Channel data generation method, device, equipment and storage medium

Abstract

A channel data generation method, device, equipment and storage medium belong to the technical field of wireless communication. The method comprises: generating virtual channel data (501) through a channel generation model; the virtual channel data is used to represent the channel condition in the channel environment; wherein the channel generation model is a machine learning model obtained by machine learning training of channel data samples. Through the above scheme, the acquisition efficiency of channel data in various channel environments is greatly improved, thereby improving the effect of channel modeling and improving the accuracy of wireless communication system research and design.

Description

Technical Field

[0001] This application relates to the field of wireless communication technology, and in particular to a channel data generation method, apparatus, device, and storage medium. Background Technology

[0002] In wireless communication systems, the channel environment is one of the main issues affecting the wireless transmission performance between communication devices.

[0003] In related technologies, in order to study the wireless transmission performance under various channel environments, it is usually necessary to obtain channel data under various channel environments in advance, and to model various channel environments based on the obtained channel data, thereby assisting in the research and design of wireless communication systems.

[0004] However, with the continuous development of wireless communication, the wireless channel environment is becoming more and more complex, and there are more and more factors to consider. It is difficult for technicians to accurately collect enough channel data in various channel environments. Summary of the Invention

[0005] This application provides a channel data generation method, apparatus, device, and storage medium. This solution can accurately and automatically generate massive amounts of channel data, thereby improving the effectiveness of channel modeling and enhancing the accuracy of wireless communication system research and design. The technical solution is as follows:

[0006] On one hand, embodiments of this application provide a channel data generation method, which is executed by a computer device, and the method includes:

[0007] Virtual channel data is generated using a channel generation model; the virtual channel data is used to characterize the channel conditions in the channel environment.

[0008] The channel generation model is a machine learning model obtained by training channel data samples.

[0009] On one hand, embodiments of this application provide a channel data processing method, which is executed by a computer device, and the method includes:

[0010] Acquire channel data samples; the channel data samples are used to characterize the channel conditions in the sample channel environment;

[0011] Using a channel generation model as the generator and a channel discrimination model as the discriminator, the channel generation model and the channel discrimination model are trained based on the channel data samples through generative adversarial learning.

[0012] The channel generation model, after being trained to convergence, is used to generate virtual channel data; the virtual channel data is used to characterize the channel conditions in the channel environment.

[0013] On the other hand, embodiments of this application provide a channel data generation apparatus, the apparatus comprising:

[0014] The generation module is used to generate virtual channel data through a channel generation model; the virtual channel data is used to characterize the channel conditions in the channel environment.

[0015] The channel generation model is a machine learning model obtained by training channel data samples.

[0016] On the other hand, embodiments of this application provide a channel data processing apparatus, the apparatus comprising:

[0017] The acquisition module is used to acquire channel data samples; the channel data samples are used to characterize the channel conditions in the sample channel environment.

[0018] The training module is used to train the channel generation model and the channel discrimination model based on the channel data samples through generative adversarial learning, using the channel generation model as the generator and the channel discrimination model as the discriminator.

[0019] The channel generation model, after being trained to convergence, is used to generate virtual channel data; the virtual channel data is used to characterize the channel conditions in the channel environment.

[0020] On the other hand, embodiments of this application provide a computer device, which is implemented as an information reporting device, and the computer device includes a processor, a memory, and a transceiver;

[0021] The memory stores a computer program, and the processor executes the computer program to enable the computer device to implement the above-described channel data generation method or channel data processing method.

[0022] In another aspect, embodiments of this application provide a computer device, which includes a processor, a memory, and a transceiver. The memory stores a computer program, which is executed by the processor to implement the aforementioned channel data generation method or channel data processing method.

[0023] In another aspect, embodiments of this application also provide a computer-readable storage medium storing a computer program, which is loaded and executed by a processor to implement the above-described channel data generation method or channel data processing method.

[0024] In another aspect, this application also provides a chip for operation in a computer device to enable the computer device to perform the above-described channel data generation method or channel data processing method.

[0025] In another aspect, this application provides a computer program product including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the aforementioned channel data generation method or channel data processing method.

[0026] In another aspect, this application provides a computer program that is executed by the processor of a computer device to implement the above-described channel data generation method or channel data processing method.

[0027] The technical solution provided in this application can bring the following beneficial effects:

[0028] A channel generation model is pre-trained using channel data samples. This model can automatically generate virtual channel data corresponding to the channel environment through simulation and prediction, without the need for actual data collection. This greatly improves the efficiency of acquiring channel data under various channel environments, thereby enhancing the effectiveness of channel modeling and improving the accuracy of wireless communication system research and design. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a schematic diagram of a network architecture provided in one embodiment of this application;

[0031] Figure 2 This is a schematic diagram of a communication system provided in one embodiment of this application;

[0032] Figure 3 This is a schematic diagram of a neural network provided in one embodiment of this application;

[0033] Figure 4 This is a schematic diagram of a neural network provided in another embodiment of this application;

[0034] Figure 5 This is a flowchart of a channel data generation method provided in one embodiment of this application;

[0035] Figure 6 This is a flowchart of a channel data processing method provided in one embodiment of this application;

[0036] Figure 7 This is a flowchart illustrating the model training and channel data generation process according to one embodiment of this application;

[0037] Figure 8 This is a flowchart of a channel data processing and channel data generation method provided in one embodiment of this application;

[0038] Figure 9 yes Figure 8 The illustrated embodiment is a schematic diagram of a virtual channel data structure.

[0039] Figure 10 yes Figure 8 A schematic diagram of another virtual channel data structure involved in the embodiment shown;

[0040] Figure 11 yes Figure 8 The illustrated embodiment relates to a model architecture diagram of a channel generation model and a channel identification model;

[0041] Figure 12 This is a block diagram of a channel data generation apparatus provided in one embodiment of this application;

[0042] Figure 13 This is a block diagram of a channel data processing apparatus provided in one embodiment of this application;

[0043] Figure 14 This is a schematic diagram of the structure of a computer device provided in one embodiment of this application. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0045] The network architecture and business scenarios described in the embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided in the embodiments of this application. As those skilled in the art will know, with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0046] Please refer to Figure 1 This diagram illustrates a network architecture of a wireless communication system according to an embodiment of this application. The network architecture may include: a terminal 10 and a base station 20.

[0047] The number of terminals 10 is typically multiple, with one or more terminals 10 distributed within the cell managed by each base station 20. Terminals 10 may include various handheld devices, vehicle-mounted devices, wearable devices, computing devices, or other processing devices connected to a wireless modem, as well as various forms of user equipment (UE), mobile station (MS), terminal device, etc. For ease of description, in this embodiment, the devices mentioned above are collectively referred to as terminals.

[0048] Base station 20 is a device deployed in an access network to provide wireless communication functionality to terminal 10. Base station 20 can include various forms of macro base stations, micro base stations, relay stations, access points, etc. In systems employing different wireless access technologies, the name of the device with base station functionality may differ; for example, in a 5th-Generation (5G) NR system, it is called gNodeB or gNB. As communication technologies evolve, the name "base station" may change. For ease of description, in this embodiment, the device providing wireless communication functionality to terminal 10 is collectively referred to as a base station.

[0049] Optional, Figure 1 As not shown, the above network architecture also includes other network devices, such as: Central Network Control (CNC), Access and Mobility Management Function (AMF) devices, Session Management Function (SMF) or User Plane Function (UPF) devices, etc.

