:

[0183] Method 1: The encoder determines the arrangement method of channel information for each layer and between layers among X layers.

[0184] Based on this scheme, the encoder may determine an arrangement scheme of channel information for each layer and between layers in the X layers, and may further perform joint compression on the channel information of the X layers based on the arrangement scheme determined by the encoder to obtain one compressed information, and send the compressed information to the decoder.

[0185] Optionally, in Scheme 1, the encoder sends to the decoder the arrangement of channel information for each layer and between layers in the X layers. In this way, the decoder can obtain the channel information for the X layers by decoding based on the arrangement of channel information for each layer and between layers in the X layers and the received compressed information. Optionally, the encoder also sends to the decoder an index value indicating the arrangement of channel information for each layer and between layers in the X layers. For example, there is a correspondence between the index value and the arrangement of channel information for each layer and between layers in the X layers. Thus, the encoder sends the index value to the decoder, and the decoder may determine the arrangement scheme corresponding to the index value based on the correspondence between the index value and the arrangement of channel information for each layer and between layers in the X layers. For example, the layer-before-subband arrangement scheme corresponds to a first index value, and the subband-before-subband arrangement scheme corresponds to a second index value. In this case, if the placement scheme is the layer-before-subband placement scheme, the encoder sends the first index value to the decoder, and if the placement scheme is the subband-before-subband placement scheme, the encoder sends the second index value to the decoder.

[0186] Method 2: The decoder determines the arrangement method of channel information for each layer and between layers among the X layers.

[0187] Based on this scheme, the decoder can determine the arrangement scheme of channel information for each layer and between layers in the X layers, and further obtain the channel information of the X layers by decoding based on the arrangement scheme determined by the decoder and the received compressed information.

[0188] Optionally, in Scheme 2, the decoder sends the arrangement manner of the channel information of each layer and between layers in the X layers to the encoder. In this way, the encoder can perform joint compression on the channel information of the X layers based on the arrangement manner of the channel information of each layer and between layers in the X layers and indicated by the network device to obtain one compressed information, and send the compressed information to the decoder. Furthermore, optionally, the decoder sends an index value to the encoder, where the index value indicates the arrangement manner of the channel information of each layer and between layers in the X layers. For details, please refer to the description of Scheme 1. The details will not be described again here.

[0189] Method 3: Another device determines the arrangement method of channel information for each layer and between layers among the X layers.

[0190] Based on this scheme, another device may determine an arrangement scheme of channel information for each layer and between layers in the X number of layers. Based on this scheme, another device may determine an arrangement scheme of channel information for each layer and between layers in the X number of layers.

[0191] In Method 3, another device may send the arrangement manner of channel information for each layer and between layers in the X layers to the decoder and / or encoder. For example, the other device sends the arrangement manner of channel information for each layer and between layers in the X layers to the decoder and the encoder separately. In another example, the other device sends the arrangement manner of channel information for each layer and between layers in the X layers to the decoder. After receiving the arrangement manner of channel information for each layer and between layers in the X layers, the decoder sends the arrangement manner of channel information for each layer and between layers in the X layers to the encoder. In another example, the other device sends the arrangement manner of channel information for each layer and between layers in the X layers to the encoder. After receiving the arrangement manner of channel information for each layer and between layers in the X layers, the encoder sends the arrangement manner of channel information for each layer and between layers in the X layers to the decoder. Furthermore, optionally, another device may send an index value to the decoder and / or encoder, and the index value indicates the arrangement manner of the channel information of each layer and between layers in the X layers. For details, please refer to the description of Scheme 1. The details will not be described again here.

[0192] Method 4: Predefined For example, the arrangement method of channel information in each layer and between layers among X layers is predefined in the standard.

[0193] Based on this scheme, the encoder and decoder can determine, based on a predefinition, the arrangement manner of channel information for each layer and between layers in the X layers. In this way, the encoder and decoder can directly determine, based on a predefinition, the arrangement manner of channel information for each layer and between layers in the X layers.

[0194] Optionally, the encoder may determine an AI model based on the arrangement of channel information in each layer and between layers in the X layers.

[0195] The encoder may calculate indicators such as the accuracy of feeding back channel information in the joint compression scheme based on the arrangement of channel information for each layer and between layers, and determine or update the AI ​​model used for the joint compression.

[0196] For example, to calculate the accuracy of feeding back channel information in the joint compression scheme, an encoder may input channel information to the encoder to obtain compressed information. Then, the encoder inputs the compressed information to a reference decoder to obtain an output result, and parses the output result of the reference decoder based on the arrangement manner of the channel information of each layer and between layers in the X layers to recover the channel eigenvectors of the X layers, and compares the channel eigenvectors with the accurate channel eigenvectors (i.e., the channel eigenvectors corresponding to the channel information input by the encoder) to estimate the accuracy of feeding back channel information in the joint compression scheme. The reference decoder may represent a decoder known on the encoder side, and the function implemented by the reference decoder may be basically the same as that implemented by the decoder on the decoder side.

[0197] Optionally, the method 200 further includes: the encoder and the decoder know the sequence of channel information of different layers during the joint compression performed on some channel information of the M layers. In this way, the encoder can compress the jointly compressed channel information based on the sequence of channel information of different layers during the joint compression, and the decoder can correctly parse the decoder output based on the sequence of channel information of different layers during the joint compression.

[0198] For example, assuming M=3, the compression scheme of channel information of M layers is represented as [Layer 1, Layer 2] and [Layer 3], which indicates that joint compression is performed on the channel information of the first layer and the second layer in the first set, and individual compression is performed on the channel information of the third layer in the second set. For example, the encoder may perform joint compression on the channel information corresponding to the [Layer 1, Layer 2] set by using Model A in FIG. 5, and the encoder may perform individual compression on the channel information corresponding to the [Layer 3] set by using Model B in FIG.

[0199] 5 is a diagram of encoding and decoding for compressing M layers of channel information. As shown in FIG. 5, the AI ​​models deployed on the encoder side and the decoder side include two models, for example, Model A and Model B. When the encoder performs joint compression on the channel information of the first layer and the second layer by using Model A in FIG. 5, in order to correctly parse the decoder output, the correspondence between Layer x and Layer y and Layer 1 and Layer 2 in the output sequence of Model A of the decoder needs to be specified. In other words, the decoder needs to know the sequence of channel information output by Model A. For example, in the set [Layer 1, Layer 2], Layer 1 precedes Layer 2, and Layer 1 precedes Layer 2, indicating that Layer x corresponds to Layer 1 and Layer y corresponds to Layer 2.

[0200] Figure 6 is a diagram of the decoder output. For example, the decoder output of model A is shown in (1) in Figure 6, which is a layer-before-subband arrangement scheme. The sequence of channel information of different layers is such that the first layer 1 precedes layer 2, i.e., the eigenvectors of different layers in the same subband are arranged adjacently in a sequence in which layer 1 precedes layer 2 to form an eigenvector group of the subband, and then the eigenvector groups of different subbands are arranged sequentially. The decoder arrangement of model B is shown in (2) in Figure 6, i.e., the eigenvectors of different subbands of one layer are arranged adjacently.