[0050] The "5G NR system" in this disclosure can also be referred to as a 5G system or a New Radio (NR) system, but those skilled in the art will understand its meaning. The technical solutions described in this disclosure are applicable to 4G systems, 5G NR systems, and subsequent evolution systems of 5G NR systems.

[0051] To facilitate understanding, the following is an introduction to some relevant terms or background concepts involved in this application:

[0052] I. Wireless Communication

[0053] Please refer to Figure 2 This illustrates a schematic diagram of a wireless communication system provided in one embodiment of this application. Figure 2As shown, in a wireless communication system, the basic workflow is as follows: the transmitter encodes, modulates, and encrypts the source information at the transmitting end to form the information to be transmitted. The transmitted information is transmitted to the receiving end via wireless space, where the receiving end decodes, decrypts, and demodulates the received information to ultimately recover the source information.

[0054] In the above process, the encoding, modulation, encryption, decoding, demodulation, and decryption operations at the transmitting and receiving ends are controllable, but the channel environment in the space environment is uncontrollable, complex, and variable.

[0055] II. Artificial Intelligence (AI)

[0056] In recent years, artificial intelligence research, represented by neural networks, has achieved remarkable results in many fields, and it will play an important role in people's production and life for a long time to come.

[0057] Please refer to Figure 3 This illustrates a schematic diagram of a neural network provided in one embodiment of this application. Figure 3 As shown, the basic structure of a simple neural network includes an input layer, hidden layers, and an output layer. The input layer receives data, the hidden layers process the data, and the final result is generated in the output layer. Each node represents a processing unit, which can be considered as simulating a neuron. Multiple neurons form a layer of the neural network, and the information transmission and processing across multiple layers construct a complete neural network.

[0058] With the continuous development of neural network research, deep learning algorithms for neural networks have been proposed in recent years. More hidden layers have been introduced, and feature learning is carried out by training the neural network layer by layer through multiple hidden layers. This has greatly improved the learning and processing capabilities of the neural network and has been widely used in pattern recognition, signal processing, optimization and combination, anomaly detection and other fields.

[0059] Similarly, with the development of deep learning, convolutional neural networks have been further studied. Please refer to [link / reference]. Figure 4 This illustrates a neural network diagram provided in another embodiment of this application. Figure 4 As shown, its basic structure includes: an input layer, multiple convolutional layers, multiple pooling layers, a fully connected layer, and an output layer. The introduction of convolutional and pooling layers effectively controls the dramatic increase in network parameters, limits the number of parameters, and taps into the characteristics of local structures, thereby improving the robustness of the algorithm.

[0060] III. Channel Samples

[0061] Understanding and utilizing wireless channels is fundamental to building wireless communication systems. The most direct way to study basic wireless communication channels is to collect data from actual wireless channels. This can be done by using paired transmitters and receivers to obtain wireless channel information, or by using a specific receiver to collect signals from a third-party transmitter (such as a cellular network base station). These methods allow for the direct acquisition of the propagation characteristics of wireless channels, thus aiding in the design of wireless communication systems.

[0062] IV. Traditional Channel Modeling

[0063] For real-world wireless channel acquisition, considering the difficulties and costs involved, it is difficult to collect data for all scenarios and features. Based on real-world wireless channel acquisition, traditional wireless channel modeling work can extract relevant transmission features of a given channel from a limited number of wireless channel samples (i.e., channel data samples), such as large-scale parameters and small-scale parameters, including multipath information, delay power spectral density, transmission angle / angle of arrival, etc.

[0064] With the development of wireless communication systems, frequency bands are gradually moving towards higher frequencies, scenarios are gradually moving towards more complex special environments such as air, land, sea, and air, and application scope is expanding to more scenarios such as human-computer interaction, Internet of Things interaction, industrial applications, and special applications, making the wireless channel environment that current wireless communication systems need to deal with increasingly complex.

[0065] Under the aforementioned circumstances, actual sampling of wireless channels becomes extremely difficult, presenting both technical and operational challenges. Simultaneously, mathematical modeling of these complex channels also faces significant challenges. The complexity of frequency bands, environments, and scenarios directly leads to the complexity of channel modeling. Nonlinear channel characteristics and difficult-to-fit channel propagation properties pose difficulties and challenges to traditional mathematical modeling methods for studying channels. For example, high-frequency channel modeling remains a problem that urgently needs to be solved. Furthermore, the discrepancy between actual channel environment modeling and ideal channel environment modeling in complex scenarios and application environments will continue to increase dramatically in future wireless communication research as the complexity of the channel environment increases.

[0066] Therefore, it is evident that obtaining channel characteristic information through channel sampling and traditional mathematical modeling is a significant challenge in complex frequency bands, environments, and scenarios.

[0067] At the same time, artificial intelligence is increasingly being used in wireless communication systems. Currently, a large amount of research is based on the combination of artificial intelligence and wireless communication. This work is highly dependent on and requires the wireless channel itself and the data associated with the wireless channel. It can be said that wireless channel data is the key to determining the performance gain of the combination of artificial intelligence and wireless communication.

[0068] Under these circumstances, when a large amount of wireless channel data is needed as the dataset required for AI-based wireless communication solutions, on the one hand, traditional methods of channel acquisition and channel modeling have significant problems in terms of feasibility and reliability; on the other hand, implementation cost is also an issue that needs to be addressed. AI solutions are highly dependent on model training datasets. When channel data is needed as a model training set, it often requires thousands, tens of thousands, hundreds of thousands, or even larger amounts of channel data, and the cost of acquiring such a large dataset is extremely high.

[0069] In summary, how to acquire and construct effective channel datasets to support research on the integration of artificial intelligence and wireless communication systems is a critical issue that urgently needs to be addressed.

[0070] Please refer to Figure 5 The diagram illustrates a flowchart of a channel data generation method provided in one embodiment of this application. This method can be executed by a computer device and may include the following steps:

[0071] Step 501: Generate virtual channel data through the channel generation model. The virtual channel data is used to characterize the channel conditions in the channel environment. The channel generation model is a machine learning model obtained by machine learning training through channel data samples.

[0072] The channel environment represented by the aforementioned virtual channel data can be a simulated channel environment. In other words, the channel generation model can generate channel data corresponding to various different channel environments through simulation and prediction. In this process, it is not necessary to collect channel data from the actual channel environment, so that the acquisition of channel data does not depend on the actual channel environment.

[0073] The virtual channel data generated by the channel generation model in this embodiment can be used to construct a channel environment for the research and design of wireless communication systems. Alternatively, it can be used as sample data for machine learning models in wireless communication system design to improve the performance gain of combining artificial intelligence and wireless communication.

[0074] In summary, the solution shown in the embodiments of this application can pre-train a channel generation model using channel data samples. Through this channel generation model, virtual channel data corresponding to the channel environment can be automatically generated through simulation and prediction. Channel data corresponding to various channel environments can be quickly obtained without actual collection, thereby greatly improving the efficiency of acquiring channel data under various channel environments, thus improving the effect of channel modeling and the accuracy of wireless communication system research and design.

[0075] The aforementioned channel generation model can be trained using generative adversarial learning, thereby enabling the virtual channel data generated by the channel generation model to accurately simulate and predict channel conditions in various channel environments.

[0076] Please refer to Figure 6 The diagram illustrates a flowchart of a channel data processing method provided in an embodiment of this application, which can be executed by a computer device; the method may include the following steps:

[0077] Step 601: Obtain channel data samples. Channel data samples are used to characterize the channel conditions in the sample channel environment.

[0078] The aforementioned channel data samples can be samples obtained by collecting channel data in an actual channel environment; or, channel data samples can be samples manually constructed by technicians based on an actual channel environment; or, the aforementioned channel data samples can be samples automatically constructed by technicians using other channel data construction tools.

[0079] Step 602: Using the channel generation model as the generator and the channel discrimination model as the discriminator, the channel generation model and the channel discrimination model are trained based on channel data samples through generative adversarial learning. The channel generation model, after being trained to convergence, is used to generate virtual channel data, which is used to characterize the channel conditions in the channel environment.

[0080] In this embodiment of the application, during the training phase of the channel generation model, two machine learning models can be set in the computer device. One machine learning model A (corresponding to the channel generation model above) is responsible for generating channel data, and the other machine learning model B (corresponding to the channel discrimination model above) is responsible for determining whether the input channel data is true (or, determining whether the input channel data is naturally existing channel data or channel data generated by the machine). The channel data samples are used as training samples, and the two machine learning models are trained through generative adversarial learning until both machine learning models are trained to convergence.

[0081] In this context, due to the existence of channel data samples, the trained machine learning model B has a certain ability to judge whether the input channel data samples are real. When the accuracy of machine learning model B is high enough (i.e. convergence), if the channel data generated by machine learning model A cannot be accurately judged by machine learning model B, it is considered that the channel data generated by machine learning model A is close enough to the real channel data, and machine learning model A has also reached the convergence state. At this time, machine learning model A can be used as the converged channel generation model for the generation of subsequent channel data.

[0082] In summary, the scheme shown in this application pre-trains a channel generation model based on channel data samples using generative adversarial learning. This channel generation model can automatically generate virtual channel data corresponding to the channel environment through simulation and prediction. It can quickly obtain channel data corresponding to various channel environments without actual collection, thereby greatly improving the efficiency of acquiring channel data under various channel environments, thus improving the effect of channel modeling and the accuracy of wireless communication system research and design.

[0083] As stated above in this application Figure 5 and Figure 6 As shown, the proposed solution includes a model training phase and a model application phase. Please refer to... Figure 7 This illustrates a flowchart of the model training and channel data generation process provided in one embodiment of this application. Figure 7 As shown, the model training phase and the model application phase described above can be executed by the model training device and the channel data generation device, respectively. Figure 7 As shown, the process includes the following steps:

[0084] Step 1: In the model training phase, the model training device 71 acquires channel data sample 71a, an initialized machine learning model A, and an initialized machine learning module B.

[0085] Specifically, the format of the data output by the initialized machine learning model A and the format of the data input by the initialized machine learning model B can be matched with the data format of the channel data. For example, the format of the data output by the initialized machine learning model A or the format of the data input by the initialized machine learning model B can be the same as the data format of the channel data; or, the format of the data output by the initialized machine learning model A or the format of the data input by the initialized machine learning model B can be converted into the data format of the channel data through a pre-designed conversion method.

[0086] The format of the data output by the initialized machine learning model A and the format of the data input to the initialized machine learning model B can be pre-designed by the developers.

[0087] Step 2: The model training device generates predicted virtual channel data 71b using machine learning model A.

[0088] In this embodiment of the application, the model training device can output data that meets the data format of channel data through machine learning model A as the predicted virtual channel data.

[0089] Step 3: The model training device uses channel data sample 71a and predicted virtual channel data 71b as positive and negative samples to train machine learning model A and machine learning model B in an adversarial learning manner.

[0090] The predicted virtual channel data can be used as a negative sample, and its corresponding training label is the first label. The first label can indicate that the predicted virtual channel data is non-natural channel data (or channel data generated through simulation prediction).

[0091] Accordingly, the aforementioned channel data samples can be used as positive samples, and their corresponding training labels are second labels, which can indicate that the channel data samples are naturally existing channel data.

[0092] During training, the model training device can train machine learning model A and machine learning model B alternately. In the adversarial learning process, the accuracy of the predicted virtual channel data output by machine learning model A is not high at the beginning of training, and the accuracy of machine learning model B in judging whether the input channel data is naturally existing channel data is also not high. As adversarial learning progresses, the accuracy of machine learning model B's judgment increases, and correspondingly, the predicted virtual channel data generated by machine learning model A becomes closer and closer to naturally existing channel data. When both machine learning models approach convergence, machine learning model B can accurately determine that the channel data sample is naturally existing channel data, but it cannot accurately distinguish whether the predicted virtual channel data is naturally existing channel data. At this point, it can be considered that the predicted virtual channel data generated by machine learning model A is sufficiently close to naturally existing channel data.

[0093] Step 4: After both machine learning model A and machine learning model B converge, the model training device outputs machine learning model A as channel generation model 72; this channel generation model can be deployed to channel data generation device 73.

[0094] Step 5, in the model application stage channel, the channel data generation device 73 generates virtual channel data 72a through the channel generation model 72.

[0095] The aforementioned model training device and channel data generation device can be implemented as the same physical device, such as the same personal computer, workstation, or server.

[0096] Alternatively, the model training device and the channel data generation device described above can be implemented as different physical devices. For example, the model training device can be implemented as a personal computer, workstation, or server used by developers, and the data generation device can be a personal computer, workstation, or server used by designers of wireless communication systems.

[0097] Please refer to Figure 8 This document illustrates a flowchart of a channel data processing and channel data generation method according to an embodiment of this application. The method can be executed by a computer device, for example, it can be interactively executed by a model training device and a channel data generation device; the method may include the following steps:

[0098] Step 801: During the model training phase, the model training device acquires channel data samples; the channel data samples are used to characterize the channel conditions in the sample channel environment.

[0099] In this embodiment of the application, during the model training phase, the developer can collect several channel data samples in advance and input the collected channel data samples into the model training device.

[0100] The aforementioned channel data samples can be channel data collected in actual channel environments, or channel data constructed manually or by machine and considered to exist naturally.

[0101] After obtaining channel data samples and the channel generation and discrimination models pre-built and initialized by the developers (e.g., by randomly setting parameters), the channel generation model can be used as the generator, and the channel discrimination model as the discriminator. Based on the channel data samples, the channel generation and discrimination models can be trained using generative adversarial learning. This training process can be referred to steps 802 to 807 of the harness.

[0102] Step 802: During the training phase of the channel identification model, the model training device generates predicted virtual channel data through the channel generation model.

[0103] In one possible implementation, the channel generation model includes at least one of the following four types of networks: fully connected network, convolutional neural network, residual network, or self-attention mechanism network.

[0104] In one possible implementation, the model training device can input the input information into the channel generation model to obtain the predicted virtual channel data output by the channel generation model after processing the input information samples.

[0105] The channel generation model can have an input port. During model training and application, the channel generation model can process the input information layer by layer and finally output data that meets a certain data format as virtual channel data.

[0106] In one possible implementation, the input information includes at least one of the following four types of information:

[0107] Noise information, random number information, channel type indication information, or channel data sample information;

[0108] Among them, the channel type indication information is used to indicate the channel type;

[0109] Channel data sample information is information constructed based on channel data samples.

[0110] In one exemplary embodiment of this application, the input to the channel generation model can be any input, such as any noise or any random number. The channel generation model can be triggered by any input, and can then perform subsequent layer-by-layer processing to finally output virtual channel data. The virtual channel data output during the model training phase is the aforementioned predicted virtual channel data.