[0201] The encoder and decoder may know the sequence of channel information of different layers during joint compression in several ways:

[0202] Method 1: The encoder determines the sequence of channel information of different layers during joint compression.

[0203] Based on this scheme, the encoder may determine a sequence of channel information of different layers during joint compression, and may further compress the jointly compressed channel information based on the sequence of the channel information of different layers during joint compression and determined by the encoder.

[0204] Optionally, in Scheme 1, the encoder sends a sequence of channel information of different layers during joint compression to the decoder. In this way, the decoder can correctly parse the decoder output based on the sequence of channel information of different layers during joint compression. Further, optionally, the encoder sends an index value to the decoder, where the index value indicates the sequence of channel information of different layers during joint compression. For example, there is a correspondence between the index value and the sequence of channel information of different layers during joint compression. Therefore, the encoder sends the index value to the decoder, and the decoder can determine the sequence corresponding to the index value based on the index value and the correspondence between the index value and the sequence of channel information of different layers during joint compression.

[0205] Method 2: The decoder determines the sequence of channel information of different layers during joint compression.

[0206] Based on this scheme, the decoder can determine the sequence of channel information of different layers during joint compression, and then correctly parse the decoder output based on the sequence of the channel information of different layers during joint compression and determined by the decoder.

[0207] Optionally, in Scheme 2, the decoder sends the sequence of channel information of different layers during joint compression to the encoder. In this way, the encoder can compress the jointly compressed channel information based on the sequence of the channel information of different layers during joint compression and indicated by the network device. Furthermore, optionally, the decoder sends an index value to the encoder, where the index value indicates the sequence of channel information of different layers during joint compression. For details, please refer to the description of Scheme 1. The details will not be described again here.

[0208] Method 3: Predefined. For example, the sequence of channel information of different layers during joint compression is predefined in the standard.

[0209] Based on this scheme, the encoder and decoder can determine the sequence of channel information of different layers during joint compression based on a predefinition. In this way, the encoder and decoder can directly determine the sequence of channel information of different layers during joint compression based on a predefinition.

[0210] The above-mentioned scheme is an example for explanation, and is not limited thereto. For example, another device may determine the sequence of channel information of different layers during joint compression, and send the sequence of channel information of different layers during joint compression to the encoder and / or decoder.

[0211] Furthermore, when the encoder notifies the decoder of at least two of the following information, namely, the compression method of the channel information of the M layers, the arrangement method of the channel information of each layer and among the X layers, and the sequence of the channel information of different layers during the joint compression, the above information may be carried by the same signaling or by different signaling. This is not limited. Similarly, when the decoder notifies the encoder of at least two of the following information, namely, the compression method of the channel information of the M layers, the arrangement method of the channel information of each layer and among the X layers, and the sequence of the channel information of different layers during the joint compression, the above information may be carried by the same signaling or by different signaling. This is not limited.

[0212] For ease of understanding, the following uses an example in which the encoder is a terminal device and the decoder is a network device to describe with reference to different procedures. For simplicity, the compression scheme of M layers of channel information is hereinafter referred to as M layer compression scheme for short, and the arrangement scheme of channel information of each layer and between layers in X layers is hereinafter referred to as arrangement scheme for short.

[0213] First, a possible procedure applicable to Solution 1 is described with reference to FIG.

[0214] 7 is a diagram of a communication method 700 according to an embodiment of the present application. As shown in FIG. 7, the method 700 is described by using an interaction between a terminal device and a network device as an example. The method 700 shown in FIG. 7 is applicable to a diagram of determining a compression scheme by a terminal device according to an embodiment of the present application. The method 700 shown in FIG. 7 may include the following steps:

[0215] Optionally, the method 700 includes step 701: a network device sends a deployment scheme to a terminal device.

[0216] In step 701, the network device may send a placement scheme to the terminal device. In this way, when the terminal device needs to perform joint compression on at least two layers of channel information, the terminal device may perform joint compression based on the placement scheme indicated by the network device.

[0217] It can be understood that an example in which a network device sends a disposition scheme to a terminal device is used for the description in this specification. As described in the embodiment of FIG. 2, the terminal device may alternatively determine the disposition scheme, or the disposition scheme may be predefined.

[0218] For the arrangement method, please refer to the description of the embodiment in Fig. 2. The details will not be described again here.

[0219] Optionally, the method 700 includes step 702. The terminal device sends information about at least one AI model to the network device.

[0220] The terminal device sends information about at least one AI model to the network device, so that the network device can process information from the terminal device by using an AI model corresponding to the at least one AI model. For example, the terminal device compresses channel information by using the at least one AI model and sends the compressed channel information to the network device. The network device can decode the compressed channel information by using an AI model corresponding to the at least one AI model.

[0221] For example, the terminal device may send at least one AI model to the network device, for example, in the form of an AI model list. In another example, the terminal device may send an identifier of at least one AI model to the network device, and a corresponding AI model may be learned based on the identifier. The at least one AI model may be all AI models deployed on the terminal device or a subset of AI models among all AI models of the terminal device. The subset of AI models may be AI models available on the terminal device. This is not limited to this.

[0222] Optionally, the terminal device monitors the performance of the AI ​​model deployed on the terminal device. For example, the terminal device monitors the performance of the AI ​​model deployed on the terminal device in a model monitoring phase (e.g., within a preset time period). For example, if the terminal device finds in the model monitoring phase that the performance of the AI ​​model has deteriorated to below a threshold, the terminal device may update the AI ​​model. Before the AI ​​model is updated, the AI ​​model becomes temporarily unavailable. Optionally, the terminal device may determine or update the AI ​​model based on the deployment scheme received in step 701. For details, please refer to the related description of method 200. The details will not be described again here.

[0223] In a possible implementation, the terminal device may periodically send at least one AI model to the network device. Alternatively, in another possible implementation, the terminal device may send at least one AI model to the network device when a preset condition is met. For example, the preset condition may be that the AI ​​model of the terminal device is updated, or that the terminal device receives a deployment method sent by the network device.

[0224] 703: The network device sends a downlink reference signal to the terminal device.

[0225] The downlink reference signal may be, for example, a channel status information reference signal (CSI-RS).

[0226] A terminal device may receive the downlink reference signals, and then the terminal device may perform channel estimation based on the received downlink reference signals to obtain corresponding channel eigenvectors.

[0227] The following briefly describes a calculation method of the channel eigenvector: It can be understood that the following method is only a possible implementation and does not limit the protection scope of the embodiments of the present application.