[0111] The noise information mentioned above can come from the real environment or be generated artificially.

[0112] The random number information mentioned above can be a random number sequence or a pseudo-random number sequence.

[0113] The noise or random number information can be in the format of a one-dimensional vector, a two-dimensional matrix, or a high-dimensional set of noise or random numbers. The format of the noise and random number information can be agreed upon in advance or be consistent with the format of the virtual channel data to be generated.

[0114] In another exemplary embodiment of this application, the input to the channel generation model can also be information with a specified meaning. For example, it can be information indicating a certain channel type (the purpose is to make the output virtual channel data simulate the virtual channel of the corresponding channel type), or it can be information derived from the channel data sample (the purpose is to make the channel environment of the virtual channel corresponding to the output virtual channel data similar to the channel environment corresponding to the channel data sample, or in other words, to make the channel environment of the virtual channel corresponding to the output virtual channel data improved based on the channel environment corresponding to the channel data sample).

[0115] In one possible implementation, during the model training phase, the aforementioned channel type indication information is used to indicate the channel type corresponding to the channel data sample.

[0116] In the embodiments of this application, during the model training phase, the channel type indicated by the aforementioned channel type indication information can be consistent with the channel type corresponding to the channel data sample. In this way, even when random numbers or random noise are mixed in, the channel generation model can generate multiple virtual channel data that match the input channel type indication information during the training process and subsequent applications.

[0117] In one possible implementation, the channel type indication information includes at least one of the following five types of information:

[0118] Temporal feature information, frequency feature information, spatial feature information, environmental feature information, or scene feature information.

[0119] In this embodiment of the application, the channel type indication information can indicate the frequency information, environmental information, and scene information corresponding to the channel, such as: high frequency, low frequency, indoor, outdoor, dense community, open field, Internet of Things scene, industrial scene, etc.

[0120] Among them, the aforementioned time-domain characteristic information, frequency characteristic information, and spatial domain characteristic information can be referred to as the channel's index characteristic information, such as time delay power spectrum information, multipath information, angle information, velocity information, etc.

[0121] The above environmental characteristics information can indicate indoor environment, outdoor environment, open field, etc.

[0122] The aforementioned scene feature information can indicate scene categories such as line-of-sight (LOS), not-line-of-sight (NLOS), high-speed, and low-speed.

[0123] In one possible implementation, the channel data sample information includes at least one of the following three types of information:

[0124] Information obtained by mixing noise into channel data samples;

[0125] Information obtained by mixing random numbers into channel data samples;

[0126] Alternatively, information obtained by mixing noise and random numbers in channel data samples.

[0127] For example, model training devices can mix noise and channel data samples as input to the channel generation model, or mix random numbers and channel data samples as input to the channel generation model, and so on.

[0128] In one possible implementation, the virtual channel data includes channel data corresponding to at least one granularity in at least one dimension.

[0129] Wherein, when the virtual channel data includes channel data corresponding to each granularity in at least two dimensions, the virtual channel data includes a matrix in at least two dimensions; or, when the virtual channel data includes channel data corresponding to each granularity in at least two dimensions, the virtual channel data includes one-dimensional data obtained by arranging the channel data corresponding to each granularity in at least two dimensions.

[0130] Taking virtual channel data, which includes channel data corresponding to various granularities in two dimensions, as an example, a single sample of the aforementioned virtual channel data can be composed of a matrix of size M*N. This matrix has M first granularities in the first dimension and N second granularities in the second dimension. M and N can be equal or unequal, and the specific values ​​within the matrix indicate the channel quality. Alternatively, the two-dimensional M*N data can be combined into a one-dimensional dataset of size 1*(M*N) or (M*N)*1. The transformation can be either first the first dimension followed by the second dimension, or vice versa; this difference lies in the format of the representation.

[0131] In one possible implementation, at least one dimension includes at least one of the following four dimensions:

[0132] Frequency domain dimension, time domain dimension, spatial domain dimension, or real / imaginary part dimension.

[0133] In one possible implementation, when at least one dimension includes a frequency domain dimension, a granularity on the frequency domain dimension includes:

[0134] At least one radio bearer (RB) or at least one subcarrier.

[0135] For example, a single sample of virtual channel data is composed of a first dimension with granularity m. The first dimension can be a frequency domain dimension. When the first dimension is a frequency domain dimension, the granularity m can be a RBs (a is greater than or equal to 1, such as 2 RBs, 4 RBs, 8 RBs), or it can be b subcarriers (b is greater than 1, such as 4 subcarriers, 6 subcarriers, 18 subcarriers). When the first dimension is a frequency domain dimension, the frequency domain range indicated by a single sample of virtual channel data is an M*m frequency domain range.

[0136] In one possible implementation, when at least one dimension includes a time-domain dimension, a granularity on the time-domain dimension includes:

[0137] p1 is a microsecond, the length of at least one symbol, or the number of sampling points for at least one symbol; where p1 is a positive number.

[0138] For example, a single sample of virtual channel data can also be composed of a first dimension with granularity p. This first dimension can be a time-domain dimension. When the first dimension is a time-domain dimension, the granularity p can be a delay granularity, such as a delay granularity of p1 microseconds, p2 symbol lengths, or p3 symbol sampling points. Here, a symbol can be an Orthogonal Frequency Division Multiplexing (OFDM) symbol. When the first dimension is a time-domain dimension, the time-domain range (or delay range) indicated by a single sample in the training set is an M*p time-domain range.

[0139] In one possible implementation, when at least one dimension includes a spatial domain dimension, a granularity on the spatial domain dimension includes:

[0140] At least one pair of transceiver antennas, at least one receiving antenna, at least one transmitting antenna, or an angular range of target angular intervals.

[0141] For example, a single sample of virtual channel data consists of a second dimension with a granularity of n. The second dimension can be a spatial domain dimension, specifically an antenna dimension. For instance, the second dimension consists of N antenna pairs, and the second granularity is a pair of transmit and receive antennas.

[0142] For example, a single sample of virtual channel data can also be composed of a second dimension with a granularity of q. The second dimension can be a spatial domain dimension, specifically an angular domain dimension. For example, the second dimension can be composed of N angles, and the second granularity is the size of the angular interval between the aforementioned N angles.

[0143] In a single sample of virtual channel data, the channel quality indication on a specific combination of dimensions can represent the channel quality indication under that specific combination of dimensions.

[0144] For example, please refer to Figure 9 The diagram illustrates a virtual channel data structure according to an embodiment of this application. In an M*N matrix, the indicator value X in the 3rd row and 6th column can be used to represent the channel quality at the third specific granularity bandwidth at the sixth spatial granularity.

[0145] For example, please refer to Figure 10 This diagram illustrates another virtual channel data structure according to an embodiment of this application. In an M*N matrix, the indicator value Y in the 4th row and 5th column can be used to represent the channel quality of the 4th specific granularity delay at the 5th spatial granularity (e.g., angle of arrival).

[0146] Since both the virtual channel and the channel feature information obtained from the virtual channel can be represented by complex numbers, the output of the channel generation model described above can have an additional dimension on top of the above description. This dimension is caused by independently representing the imaginary and real parts of the virtual channel (or the channel feature information obtained from the virtual channel). For example, in addition to the first and second dimensions, there can be a third dimension, which comes from the real and imaginary parts of the channel.