[0228] The terminal device performs channel estimation to obtain channel information. The terminal device processes the channel information, and the dimension of the channel information is [N tx ,N rx ,N RB ] where N tx represents the number of antennas or antenna ports at the transmitting end, and N rx represents the number of antennas or antenna ports at the receiving end, and N RB represents the number of frequency domain units, e.g., N RB represents the quantity of RB. [N tx ,N rx ,N RB The original channel of dimension ] can be divided into M [N tx ,N sb ]-dimensional eigensubspace, where M represents the number of layers (also called the number of streams or the number of ranks), and N sb represents the number of frequency domain units, e.g., N sb represents the number of frequency domain subbands. Typical numbers of frequency domain subbands are 1 RB, 2 RBs, or 4 RBs. For example, 4 RBs are used, and N sb =N RB / 4. N RB and N sb Both represent the quantity of frequency domain units, and N RB and N sb It can be understood that N may correspond to different values. RB represents the quantity of RB, and N sb represents the number of subbands. The processing in different layers is as follows:

[0229] The specific processing process in the Lth layer is as follows:

[0230] Each subband m contains T RBs, and the channels of the T RBs are combined to calculate the equivalent channel. The channel of the i-th RB is H i Assuming that , the equivalent channel in a subband can be expressed as:

[0231]

number

[0232] The SVD decomposition is

[0233]

number

[0234] is performed on and the following is obtained:

[0235]

number

[0236] In the above formula, each H i The dimension of is (N tx ,N rx ) and

[0237]

number

[0238] The dimension of (N tx ,N tx ), and the channel eigenvector value of the m-th subband in the L-th layer is

[0239]

number

[0240] is the L-th column of the subband, and the subband dimension is (N tx,1), that is, the eigenvector of the m-th subband in the L-th layer is

[0241]

number

[0242] where,

[0243]

number

[0244] is the value

[0245]

number

[0246] The colon ":" in this specification is only a possible notational method,

[0247]

number

[0248] It can be understood that any representation that can represent the L-th column of is applicable to this embodiment of the present application.

[0249] The channel eigenvectors corresponding to each layer can be obtained by performing the above operations.

[0250] Optionally, the method 700 includes step 704. The terminal device determines a number M of layers.

[0251] In this way, when determining the compression method for the M layers, the terminal device may further refer to the number M of layers.

[0252] After receiving the downlink reference signal, the terminal device may determine the number of layers based on the result of channel estimation. For example, the terminal device may determine the number of layers suitable for the terminal device based on the result of channel estimation and some communication parameters (e.g., signal-to-interference-and-noise ratio). The number of layers determined by the terminal device is M, where M is assumed to be an integer greater than 1.

[0253] 705: The terminal device determines the compression methods of the M layers.

[0254] The terminal device may determine the compression scheme of the M layers based on at least one of the capability information of the terminal device, the capability information of the network device, the performance of performing joint compression on the channel information of some layers in the channel information of the M layers, the performance of performing individual compression on the channel information of some layers in the channel information of the M layers, and the value of M. The capability information of the network device is sent by the network device to the terminal device, and may be sent to the terminal device together with the deployment scheme in step 701 (i.e., the capability information of the network device and the deployment scheme are carried in one signaling) or may be sent to the terminal device separately. Alternatively, the capability information of the network device may be estimated by the terminal device, for example, based on a historical communication status. For specific determination schemes, please refer to the relevant description of the above method 200. Details will not be described again here.

[0255] The compression schemes of the M layers may be represented by using a set. For example, assume M=3. For example, the compression schemes of the M layers may be [Layer 1], [Layer 2], and [Layer 3], i.e., separate compression is performed on the channel eigenvectors of different layers. In another example, the compression schemes of the M layers may be [Layer 1, Layer 2] and [Layer 3], i.e., joint compression is performed on the channel eigenvectors of the first layer and the second layer, and separate compression is performed on the channel eigenvectors of the third layer. In another example, the compression schemes of the M layers may be [Layer 1, Layer 2, Layer 3], i.e., joint compression is performed on the channel eigenvectors of the three layers.

[0256] 706: The terminal device sends the compression schemes of the M layers to the network device.

[0257] The terminal device feeds back the compression schemes of the M layers to the network device, so that the network device can select a corresponding AI model for decoding based on the compression schemes of the M layers.

[0258] Optionally, the terminal device further sends the value of M to the network device. For example, if method 700 includes step 704, the terminal device further sends the value of M to the network device. The compression schemes of the M layers and the value of M may be conveyed in the same signaling or in different signaling. This is not limited thereto.

[0259] Optionally, the terminal device further sends the sequence of channel information of different layers during joint compression to the network device. During joint compression, the sequence of channel information of different layers and one or more of the compression schemes of the M layers and the value of M can be conveyed in the same signaling or in different signaling. This is not limited.

[0260] 707: The terminal device compresses the M layers of channel information based on the M layers of compression schemes to obtain N pieces of compressed information.

[0261] For example, the compression schemes of the M layers determined by the terminal device are [Layer 1, Layer 2] and [Layer 3]. Therefore, the terminal device may perform joint compression on the channel eigenvectors of the first layer and the second layer by using Model A in Fig. 5, and the terminal device may perform individual compression on the channel eigenvectors of the third layer by using Model B in Fig. 5 to obtain two pieces of compressed information. One piece of compressed information corresponds to the joint compressed information of the channel eigenvectors of the first layer and the second layer, and the other piece of compressed information corresponds to the compressed information of the channel eigenvectors of the third layer.

[0262] Furthermore, the terminal device may perform compression based on the constellation scheme received in step 701 and the sequence of channel information of different layers. For example, if the constellation scheme is a layer-before-subband constellation scheme, the sequence of channel information of different layers is a sequence in which Layer 1 precedes Layer 2, i.e., eigenvectors of different layers in the same subband are adjacently arranged in a sequence in which Layer 1 precedes Layer 2. In this case, the constellation scheme of the channel eigenvectors that may be input to the AI ​​model by the terminal device and are of the first and second layers is a layer-before-subband constellation scheme, and eigenvectors of different layers in the same subband are adjacently arranged in a sequence in which Layer 1 precedes Layer 2. In another example, if the constellation scheme is a subband-before-layer constellation scheme, the sequence of channel information of different layers is a sequence in which Layer 1 precedes Layer 2, i.e., eigenvectors of different subbands in the same layer are adjacently arranged, and different layers are arranged in a sequence in which Layer 1 precedes Layer 2. In this case, the arrangement manner of the channel eigenvectors, which may be input to the AI ​​model by the terminal device and are of the first and second layers, is a subband-before-layer arrangement manner, and the different layers are arranged in a sequence in which Layer 1 precedes Layer 2. The sequence of the channel information of different layers may be indicated by the network device (e.g., indicated in step 701 or indicated separately), or may be determined or predefined by the terminal device. For details, please refer to the related description of method 200. The details will not be described again here.

[0263] 708: The terminal device sends the N compressed pieces of information to the network device.

[0264] 709: The network device decodes the N compressed information based on the M-layer compression scheme to obtain M-layer channel information.

[0265] The above example is still used as an example. Assuming that the compression schemes of the M layers are [Layer 1, Layer 2] and [Layer 3], the network device may know that the terminal device will send two pieces of compressed information, denoted as c1 and c2, respectively. Furthermore, the network device may know that c1 corresponds to the joint compressed information of [Layer 1, Layer 2] and therefore perform decoding by using a decoder for two layers (e.g., Model A in FIG. 5), and may know that c2 corresponds to the compressed information of [Layer 3] and therefore perform decoding by using a decoder for a single layer (e.g., Model B in FIG. 5). Furthermore, based on the sequence of channel information of different layers during joint compression and the arrangement manner of the outputs of the decoders of Model A and Model B, the network device may know that the channel information shown in FIG. 8 is obtained by inputting c1 into Model A, and the channel information shown in FIG. 9 is obtained by inputting c2 into Model B.

[0266] The above describes an example of a procedure in which a terminal device determines a compression scheme, with reference to steps 701 to 709 shown in FIG. 7. It should be understood that the above steps are merely illustrative examples and are not strictly limited. Furthermore, the sequence numbers of the above processes do not imply an execution sequence. The execution sequence of the processes should be determined based on the function and internal logic of the processes and does not constitute any limitation on the implementation process of the embodiments of the present application. For example, step 701 and step 706 may be performed simultaneously. Specifically, the terminal device sends the placement scheme, the compression schemes for the M layers, and the value of M to the network device. Furthermore, one or more of the placement scheme, the compression schemes for the M layers, and the value of M may be conveyed in the same signaling or in different signaling. This is not limited.

[0267] Further, for example, steps 701 and 702 may be considered a model deployment phase or a model monitoring phase, and steps 703 through 709 may be considered a model inference phase or a model running phase.

[0268] Based on the above technical solutions, the terminal device may determine compression schemes for the M layers and dynamically select an appropriate compression scheme based on, for example, the computing power of the terminal device, available AI models, etc. The terminal device may perform compression based on the arrangement scheme indicated by the network device, the compression scheme determined by the terminal device, and the sequence of channel information for different layers. The decoder performs decoding based on the arrangement scheme, the compression scheme, and the sequence of channel information for different layers, so that the AI ​​model used for decoding on the network device side can be matched to the AI ​​model used for compression on the terminal device side, and the network device side can correctly parse the output of the AI ​​model used for decoding.

[0269] The following describes a possible procedure applicable to Solution 2 with reference to FIG.

[0270] 10 is a diagram of a communication method 1000 according to an embodiment of the present application. As shown in FIG. 10, the method 1000 is described by using an interaction between a terminal device and a network device as an example. The method 1000 shown in FIG. 10 is applicable to a diagram of determining a compression scheme by a network device according to an embodiment of the present application. The method 1000 shown in FIG. 10 may include the following steps:

[0271] Optionally, the method 1000 includes step 1001. A network device sends a deployment scheme to a terminal device.

[0272] Optionally, the method 1000 includes step 1002. The terminal device sends information about at least one AI model to the network device.

[0273] 1003: The network device sends a downlink reference signal to the terminal device.

[0274] Optionally, the method 1000 includes step 1004. The terminal device determines a number M of layers.

[0275] Steps 1001 to 1004 are similar to steps 701 to 704 and will not be described again in detail here.

[0276] Optionally, the method 1000 includes step 1005. The terminal device sends the value of M to the network device.

[0277] 1006: The network device determines the compression methods for the M layers.

[0278] The network device may determine a compression scheme for the M layers based on at least one of the capability information of the terminal device, the capability information of the network device, a performance of performing joint compression on channel information of some layers in the channel information of the M layers, a performance of performing individual compression on channel information of some layers in the channel information of the M layers, and the value of M. The capability information of the terminal device may be sent by the terminal device to the network device, for example, together with the at least one AI model in step 1002 (i.e., the capability information of the terminal device and the at least one AI model are carried in one signaling), or together with the number of layers in step 1005 (i.e., the capability information of the terminal device and the number of layers are carried in one signaling), or separately sent to the network device. Alternatively, the capability information of the terminal device may be estimated by the network device, for example, based on a historical communication status or information about the at least one AI model sent by the terminal device. For the specific determination manner, please refer to the relevant description of the above method 200, and the details will not be described again here.

[0279] 1007: The network device sends the M layers of compression schemes to the terminal device.

[0280] The compression schemes of the M layers may be represented by using a set. For example, assume M=3. For example, the compression schemes of the M layers may be [Layer 1], [Layer 2], and [Layer 3], i.e., separate compression is performed on the channel eigenvectors of different layers. In another example, the compression schemes of the M layers may be [Layer 1, Layer 2] and [Layer 3], i.e., joint compression is performed on the channel eigenvectors of the first layer and the second layer, and separate compression is performed on the channel eigenvectors of the third layer. In another example, the compression schemes of the M layers may be [Layer 1, Layer 2, Layer 3], i.e., joint compression is performed on the channel eigenvectors of the three layers.

[0281] 1008: The terminal device compresses the M layers of channel information based on the M layers of compression schemes to obtain N pieces of compressed information.

[0282] The terminal device may compress the M layers of channel information based on the compression schemes for the M layers received in step 1007.

[0283] Step 1008 is similar to step 707 and will not be described again in detail here.

[0284] 1009: The terminal device sends the N compressed pieces of information to the network device.

[0285] 1010: The network device decodes the N compressed information based on the M-layer compression scheme to obtain M-layer channel information.

[0286] The network device may decode the N compressed information received in step 1009 based on the compression scheme of the M layers determined by the network device.

[0287] Step 1010 is similar to step 709 and will not be described again in detail here.

[0288] The above describes an example of a procedure in which a network device determines a compression scheme, with reference to steps 1001 to 1010 shown in FIG. 10 . It should be understood that the above steps are merely illustrative examples and are not strictly limited. Furthermore, the sequence numbers of the above processes do not imply an execution sequence. The execution sequence of the processes should be determined based on the function and internal logic of the processes and does not constitute any limitation on the implementation process of the embodiments of the present application. For example, step 1001 and step 1007 can be performed simultaneously, i.e., the network device sends the M-layer configuration scheme and compression scheme to the terminal device. Furthermore, the configuration scheme and the M-layer compression scheme can be conveyed in the same signaling or in different signaling. This is not limited.

[0289] Further, for example, steps 1001 and 1002 may be considered a model deployment phase or a model monitoring phase, and steps 1003 through 1010 may be considered a model inference phase or a model running phase.

[0290] Based on the above technical solution, the network device may determine compression schemes for the M layers, dynamically select an appropriate compression scheme based on, for example, the computing power of the network device, an available AI model, etc., and indicate the compression scheme to the terminal device. The terminal device may perform compression based on the arrangement scheme indicated by the network device, the compression scheme indicated by the network device, and the sequence of channel information for different layers. The decoder performs decoding based on the arrangement scheme, the compression scheme, and the sequence of channel information for different layers, so that the AI ​​model used for decoding on the network device side can be matched to the AI ​​model used for compression on the terminal device side, and the network device side can correctly parse the output of the AI ​​model used for decoding.