[0147] Furthermore, it should be noted that the output of the aforementioned channel generation model can also be split and combined based on the first, second, and third dimensions. For example, when the second dimension is an antenna pair dimension, it can be split into a transmit antenna sub-dimension and a receive antenna sub-dimension, thereby expanding the dimensions of the aforementioned virtual channel output form.

[0148] In the above description, for the sake of simplicity, a two-dimensional virtual channel consisting of the first and second dimensions is used as an example. The dimensions of the virtual channel involved in the embodiments of this application are not limited to two dimensions.

[0149] In one possible implementation, the virtual channel data includes at least one of the following two types of information: the original channel information, or the channel feature vector;

[0150] Channel feature vectors are obtained by transforming the original channel information.

[0151] In one possible implementation, the original channel information includes channel quality information.

[0152] In one possible implementation, the channel feature vector is obtained by performing Singular Value Decomposition (SVD) on the original channel information.

[0153] The output information of the aforementioned channel generation model can also be channel feature information obtained by mathematical transformation of the original channel information, such as channel feature vector information obtained by SVD decomposition. This can be single-stream channel feature vector information or multi-stream channel feature vector information, such as 2-stream, 4-stream, and 8-stream channel feature vector information.

[0154] Step 803: The model training device inputs the predicted virtual channel data and channel data samples into the channel identification model to obtain the first identification result of the channel identification model; the first identification result is used to indicate whether the predicted virtual channel data and channel data samples are channel data generated by the model.

[0155] In one possible implementation, the channel discrimination model described above includes one or more of the following four types of networks: fully connected networks, convolutional neural networks, residual networks, or self-attention mechanism networks.

[0156] Please refer to Figure 11 It shows a model architecture diagram of a channel generation model and a channel identification model involved in the embodiments of this application.

[0157] Step 804: The model training device updates the model parameters of the channel identification model based on the first identification result.

[0158] In the embodiments of this application, during the training stage of the channel discrimination model, the model training device can calculate the loss function value based on the first discrimination result, the predicted virtual channel data, and the respective labels of the channel data samples, and then update the model parameters of the channel discrimination model based on the loss function value.

[0159] Step 805: During the training phase of the channel generation model, the model training device generates predicted virtual channel data through the channel generation model.

[0160] Step 806: The model training device inputs the predicted virtual channel data into the channel identification model to obtain the second identification result of the channel identification model; the second identification result is used to indicate whether the predicted virtual channel data is channel data generated by the model.

[0161] Step 807: The model training device updates the model parameters of the channel generation model based on the second identification result.

[0162] In the embodiments of this application, during the training stage of the channel generation model, the model training device can calculate the loss function value using the second identification result and the label of the predicted virtual channel data, and then update the model parameters of the channel generation model using the loss function value.

[0163] Iteratively execute steps 802 to 807 above until both models converge.

[0164] Once the generator and discriminator reach a stable state through neural network training, the generator can be extracted separately for generating virtual channel data.

[0165] The construction of the channel generation model requires the simultaneous construction of two parts: the channel generation model and the channel identification model. The channel generation model is used to generate virtual channel data, while the channel identification model is used to determine the difference between the virtual channel and the real channel.

[0166] The model training device utilizes Generative Adversarial Networks (GANs) as the foundational structure for both the channel generation and channel discrimination models. Input information (e.g., random numbers) serves as the input to the channel generation model, and the given neural network structure forms the basic framework for generating the current virtual channel input. The channel discrimination model then determines the difference between the virtual and real channels. If a difference is identified, the process continues, updating the channel generation model parameters and generating new virtual channel outputs. This process continues until the channel discrimination model can no longer distinguish between the virtual and real channels. At this point, the channel generation model is considered complete, capable of generating a virtual channel that closely approximates the real channel.

[0167] Based on the construction of the channel generation model described above, only a small amount of real channel information can be used to assist in the construction of the channel identification model, and finally a large amount of virtual channel data can be generated based on the channel generation model.

[0168] This solution addresses the challenges of obtaining large datasets and effectively fitting complex nonlinear channel models in complex frequencies, scenarios, and environments using traditional channel modeling and estimation methods. By employing a design that utilizes only a small amount of real-world data, a channel generation model can be constructed. This model, in turn, allows for the creation of a large amount of virtual channel data, significantly reducing the difficulty and cost of manual data collection and avoiding the ineffectiveness issues of traditional data modeling in complex channels. This virtual channel data can then be used in AI-based wireless communication solutions to rapidly build datasets across multiple frequency bands, scenarios, and environments, supporting on-demand model retraining and updates.

[0169] Step 808: In the model application stage channel, the channel data generation device generates virtual channel data through the channel generation model.

[0170] In one possible implementation, virtual channel data is generated through a channel generation model, including:

[0171] Input information is fed into the channel generation model to obtain virtual channel data output by the channel generation model after processing the input information.

[0172] The inputs and outputs of the channel generation model described above can be found in the description under step 802 above, and will not be repeated here.

[0173] In this embodiment of the application, the channel generation model in the channel data generation device can be encapsulated in the channel data generator.

[0174] In one possible implementation, the aforementioned channel data generator may be provided with an input interface, the input of which is the input information of the aforementioned channel generation model, which will not be elaborated here.

[0175] In another possible implementation, when the input information of the channel generation model is noise information and / or random number information, the noise generator and / or random number generator can also be encapsulated in the channel data generator. In this case, the channel data generator can be without input, or in other words, the channel data generator does not need to input noise information and / or random number information.

[0176] Alternatively, in another possible implementation, a noise generator and / or a random number generator can be encapsulated in the channel generation model. That is, when the information processed by the channel generation model is noise information and / or random number information, the channel generation model can be without input. In this case, the channel generation model automatically generates noise information and / or random number information, processes the noise information and / or random number information, and outputs virtual channel data.

[0177] This application provides a method for constructing a virtual channel, which replaces the actual channel and the actual environment with a virtual environment, thereby reducing the dependence on actual channel environment data when conducting research and development on the integration of artificial intelligence and wireless communication systems.

[0178] The virtual channel environment provided in this application relies on a channel generation model for construction. This model can generate wireless channel data corresponding to one or more frequency bands, scenarios, and environments. The wireless channel data generated by the channel generation model can be used to construct datasets for AI solutions of wireless communication systems, or for channel analysis and modeling of wireless communication systems.

[0179] The virtual channel data output by the channel generation model can be used to simulate channel information under different frequencies, environments, and scenarios, such as high frequency, low frequency, indoor, outdoor, dense residential areas, open outdoor fields, IoT scenarios, and industrial scenarios.

[0180] Please refer to Figure 12 This diagram illustrates a block diagram of a channel data generation apparatus according to an embodiment of this application. The apparatus has the following features: Figure 5 or Figure 8 In the method shown, the function is performed by the channel data generation device. For example... Figure 12 As shown, the device may include:

[0181] The generation module 1201 is used to generate virtual channel data through a channel generation model; the virtual channel data is used to characterize the channel conditions in the channel environment.

[0182] The channel generation model is a machine learning model obtained by training channel data samples.

[0183] In one possible implementation, the generation module 1201 is used to input input information into the channel generation model to obtain the virtual channel data output by the channel generation model after processing the input information.

[0184] In one possible implementation, the input information includes at least one of the following four types of information:

[0185] Noise information, random number information, channel type indication information, or channel data sample information;

[0186] The channel type indication information is used to indicate the channel type;

[0187] The channel data sample information is information constructed based on the channel data sample.