[0291] It can be understood that in the embodiments of Figures 7 and 10, the interaction between the terminal device and the network device is mainly used as an example for explanation. The present application is not limited thereto. The terminal device can be replaced with a receiving end device, and the receiving end device can be a terminal device or a network device. The network device can be replaced with a transmitting end device, and the transmitting end device can be a terminal device or a network device. For example, a "terminal device" can be replaced with a "first terminal device", and a "network device" can be replaced with a "second terminal device".

[0292] It may be further understood that in some of the above embodiments, one channel information or at least two channel information is referred to. One channel information represents one layer of channel information, i.e., one channel information can also be replaced with one layer of channel information. Similarly, at least two channel information represents at least two layers of channel information, i.e., at least two channel information can also be replaced with at least two layers of channel information.

[0293] It may be further understood that in some of the above embodiments, an example in which a terminal device provides information about an AI model for a network device is used for explanation. This is not limited. For example, the terminal device may provide a compression scheme to the network device, and the network device may determine a corresponding AI model based on the compression scheme. Similarly, the terminal device may provide information about an AI model to the network device, and the network device may also know the compression scheme based on the information about the AI ​​model and provided by the terminal device.

[0294] It may be further understood that some optional features in the embodiments of the present application may be independent of other features in some scenarios or may be combined with other features in some scenarios, without limitation.

[0295] It can be further understood that the solutions in the embodiments of the present application can be appropriately combined for use, and the explanations or descriptions of terms in the embodiments can be mutually referenced or explained in the embodiments. This is not limited.

[0296] In the above method embodiments, it may be further understood that the methods and operations implemented by the encoder may alternatively be implemented by components (e.g., chips or circuits) of the encoder. Furthermore, the methods and operations implemented by the decoder may alternatively be implemented by components (e.g., chips or circuits) of the decoder. This is not limited thereto.

[0297] The method provided in the embodiment of the present application has been described in detail above with reference to Figures 2 to 10. The following will describe in detail the data transmission device in the embodiment of the present application with reference to Figures 11 to 13. It should be understood that the description of the device embodiment corresponds to the description of the method embodiment. Therefore, for the contents not described in detail, please refer to the above method embodiment. For the sake of brevity, the details will not be described again here.

[0298] 11 is a block diagram of a communication device 1100 according to an embodiment of the present application. The device 1100 includes a transceiver unit 1110 and a processing unit 1120. The transceiver unit 1110 may be configured to implement corresponding communication functions. The transceiver unit 1110 may also be referred to as a communication interface or a communication unit. The processing unit 1120 may be configured to perform data processing.

[0299] Optionally, the apparatus 1100 may further include a storage unit. The storage unit may be configured to store instructions and / or data. The processing unit 1120 may read the instructions and / or data in the storage unit to enable the apparatus to implement the above method embodiments.

[0300] In design, apparatus 1100 is configured to perform steps or procedures performed by an encoder in the above method embodiments, e.g., steps or procedures performed by an encoder in the embodiment shown in Figure 2 and steps or procedures performed by a terminal device in the embodiment shown in Figure 7 or Figure 10. Transceiver unit 1110 is configured to perform transceiver-related operations at the encoder side in the above method embodiments, and processing unit 1120 is configured to perform processing-related operations at the encoder side in the above method embodiments.

[0301] In a possible implementation, the processing unit 1120 is configured to compress the M layers of channel information by using at least two artificial intelligence AI models to obtain N pieces of compressed information, where each of the N pieces of compressed information is obtained by compressing channel information of some layers in the M layers of channel information by using at least one of the at least two AI models, and N and M are integers greater than 1 and N is less than M. The transceiver unit 1110 is configured to send the N pieces of compressed information to a decoder.

[0302] For example, the N pieces of compressed information include at least one first compressed information and at least one second compressed information, where the first compressed information is obtained by performing joint compression on at least two layers of channel information among the M layers of channel information by using a first AI model among the at least two AI models, and the second compressed information is obtained by performing individual compression on one layer of channel information among the M layers of channel information by using a second AI model among the at least two AI models.

[0303] In another example, the compression scheme for the channel information of the M layers is determined based on at least one of the following information: the computing resources of the encoder, the computing resources of the decoder, the AI ​​model of the encoder, the AI ​​model of the decoder, the performance of performing joint compression on the channel information of some layers among the channel information of the M layers, the performance of performing individual compression on the channel information of some layers among the channel information of the M layers, and the value of M. The compression scheme for the channel information of the M layers includes joint compression, or the compression scheme for the channel information of the M layers includes joint compression and individual compression.

[0304] In another example, the processing unit 1120 is further configured to determine a compression scheme for the channel information of the M layers, where the compression scheme for the channel information of the M layers includes joint compression, or the compression scheme for the channel information of the M layers includes joint compression and individual compression.

[0305] In another example, the transceiver unit 1110 is further configured to send a compressed version of the channel information of the M layers to a decoder.

[0306] In another example, the transceiver unit 1110 is further configured to receive a compression scheme of the channel information of M layers, where the compression scheme of the channel information of the M layers includes joint compression, or the compression scheme of the channel information of the M layers includes joint compression and individual compression.

[0307] In another example, the transceiver unit 1110 is further configured to receive a reference signal from a decoder, and the processing unit 1120 is further configured to perform channel measurements based on the reference signal to obtain channel information of the M layers.

[0308] In another example, the processing unit 1120 is further configured to determine the value of M based on the results of the channel measurements.

[0309] In another example, the processing unit 1120 is specifically configured to: determine a first AI model based on an arrangement manner of channel information of each layer and between layers in the X layers, where at least two AI models include the first AI model, and the M layers include X layers, where X is an integer greater than 1 and less than M; perform joint compression on the channel information of the X layers by using the first AI model to obtain first compressed information, where the N pieces of compressed information include the first compressed information; and the arrangement manner of the channel information of each layer and between layers in the X layers includes any one of adjacent arrangement of channel information of different subbands in the same layer and adjacent arrangement of channel information of different layers in the same subband.

[0310] In another example, the transceiver unit 1110 is further configured to receive from the decoder an arrangement manner of channel information for and between each layer in the X layers, or to send to the decoder an arrangement manner of channel information for and between each layer in the X layers.

[0311] In another example, the transceiver unit 1110 is further configured to send first information to the decoder, where the first information indicates a sequence of channel information of different layers during joint compression performed on channel information of at least two layers among the channel information of the M layers.

[0312] In another example, the encoder is a terminal device and the decoder is a network device.

[0313] In another design, apparatus 1100 is configured to perform steps or procedures performed by a decoder in the above method embodiments, e.g., steps or procedures performed by a decoder in the embodiment shown in FIG. 2 and steps or procedures performed by a network device in the embodiment shown in FIG. 7 or 10. Transceiver unit 1110 is configured to perform transceiver-related operations at the decoder side in the above method embodiments, and processing unit 1120 is configured to perform processing-related operations at the decoder side in the above method embodiments.

[0314] In a possible implementation, the transceiver unit 1110 is configured to receive N pieces of compressed information from an encoder, where the N pieces of compressed information are obtained by compressing M layers of channel information by using at least two artificial intelligence AI models, and each of the N pieces of compressed information is obtained by compressing channel information of some layers in the M layers of channel information by using at least one of the at least two AI models, where N and M are integers greater than 1 and N is less than M. The processing unit 1120 is configured to decode the N pieces of compressed information to obtain the M layers of channel information.

[0315] For example, the N pieces of compressed information include at least one first compressed information and at least one second compressed information, where the first compressed information is obtained by performing joint compression on at least two layers of channel information among the M layers of channel information by using a first AI model among the at least two AI models, and the second compressed information is obtained by performing individual compression on one layer of channel information among the M layers of channel information by using a second AI model among the at least two AI models.

[0316] In another example, the compression scheme for the channel information of the M layers is determined based on at least one of the following information: the computing resources of the encoder, the computing resources of the decoder, the AI ​​model of the encoder, the AI ​​model of the decoder, the performance of performing joint compression on the channel information of some layers among the channel information of the M layers, the performance of performing individual compression on the channel information of some layers among the channel information of the M layers, and the value of M. The compression scheme for the channel information of the M layers includes joint compression, or the compression scheme for the channel information of the M layers includes joint compression and individual compression.

[0317] In another example, the processing unit 1120 is specifically configured to decode the N compressed information pieces based on a compression scheme of the channel information of the M layers, where the compression scheme of the channel information of the M layers includes joint compression, or the compression scheme of the channel information of the M layers includes joint compression and individual compression.

[0318] In another example, the processing unit 1120 is further configured to determine a compression scheme for the channel information of the M layers.

[0319] In another example, the transceiver unit 1110 is further configured to send a compression scheme of the channel information of the M layers to the encoder.

[0320] In another example, the transceiver unit 1110 is further configured to receive a compression scheme of the channel information for the M layers from the encoder.

[0321] In another example, the processing unit 1120 is specifically configured to decode the first compressed information based on an arrangement manner of channel information of each layer and between layers in the X layers to obtain channel information of X layers, where the first compressed information is obtained by performing joint compression on the channel information of the X layers by using a first AI model among the at least two AI models, the N compressed information includes the first compressed information, the M layers include the X layers, and X is an integer greater than 1 and less than M. The arrangement manner of channel information of each layer and between layers in the X layers includes any one of adjacent arrangement of channel information of different subbands in the same layer and adjacent arrangement of channel information of different layers in the same subband.

[0322] In another example, the transceiver unit 1110 is further configured to send to the encoder an arrangement manner of channel information for and between each layer in the X layers, or to receive from the encoder an arrangement manner of channel information for and between each layer in the X layers.

[0323] In another example, the transceiver unit 1110 is further configured to receive first information from the encoder, where the first information indicates a sequence of channel information of different layers during joint compression performed on channel information of at least two layers among the channel information of the M layers.

[0324] In another example, the encoder is a terminal device and the decoder is a network device.

[0325] It should be understood that the specific processes by which the units perform the above corresponding steps have been described in detail in the above method embodiments, and for the sake of brevity, the details will not be described again here.

[0326] It should be understood that the apparatus 1100 herein is presented in the form of a functional unit. The term "unit" herein may refer to an application-specific integrated circuit (ASIC), an electronic circuit, a processor (e.g., a shared processor, a dedicated processor, or a group processor) configured to execute one or more software or firmware programs, a memory, a merged logic circuit, and / or another suitable component supporting the described functionality. Those skilled in the art will understand that, in an optional example, the apparatus 1100 may specifically be the first terminal device in the above embodiment and may be configured to perform procedures and / or steps corresponding to the first terminal device in the above method embodiment. Alternatively, the apparatus 1100 may specifically be the second terminal device in the above embodiment and may be configured to perform procedures and / or steps corresponding to the second terminal device in the above method embodiment. To avoid repetition, details will not be described again here.

[0327] The apparatus 1100 in the above solution has a function of implementing a corresponding step performed by a first terminal device in the above method, or the apparatus 1100 in the above solution has a function of implementing a corresponding step performed by a second terminal device in the above method. The function may be implemented by hardware or by executing corresponding software. The hardware or software includes one or more modules corresponding to the above function. For example, a transceiver unit may be replaced by a transceiver (e.g., a sending unit in the transceiver unit may be replaced by a transmitter, and a receiving unit in the transceiver unit may be replaced by a receiver). Another unit, such as a processing unit, may be replaced by a processor to separately perform sending and receiving operations and related processing operations in each method embodiment.

[0328] Furthermore, the transceiver unit 1110 may alternatively be a transceiver circuit (eg, which may include a receiving circuit and a transmitting circuit), and the processing unit may be a processing circuit.

[0329] It should be noted that the apparatus in Fig. 11 may be a network element or device in the above embodiments, or may be a chip or a chip system, for example, a system-on-a-chip (SoC). The transceiver unit may be an input / output circuit and / or a communication interface. The processing unit may be an integrated processor, a microprocessor, or an integrated circuit on a chip. This is not limited here.

[0330] 12 is a diagram of another communication device 1200 according to an embodiment of the present application. The device 1200 includes a processor 1210. The processor 1210 is coupled to a memory 1220. The memory 1220 is configured to store computer programs or instructions and / or data. The processor 1210 is configured to execute computer programs or instructions stored in the memory 1220 or read data stored in the memory 1220 to implement the methods in the above method embodiments.

[0331] Optionally, there are one or more processors 1210 .

[0332] Optionally, one or more memories 1220 are present.

[0333] Optionally, memory 1220 is integrated with processor 1210 or disposed separately.

[0334] 12, the apparatus 1200 may further include a transceiver 1230. The transceiver 1230 is configured to receive and / or transmit signals. For example, the processor 1210 is configured to control the transceiver 1230 to receive and / or transmit signals.

[0335] For example, the processor 1210 may have the functionality of the processing unit 1120 shown in FIG. 11, the memory 1220 may have the functionality of a storage unit, and the transceiver 1230 may have the functionality of the transceiver unit 1110 shown in FIG. 11.

[0336] In the solution, the apparatus 1200 is configured to implement the operations performed by the encoder in the above method embodiments.

[0337] For example, the processor 1210 is configured to execute computer programs or instructions stored in the memory 1220 to implement relevant operations of the encoder in the above method embodiments, such as the method performed by the encoder in the embodiment shown in FIG. 2 or the method performed by the terminal device in the embodiment shown in either FIG. 7 or FIG. 10.

[0338] In another solution, the apparatus 1200 is configured to implement the operations performed by the decoder in the above method embodiments.

[0339] For example, the processor 1210 is configured to execute computer programs or instructions stored in the memory 1220 to implement relevant operations of the decoder in the method embodiments described above, such as the method performed by the decoder in the embodiment shown in FIG. 2 or the method performed by the network device in the embodiments shown in either FIG. 7 or FIG. 10.

[0340] It should be understood that the processor referred to in the embodiments of the present application may be a central processing unit (CPU), and may also be another general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc. The general-purpose processor may be a microprocessor, or the processor may be any conventional processor, etc.