[0188] In one possible implementation, the channel type indication information includes at least one of the following five types of information:

[0189] Temporal feature information, frequency feature information, spatial feature information, environmental feature information, or scene feature information.

[0190] In one possible implementation, the channel data sample information includes at least one of the following three types of information:

[0191] Information obtained by mixing noise into the channel data samples;

[0192] Information obtained by mixing random numbers into the channel data sample;

[0193] Alternatively, the information obtained by mixing noise and random numbers into the channel data samples.

[0194] In one possible implementation, the virtual channel data includes channel data corresponding to at least one granularity in at least one dimension.

[0195] In one possible implementation, the at least one dimension includes at least one of the following four dimensions:

[0196] Frequency domain dimension, time domain dimension, spatial domain dimension, or real / imaginary part dimension.

[0197] In one possible implementation, when the at least one dimension includes a frequency domain dimension, a granularity on the frequency domain dimension includes:

[0198] At least one RB or at least one subcarrier.

[0199] In one possible implementation, when the at least one dimension includes a time-domain dimension, a granularity of the time-domain dimension includes:

[0200] p1 is a microsecond, the length of at least one symbol, or the number of sampling points for at least one symbol; where p1 is a positive number.

[0201] In one possible implementation, when the at least one dimension includes a spatial domain dimension, a granularity on the spatial domain dimension includes:

[0202] At least one pair of transceiver antennas, at least one receiving antenna, at least one transmitting antenna, or an angular range of target angular intervals.

[0203] In one possible implementation, when the virtual channel data includes channel data corresponding to each granularity in at least two dimensions...

[0204] The virtual channel data includes a matrix with at least two dimensions;

[0205] Alternatively, the virtual channel data may comprise one-dimensional data obtained by arranging channel data corresponding to each granularity in the at least two dimensions.

[0206] In one possible implementation, the virtual channel data includes at least one of the following two types of information: original channel information, or a channel feature vector;

[0207] The channel feature vector is obtained by performing data transformation on the original channel information.

[0208] In one possible implementation, the original channel information includes channel quality information.

[0209] In one possible implementation, the channel feature vector comprises singular value decomposition of the original channel information.

[0210] In one possible implementation, the channel generation model includes at least one of the following four networks:

[0211] Fully connected networks, convolutional neural networks, residual networks, or self-attention mechanism networks.

[0212] In summary, the solution presented in this application pre-trains a channel generation model using channel data samples. This model can automatically generate virtual channel data corresponding to the channel environment through simulation and prediction, without requiring actual data collection. This greatly improves the efficiency of acquiring channel data under various channel environments, thereby enhancing the effectiveness of channel modeling and improving the accuracy of wireless communication system research and design.

[0213] Please refer to Figure 13This diagram illustrates a block diagram of a channel data generation apparatus according to an embodiment of this application. The apparatus has the following features: Figure 6 or Figure 8 The function shown is performed by the model training device in the method described. For example... Figure 13 As shown, the device may include:

[0214] Acquisition module 1301 is used to acquire channel data samples; the channel data samples are used to characterize the channel conditions in the sample channel environment;

[0215] Training module 1302 is used to train the channel generation model and the channel discrimination model based on the channel data samples by using the channel generation model as a generator and the channel discrimination model as a discriminator, through generative adversarial learning.

[0216] The channel generation model, after being trained to convergence, is used to generate virtual channel data; the virtual channel data is used to characterize the channel conditions in the channel environment.

[0217] In one possible implementation, the training module 1302 is used for,

[0218] During the training phase of the channel discrimination model, predicted virtual channel data is generated through the channel generation model.

[0219] The predicted virtual channel data and the channel data sample are input into the channel discrimination model to obtain a first discrimination result of the channel discrimination model; the first discrimination result is used to indicate whether the predicted virtual channel data and the channel data sample are channel data generated by the model.

[0220] The model parameters of the channel identification model are updated based on the first identification result.

[0221] In one possible implementation, the training module 1302 is used for,

[0222] During the training phase of the channel generation model, predicted virtual channel data is generated through the channel generation model.

[0223] The predicted virtual channel data is input into the channel discrimination model to obtain a second discrimination result of the channel discrimination model; the second discrimination result is used to indicate whether the predicted virtual channel data is channel data generated by the model.

[0224] The model parameters of the channel generation model are updated based on the second identification result.

[0225] In one possible implementation, the training module 1302 is used to input input information into the channel generation model to obtain the predicted virtual channel data output by the channel generation model after processing the input information sample.

[0226] In one possible implementation, the input information includes at least one of the following four types of information:

[0227] Noise information, random number information, channel type indication information, or channel data sample information;

[0228] The channel type indication information is used to indicate the channel type;

[0229] The channel data sample information is information constructed based on the channel data sample.

[0230] In one possible implementation, the channel type indication information is used to indicate the channel type corresponding to the channel data sample.

[0231] In one possible implementation, the channel type indication information includes at least one of the following five types of information:

[0232] Temporal feature information, frequency feature information, spatial feature information, environmental feature information, or scene feature information.

[0233] In one possible implementation, the channel data sample information includes at least one of the following three types of information:

[0234] Information obtained by mixing noise into the channel data samples;

[0235] Information obtained by mixing random numbers into the channel data sample;

[0236] Alternatively, the information obtained by mixing noise and random numbers in the channel data samples.

[0237] In summary, the scheme shown in this application pre-trains a channel generation model based on channel data samples using generative adversarial learning. This channel generation model can automatically generate virtual channel data corresponding to the channel environment through simulation and prediction. It can quickly obtain channel data corresponding to various channel environments without actual collection, thereby greatly improving the efficiency of acquiring channel data under various channel environments, thus improving the effect of channel modeling and the accuracy of wireless communication system research and design.

[0238] It should be noted that the device provided in the above embodiments is only illustrated by the division of the above functional modules when implementing its functions. In actual applications, the above functions can be assigned to different functional modules according to actual needs, that is, the content structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0239] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0240] Figure 14 A structural block diagram of a computer device 1400 illustrated in an exemplary embodiment of this application is shown. The computer device 1400 includes a Central Processing Unit (CPU) 1401, a system memory 1404 including Random Access Memory (RAM) 1402 and Read-Only Memory (ROM) 1403, and a system bus 1405 connecting the system memory 1404 and the CPU 1401. The computer device 1400 also includes a basic input / output system (I / O system) 1406 to facilitate information transfer between various devices within the computer, and a mass storage device 1407 for storing an operating system 1413, application programs 1414, and other program modules 1415.

[0241] The basic input / output system 1406 includes a display 1408 for displaying information and an input device 1409 for user input, such as a mouse or keyboard. Both the display 1408 and the input device 1409 are connected to the central processing unit 1401 via an input / output controller 1410 connected to the system bus 1405. The basic input / output system 1406 may also include the input / output controller 1410 for receiving and processing input from multiple other devices such as a keyboard, mouse, or electronic stylus. Similarly, the input / output controller 1410 also provides output to a display screen, printer, or other types of output devices.

[0242] The mass storage device 1407 is connected to the central processing unit 1401 via a mass storage controller (not shown) connected to the system bus 1405. The mass storage device 1407 and its associated computer-readable media provide non-volatile storage for the computer device 1400. That is, the mass storage device 1407 may include computer-readable media (not shown) such as a hard disk or a compact disc read-only memory (CD-ROM) drive.