[0341] It should be further understood that the memory referred to in the embodiments of the present application may be volatile memory and / or nonvolatile memory. The nonvolatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM). For example, RAM may be used as an external cache. By way of example and not limitation, RAM includes several forms such as static random access memory (static RAM, SRAM), dynamic random access memory (dynamic RAM, DRAM), synchronous dynamic random access memory (synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (double data rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhanced SDRAM, ESDRAM), synchlink dynamic random access memory (synchlink DRAM, SLDRAM), and direct rambus random access memory (direct rambus RAM, DR RAM).

[0342] It should be noted that when the processor is a general-purpose processor, a DSP, an ASIC, an FPGA or another programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, the memory (storage module) may be integrated into the processor.

[0343] It should be further noted that memory as described herein is intended to include, without being limited to, these and any other suitable types of memory.

[0344] 13 is a diagram of a chip system 1300 according to one embodiment of the present application. The chip system 1300 (or sometimes referred to as a processing system) includes a logic circuit 1310 and an input / output interface 1320.

[0345] The logic circuit 1310 may be a processing circuit in the chip system 1300. The logic circuit 1310 may be coupled to a storage unit and may retrieve instructions from the storage unit, so that the chip system 1300 can implement the methods and functions in the embodiments of the present application. The input / output interface 1320 may be an input / output circuit in the chip system 1300, which outputs information processed by the chip system 1300 or inputs data or signaling information to be processed into the chip system 1300 for processing.

[0346] Specifically, for example, if chip system 1300 is installed in an encoder and logic circuit 1310 is coupled to input / output interface 1320, logic circuit 1310 may send compressed information to a decoder via input / output interface 1320, and the compressed information may be obtained by logic circuit 1310 by compressing channel information. Alternatively, input / output interface 1320 may input a message from a second terminal device to logic circuit 1310 for processing. In another example, if chip system 1300 is installed in a decoder and logic circuit 1310 is coupled to input / output interface 1320, input / output interface 1320 may input compressed information from the encoder to logic circuit 1310 for processing.

[0347] In the solution, the chip system 1300 is configured to implement the operations performed by the encoder in the above method embodiments.

[0348] For example, logic circuitry 1310 is configured to implement processing-related operations performed by the encoder in the above method embodiments, e.g., the processing-related operations performed by the encoder in the embodiment shown in Figure 2, or the processing-related operations performed by the terminal device in the embodiment shown in either one of Figures 7 or 10. I / O interface 1320 is configured to implement sending- and / or receiving-related operations performed by the encoder in the above method embodiments, e.g., the sending- and / or receiving-related operations performed by the encoder in the embodiment shown in Figure 2, or the sending- and / or receiving-related operations performed by the terminal device in the embodiment shown in either one of Figures 7 or 10.

[0349] In another solution, the chip system 1300 is configured to implement the operations performed by the decoder in the above method embodiments.

[0350] For example, the logic circuitry 1310 is configured to implement the processing-related operations performed by the decoder in the method embodiments described above, e.g., the processing-related operations performed by the decoder in the embodiment shown in Figure 2, or the processing-related operations performed by the network device in the embodiments shown in either one of Figures 7 or 10. The input / output interface 1320 is configured to implement the sending- and / or receiving-related operations performed by the decoder in the method embodiments described above, e.g., the sending- and / or receiving-related operations performed by the decoder in the embodiment shown in Figure 2, or the sending- and / or receiving-related operations performed by the network device in the embodiments shown in either one of Figures 7 or 10.

[0351] An embodiment of the present application further provides a computer-readable storage medium, which stores computer instructions used to implement the method performed by the encoder or decoder in the above method embodiments.

[0352] For example, when the computer program is executed by a computer, it enables the computer to implement the methods performed by the encoder or decoder in the above method embodiments.

[0353] An embodiment of the present application further provides a computer program product including instructions, which, when executed by a computer, implement the method performed by the encoder or decoder in the above method embodiments.

[0354] An embodiment of the present application further provides a communication system. The communication system includes the encoder and decoder in the above embodiment. For example, the system includes the encoder and decoder in the embodiment shown in Figure 2. In another example, the system includes the terminal device and the network device in the embodiment shown in Figure 7 or Figure 10.

[0355] For the description of the relevant contents and beneficial effects of any one of the devices provided above, please refer to the corresponding method embodiments provided above, and the details will not be described again here.

[0356] In some embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods may be implemented in other manners. For example, the described device embodiments are merely examples. For example, the division into units is merely a logical functional division, and actual implementation may involve other divisions. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not implemented. Furthermore, the shown or described mutual couplings or direct couplings or communication connections may be implemented through some interfaces. Indirect couplings or communication connections between devices or units may be implemented in electrical, mechanical, or other forms.

[0357] All or part of the above embodiments may be implemented by using software, hardware, firmware, or any combination thereof. When software is used to implement an embodiment, all or part of the embodiment may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded into a computer and executed, the procedures or functions according to the embodiments of the present application are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or another programmable device. For example, the computer may be a personal computer, a server, or a network device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optics, or digital subscriber line (DSL)) or wireless (e.g., infrared, radio, or microwave) methods. The computer-readable storage medium may be any available medium accessible by a computer, or a data storage device such as a server or a data center integrated with one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, or a magnetic tape), an optical medium (e.g., a DVD), a semiconductor medium (e.g., a solid-state disk (SSD)), etc. For example, the available medium may include any medium capable of storing program code, such as, but not limited to, a USB flash drive, a removable hard disk drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0358] The above description is merely a specific implementation of the present application and does not limit the scope of protection of the present application. Any variations or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application shall fall within the scope of protection of the present application. Therefore, the scope of protection of the present application should be subject to the scope of protection of the claims.

Claims

1. 1. A communication method comprising: compressing, by an encoder, M layers of channel information by using at least two artificial intelligence (AI) models to obtain N pieces of compressed information, each of the N pieces of compressed information being obtained by compressing channel information of a portion of layers among the M layers of channel information by using at least one of the at least two AI models, where N and M are integers greater than 1 and N is less than M; sending, by the encoder, the N compressed pieces of information to a decoder; A communication method including:

2. the N compressed information pieces include at least one first compressed information piece and at least one second compressed information piece; 2. The method of claim 1, wherein the first compressed information is obtained by performing joint compression on at least two layers of channel information among the M layers of channel information by using a first AI model among the at least two AI models, and the second compressed information is obtained by performing individual compression on one layer of channel information among the M layers of channel information by using a second AI model among the at least two AI models.

3. The compression method of the channel information of the M layers is the following information: a computing resource of the encoder, a computing resource of the decoder, an AI model of the encoder, an AI model of the decoder, performance of performing joint compression on channel information of some layers among the channel information of the M layers, performance of performing individual compression on channel information of some layers among the channel information of the M layers, and a value of M; is determined based on at least one of The method of claim 1 or 2, wherein the compression scheme of the channel information of the M layers includes joint compression, or the compression scheme of the channel information of the M layers includes joint compression and individual compression.

4. The method comprises: determining, by the encoder, the compression scheme of the channel information of the M layers, wherein the compression scheme of the channel information of the M layers includes joint compression, or the compression scheme of the channel information of the M layers includes joint compression and individual compression; The method of any one of claims 1 to 3, further comprising:

5. The method comprises: sending, by the encoder, the compression scheme of the channel information of the M layers to the decoder; The method of claim 4 further comprising:

6. The method comprises: receiving, by the encoder, the compression scheme of the channel information of the M layers, wherein the compression scheme of the channel information of the M layers includes joint compression, or the compression scheme of the channel information of the M layers includes joint compression and individual compression; The method of any one of claims 1 to 3, further comprising:

7. The method comprises: receiving, by the encoder, a reference signal from the decoder; performing, by the encoder, channel measurements based on the reference signal to obtain the channel information for the M layers; The method of any one of claims 1 to 6, further comprising:

8. The method comprises: determining, by the encoder, the value of M based on the results of the channel measurements. The method of claim 7 further comprising:

9. The step of compressing, by an encoder, M layers of channel information by using at least two artificial intelligence (AI) models, comprises: determining, by the encoder, a first AI model based on an arrangement manner of channel information in each layer and between layers in the X layers, wherein the at least two AI models include the first AI model, and the M layers include the X layers, where X is an integer greater than 1 and less than M; performing, by the encoder, joint compression on the X layers of channel information by using the first AI model to obtain first compressed information, wherein the N pieces of compressed information include the first compressed information; Including, 9. The method according to claim 1, wherein the arrangement manner of the channel information in each layer and between the layers in the X layers includes any one of adjacent arrangement of channel information of different subbands in the same layer and adjacent arrangement of channel information of different layers in the same subband.

10. The method comprises: receiving, by the encoder, from the decoder, the arrangement of the channel information for each layer and between the layers in the X layers; or sending, by the encoder, the arrangement of the channel information in each layer and between the layers in the X number of layers to the decoder; The method of claim 9 further comprising:

11. The method comprises: sending, by the encoder, first information to the decoder, the first information indicating a sequence of channel information of different layers during joint compression performed on channel information of at least two layers among the channel information of the M layers; The method of any one of claims 1 to 10, further comprising:

12. 12. The method according to any one of claims 1 to 11, wherein the encoder is a terminal device and the decoder is a network device.

13. The method of any one of claims 1 to 12, wherein the channel information corresponding to each of the N compressed pieces of information is non-overlapping.

14. 1. A communication method comprising: receiving, by a decoder, N pieces of compressed information from an encoder, wherein the N pieces of compressed information are obtained by compressing channel information of M layers by using at least two artificial intelligence (AI) models, and each of the N pieces of compressed information is obtained by compressing channel information of a portion of layers among the channel information of the M layers by using at least one of the at least two AI models, where N and M are integers greater than 1 and N is less than M; decoding, by the decoder, the N compressed information to obtain the channel information of the M layers; A communication method including:

15. the N compressed information pieces include at least one first compressed information piece and at least one second compressed information piece; 15. The method of claim 14, wherein the first compressed information is obtained by performing joint compression on at least two layers of channel information among the M layers of channel information by using a first AI model among the at least two AI models, and the second compressed information is obtained by performing individual compression on one layer of channel information among the M layers of channel information by using a second AI model among the at least two AI models.

16. The compression method of the channel information of the M layers is the following information: a computing resource of the encoder, a computing resource of the decoder, an AI model of the encoder, an AI model of the decoder, performance of performing joint compression on channel information of some layers among the channel information of the M layers, performance of performing individual compression on channel information of some layers among the channel information of the M layers, and a value of M; is determined based on at least one of The method of claim 14 or 15, wherein the compression scheme of the channel information of the M layers comprises joint compression, or the compression scheme of the channel information of the M layers comprises joint compression and individual compression.

17. The step of decoding the N compressed information by the decoder comprises: decoding, by the decoder, the N pieces of compressed information based on the compression scheme of the channel information of the M layers, wherein the compression scheme of the channel information of the M layers includes joint compression, or the compression scheme of the channel information of the M layers includes joint compression and individual compression; 17. The method of any one of claims 14 to 16, comprising:

18. The method comprises: determining, by the decoder, the compression scheme of the channel information of the M layers; 18. The method of any one of claims 14 to 17, further comprising:

19. The method comprises: sending, by the decoder, the compression scheme of the channel information of the M layers to the encoder; 20. The method of claim 18 further comprising:

20. The method comprises: receiving, by the decoder, the compression scheme of the channel information of the M layers from the encoder; 20. The method of any one of claims 14 to 19, further comprising:

21. The step of decoding, by the decoder, the N compressed information to obtain the channel information of the M layers comprises: decoding, by the decoder, the first compressed information based on an arrangement manner of channel information of each layer and between layers in the X layers to obtain channel information of X layers, wherein the first compressed information is obtained by performing joint compression on the channel information of the X layers by using the first AI model among the at least two AI models, the N pieces of compressed information include the first compressed information, the M layers include the X layers, and X is an integer greater than 1 and less than M; 21. The method of any one of claims 14 to 20, comprising:

22. The method comprises: sending, by the decoder, the arrangement of the channel information in each layer and between the X layers to the encoder; or receiving, by the decoder, from the encoder, the arrangement of the channel information for each layer and between the layers in the X number of layers.

22. The method of claim 21 further comprising:

23. The method comprises: receiving, by the decoder, first information from the encoder, the first information indicating a sequence of channel information of different layers during joint compression performed on channel information of at least two layers among the channel information of the M layers; 23. The method of any one of claims 14 to 22, further comprising:

24. 24. The method of any one of claims 14 to 23, wherein the encoder is a terminal device and the decoder is a network device.

25. 25. The method of any one of claims 14 to 24, wherein the channel information corresponding to each of the N pieces of compressed information is non-overlapping.

26. A communication device comprising a module or unit adapted to carry out a method according to any one of claims 1 to 25.

27. 26. A communications device comprising a processor, the processor configured to execute computer programs or instructions stored in a memory to enable the device to perform a method according to any one of claims 1 to 25.

28. The apparatus further comprises the memory and / or a communication interface, the communication interface coupled to the processor; 28. The device of claim 27, wherein the communication interface is configured to input and / or output information.

29. A processor configured to carry out a method according to any one of claims 1 to 25.

30. 26. A computer-readable storage medium storing a computer program or instructions that, when run on a communications device, enable the communications device to perform a method according to any one of claims 1 to 25.

31. A computer program product, said computer program product comprising computer programs or instructions used to implement the method of any one of claims 1 to 25.

32. 26. A chip coupled to a memory and configured to read and execute program instructions stored in the memory to implement the method of any one of claims 1 to 25.

33. 26. A communication system comprising an encoding device and a decoding device, the encoding device being configured to implement a method according to any one of claims 1 to 13, and the decoding device being configured to implement a method according to any one of claims 14 to 25.

Images