[0243] Without loss of generality, the computer-readable medium may include computer storage media and communication media. Computer storage media include volatile and non-volatile, removable and non-removable media implemented using any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. Computer storage media include RAM, ROM, erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other solid-state storage technologies, CD-ROM, digital versatile disc (DVD) or other optical storage, magnetic tape cassettes, magnetic tape, disk storage, or other magnetic storage devices. Of course, those skilled in the art will recognize that the computer storage media are not limited to the above-mentioned types. The system memory 1404 and mass storage device 1407 described above can be collectively referred to as memory.

[0244] According to various embodiments of this disclosure, the computer device 1400 can also be connected to a remote computer on a network, such as the Internet. That is, the computer device 1400 can be connected to a network 1412 via a network interface unit 1411 connected to the system bus 1405, or the network interface unit 1411 can be used to connect to other types of networks or remote computer systems (not shown).

[0245] The memory also includes at least one computer instruction stored in the memory. The central processing unit 1401 executes the at least one instruction, at least one program, code set, or instruction set to implement all or part of the steps performed by the model training device or the channel data generation device in the methods shown in the above embodiments.

[0246] This application also provides a computer-readable storage medium storing a computer program, which is loaded and executed by a processor to implement the above-described embodiments. Figure 5 , Figure 6 or Figure 8 The methods shown include all or part of the steps performed by the model training device or the channel data generation device.

[0247] This application also provides a chip for operation in a computer device, causing the computer device to perform the above-described actions. Figure 5 , Figure 6 or Figure 8 The methods shown include all or part of the steps performed by the model training device or the channel data generation device.

[0248] This application also provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the above-described actions. Figure 5 , Figure 6 or Figure 8 The methods shown include all or part of the steps performed by the model training device or the channel data generation device.

[0249] This application also provides a computer program that is executed by the processor of a computer device to perform the above-described tasks. Figure 5 , Figure 6 or Figure 8 The methods shown include all or part of the steps performed by the model training device or the channel data generation device.

[0250] Those skilled in the art will recognize that the functions described in the embodiments of this application in one or more of the above examples can be implemented using hardware, software, firmware, or any combination thereof. When implemented using software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transfer of a computer program from one place to another. Storage media can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0251] The above description is merely an exemplary embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for generating channel data, characterized in that, The method includes: Input information is fed into a channel generation model to obtain virtual channel data output by the channel generation model after processing the input information; the virtual channel data is used to characterize the channel conditions in the channel environment; The channel generation model is a machine learning model trained using channel data samples. The input information includes at least one of channel type indication information or channel data sample information. The channel type indication information is used to make the virtual channel data output by the channel generation model simulate a virtual channel of the corresponding channel type. The channel type indication information includes at least one of the following five types of information: time-domain feature information, frequency feature information, spatial-domain feature information, environmental feature information, or scene feature information. The channel data sample information is used to make the channel environment of the virtual channel corresponding to the virtual channel data output by the channel generation model similar to the channel environment corresponding to the channel data sample. The sample information is constructed based on the channel data samples. The channel data sample information includes at least one of the following three types of information: information obtained by mixing noise into the channel data samples; information obtained by mixing random numbers into the channel data samples; or information obtained by mixing noise and random numbers into the channel data samples. The virtual channel data includes channel data corresponding to at least one granularity in at least one dimension. The at least one dimension includes at least one of the following four dimensions: frequency domain dimension, time domain dimension, spatial domain dimension, or real / imaginary part dimension. The channel quality indication of the virtual channel data in a combination of dimensions is used to represent the channel quality indication under the combination of dimensions.

2. The method according to claim 1, characterized in that, The input information also includes at least one of the following: noise information and random number information.

3. The method according to claim 1, characterized in that, When the at least one dimension includes a frequency domain dimension, a granularity of the frequency domain dimension includes: At least one RB or at least one subcarrier.

4. The method according to claim 1, characterized in that, When the at least one dimension includes a time-domain dimension, a granularity of the time-domain dimension includes: p1 is a microsecond, the length of at least one symbol, or the number of sampling points for at least one symbol; where p1 is a positive number.

5. The method according to claim 1, characterized in that, When the at least one dimension includes a spatial domain dimension, a granularity on the spatial domain dimension includes: At least one pair of transceiver antennas, at least one receiving antenna, at least one transmitting antenna, or an angular range of target angular intervals.

6. The method according to claim 1, characterized in that, When the virtual channel data includes channel data corresponding to each granularity in at least two dimensions... The virtual channel data includes a matrix with at least two dimensions; Alternatively, the virtual channel data may comprise one-dimensional data obtained by arranging channel data corresponding to each granularity in the at least two dimensions.

7. The method according to claim 1 or 2, characterized in that, The virtual channel data includes at least one of the following two types of information: original channel information, or channel feature vector; The channel feature vector is obtained by performing data transformation on the original channel information.

8. The method according to claim 7, characterized in that, The original channel information includes channel quality information.

9. The method according to claim 7, characterized in that, The channel feature vector is obtained by performing singular value decomposition on the original channel information.

10. The method according to claim 1 or 2, characterized in that, The channel generation model includes at least one of the following four types of networks: Fully connected networks, convolutional neural networks, residual networks, or self-attention mechanism networks.

11. A channel data processing method, characterized in that, The method includes: Acquire channel data samples; the channel data samples are used to characterize the channel conditions in the sample channel environment; Using a channel generation model as the generator and a channel discrimination model as the discriminator, the channel generation model and the channel discrimination model are trained based on the channel data samples through generative adversarial learning. The channel generation model, trained to convergence, is used to generate virtual channel data. This virtual channel data characterizes the channel conditions within the channel environment. The input information of the channel generation model includes at least one of channel type indication information or channel data sample information. The channel type indication information enables the virtual channel data output by the channel generation model to simulate a virtual channel of the corresponding channel type. This channel type indication information includes at least one of the following five types of information: time-domain feature information, frequency feature information, spatial-domain feature information, environmental feature information, or scene feature information. The channel data sample information enables the channel environment of the virtual channel corresponding to the virtual channel data output by the channel generation model to match the channel environment of the channel data sample. Given similar environments, the channel data sample information is constructed based on the channel data samples. The channel data sample information includes at least one of the following three types of information: information obtained by mixing noise into the channel data samples; information obtained by mixing random numbers into the channel data samples; or information obtained by mixing noise and random numbers into the channel data samples. The virtual channel data includes channel data corresponding to at least one granularity in at least one dimension. The at least one dimension includes at least one of the following four dimensions: frequency domain dimension, time domain dimension, spatial domain dimension, or real / imaginary part dimension. The channel quality indication of the virtual channel data in a combination of dimensions is used to represent the channel quality indication under the combination of dimensions.

12. The method according to claim 11, characterized in that, The process of using a channel generation model as a generator and a channel discrimination model as a discriminator, and training the channel generation model and the channel discrimination model based on the channel data samples through generative adversarial learning, includes: During the training phase of the channel discrimination model, predicted virtual channel data is generated through the channel generation model. The predicted virtual channel data and the channel data sample are input into the channel discrimination model to obtain a first discrimination result of the channel discrimination model; the first discrimination result is used to indicate whether the predicted virtual channel data and the channel data sample are channel data generated by the model. The model parameters of the channel identification model are updated based on the first identification result.

13. The method according to claim 11, characterized in that, The process of using a channel generation model as a generator and a channel discrimination model as a discriminator, and training the channel generation model and the channel discrimination model based on the channel data samples through generative adversarial learning, includes: During the training phase of the channel generation model, predicted virtual channel data is generated through the channel generation model. The predicted virtual channel data is input into the channel discrimination model to obtain a second discrimination result of the channel discrimination model; the second discrimination result is used to indicate whether the predicted virtual channel data is channel data generated by the model. The model parameters of the channel generation model are updated based on the second identification result.

14. The method according to claim 11, characterized in that, The input information also includes at least one of the following: Noise information, random number information.

15. A channel data generation apparatus, characterized in that, The device includes: The generation module is used to input input information into the channel generation model and obtain virtual channel data output by the channel generation model after processing the input information; the virtual channel data is used to characterize the channel conditions in the channel environment. The channel generation model is a machine learning model trained using channel data samples. The input information includes at least one of channel type indication information or channel data sample information. The channel type indication information is used to make the virtual channel data output by the channel generation model simulate a virtual channel of the corresponding channel type. The channel type indication information includes at least one of the following five types of information: time-domain feature information, frequency feature information, spatial-domain feature information, environmental feature information, or scene feature information. The channel data sample information is used to make the channel environment of the virtual channel corresponding to the virtual channel data output by the channel generation model similar to the channel environment corresponding to the channel data sample. The sample information is constructed based on the channel data samples. The channel data sample information includes at least one of the following three types of information: information obtained by mixing noise into the channel data samples; information obtained by mixing random numbers into the channel data samples; or information obtained by mixing noise and random numbers into the channel data samples. The virtual channel data includes channel data corresponding to at least one granularity in at least one dimension. The at least one dimension includes at least one of the following four dimensions: frequency domain dimension, time domain dimension, spatial domain dimension, or real / imaginary part dimension. The channel quality indication of the virtual channel data in a combination of dimensions is used to represent the channel quality indication under the combination of dimensions.

16. The apparatus according to claim 15, characterized in that, The input information also includes at least one of the following: Noise information, random number information.

17. The apparatus according to claim 15, characterized in that, When the at least one dimension includes a frequency domain dimension, a granularity of the frequency domain dimension includes: At least one RB or at least one subcarrier.

18. The apparatus according to claim 15, characterized in that, When the at least one dimension includes a time-domain dimension, a granularity of the time-domain dimension includes: p1 is a microsecond, the length of at least one symbol, or the number of sampling points for at least one symbol; where p1 is a positive number.

19. The apparatus according to claim 15, characterized in that, When the at least one dimension includes a spatial domain dimension, a granularity on the spatial domain dimension includes: At least one pair of transceiver antennas, at least one receiving antenna, at least one transmitting antenna, or an angular range of target angular intervals.

20. The apparatus according to claim 15, characterized in that, When the virtual channel data includes channel data corresponding to each granularity in at least two dimensions... The virtual channel data includes a matrix with at least two dimensions; Alternatively, the virtual channel data may comprise one-dimensional data obtained by arranging channel data corresponding to each granularity in the at least two dimensions.

21. The apparatus according to claim 15 or 16, characterized in that, The virtual channel data includes at least one of the following two types of information: original channel information, or channel feature vector; The channel feature vector is obtained by performing data transformation on the original channel information.

22. The apparatus according to claim 21, characterized in that, The original channel information includes channel quality information.

23. The apparatus according to claim 21, characterized in that, The channel feature vector is obtained by performing singular value decomposition on the original channel information.

24. The apparatus according to claim 15 or 16, characterized in that, The channel generation model includes at least one of the following four types of networks: Fully connected networks, convolutional neural networks, residual networks, or self-attention mechanism networks.

25. A channel data processing apparatus, characterized in that, The device includes: The acquisition module is used to acquire channel data samples; the channel data samples are used to characterize the channel conditions in the sample channel environment. The training module is used to train the channel generation model and the channel discrimination model based on the channel data samples through generative adversarial learning, using the channel generation model as the generator and the channel discrimination model as the discriminator. The channel generation model, trained to convergence, is used to generate virtual channel data. This virtual channel data characterizes the channel conditions within the channel environment. The input information of the channel generation model includes at least one of channel type indication information or channel data sample information. The channel type indication information enables the virtual channel data output by the channel generation model to simulate a virtual channel of the corresponding channel type. This channel type indication information includes at least one of the following five types of information: time-domain feature information, frequency feature information, spatial-domain feature information, environmental feature information, or scene feature information. The channel data sample information enables the channel environment of the virtual channel corresponding to the virtual channel data output by the channel generation model to match the channel environment of the channel data sample. Given similar environments, the channel data sample information is constructed based on the channel data samples. The channel data sample information includes at least one of the following three types of information: information obtained by mixing noise into the channel data samples; information obtained by mixing random numbers into the channel data samples; or information obtained by mixing noise and random numbers into the channel data samples. The virtual channel data includes channel data corresponding to at least one granularity in at least one dimension. The at least one dimension includes at least one of the following four dimensions: frequency domain dimension, time domain dimension, spatial domain dimension, or real / imaginary part dimension. The channel quality indication of the virtual channel data in a combination of dimensions is used to represent the channel quality indication under the combination of dimensions.

26. The apparatus according to claim 25, characterized in that, The process of using a channel generation model as a generator and a channel discrimination model as a discriminator, and training the channel generation model and the channel discrimination model based on the channel data samples through generative adversarial learning, includes: During the training phase of the channel discrimination model, predicted virtual channel data is generated through the channel generation model. The predicted virtual channel data and the channel data sample are input into the channel discrimination model to obtain a first discrimination result of the channel discrimination model; the first discrimination result is used to indicate whether the predicted virtual channel data and the channel data sample are channel data generated by the model. The model parameters of the channel identification model are updated based on the first identification result.

27. The apparatus according to claim 25, characterized in that, The process of using a channel generation model as a generator and a channel discrimination model as a discriminator, and training the channel generation model and the channel discrimination model based on the channel data samples through generative adversarial learning, includes: During the training phase of the channel generation model, predicted virtual channel data is generated through the channel generation model. The predicted virtual channel data is input into the channel discrimination model to obtain a second discrimination result of the channel discrimination model; the second discrimination result is used to indicate whether the predicted virtual channel data is channel data generated by the model. The model parameters of the channel generation model are updated based on the second identification result.

28. The apparatus according to claim 25, characterized in that, The input information also includes at least one of the following: Noise information, random number information.

29. A computer device, characterized in that, The computer device includes a processor, memory, and transceiver; The memory stores a computer program, and the processor executes the computer program to cause the computer device to implement the channel data generation method as shown in any one of claims 1 to 10, or to implement the channel data processing method as shown in any one of claims 11 to 14.

30. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that is executed by a processor to implement the channel data generation method as described in any one of claims 1 to 10, or to implement the channel data processing method as described in any one of claims 11 to 14.

31. A chip, characterized in that, The chip is configured to operate in a computer device to enable the computer device to perform the channel data generation method as described in any one of claims 1 to 10, or to implement the channel data processing method as described in any one of claims 11 to 14.

32. A computer program product, characterized in that, The computer program product includes computer instructions stored in a computer-readable storage medium; the processor of the computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the channel data generation method as described in any one of claims 1 to 10, or to implement the channel data processing method as described in any one of claims 11 to 14.

Images