Channel prediction method, device and program based on wireless environment information

By introducing wireless environment information and neural network technology into the channel prediction algorithm, combining simulation environment and deep probability sampling algorithm, the high-precision and low overhead problems of channel prediction in dynamic environments are solved, and are suitable for complex and changeable situations of 6G channels.

CN118944791AActive Publication Date: 2024-11-12BEIJING UNIV OF POSTS & TELECOMM

Patent Information

Application Number
CN202411097250.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-12
Estimated Expiration
2044-08-12

Smart Images

  • Figure CN118944791A_ABST
    Figure CN118944791A_ABST
Patent Text Reader

Abstract

The invention provides a channel prediction method, device and program based on wireless environment information, and relates to the technical field of wireless, and the channel prediction method comprises the steps: obtaining first CSI of a part in a first wireless scene according to a target pilot pattern, and obtaining multi-view image data of the first wireless scene; according to the first CSI, the multi-view image data and a target channel prediction model, predicting to obtain complete CSI in the first wireless scene; the target channel prediction model comprises an image feature extraction neural network, a target pilot pattern and a CSI reconstruction neural network, the image feature extraction neural network is used for extracting image features of the image data, and the CSI reconstruction neural network is used for obtaining complete CSI according to the image features and part of CSI prediction. On the premise of considering a dynamic environment scene, high-precision and low-overhead channel state information prediction is realized, and the requirements of a future communication system are met.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of wireless technology, and in particular to a channel prediction method, device and program based on wireless environment information. Background Art

[0002] In the sixth-generation (6G) massive multiple input multiple output (MIMO) system, the size of antenna arrays continues to expand and the number of antennas continues to increase. As an important part of wireless communication systems, the application of massive MIMO technology has brought many benefits to wireless communication systems, such as high spectrum efficiency, high energy efficiency, high spatial resolution, and high beamforming gain. In order to achieve these advantages, 6G massive MIMO systems need to obtain accurate channel state information (CSI).

[0003] At present, channel prediction methods based on artificial intelligence (AI) can obtain CSI in massive MIMO systems with low pilot overhead. Given that channels show certain correlations in space, time, and frequency dimensions, this method can mine the deep relationship of space-time-frequency multi-domain channels, thereby reducing the necessary multi-dimensional pilot signals.

[0004] However, the current traditional AI-based channel prediction algorithm uses the CSI of some antennas, time, and frequency to predict the entire CSI without considering the wireless environment information. This data-driven algorithm only generates a prediction model for training data and is difficult to generalize to other scenarios. Therefore, the traditional algorithm is only applicable to relatively stable static environments and cannot achieve stable prediction performance with low overhead in dynamic scenarios. In the future, when 6G channels are complex and changeable, the above algorithms can no longer meet the needs. Summary of the invention

[0005] The purpose of the present invention is to provide a channel prediction method, device and program based on wireless environment information to solve the problem that the existing channel prediction algorithm cannot achieve high-precision and low-overhead channel prediction in dynamic environment scenarios.

[0006] To solve the above technical problems, the embodiments of the present invention provide the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a channel prediction method, including:

[0008] Acquire a portion of first channel state information CSI in a first wireless scenario according to a target pilot pattern, and acquire multi-view image data of the first wireless scenario;

[0009] Predicting a complete CSI in the first wireless scenario according to the first CSI, the multi-view image data and the target channel prediction model;

[0010] Among them, the target channel prediction model includes an image feature extraction neural network, the target pilot pattern and a CSI reconstruction neural network, the image feature extraction neural network is used to extract image features of multi-view image data, and the CSI reconstruction neural network is used to obtain complete CSI based on image features and partial CSI prediction, and the image feature extraction neural network, the target pilot pattern and the CSI reconstruction neural network are trained based on the complete second CSI in the simulation environment of the first wireless scenario and the multi-view image data in the simulation environment of the first wireless scenario.

[0011] Optionally, the method further comprises:

[0012] Acquire the second CSI of the receiver at a target position in the simulation environment of the first wireless scenario, and acquire multi-view image data of the receiver at the target position in the simulation environment of the first wireless scenario, wherein the target position includes a first position and a second position;

[0013] According to the first network hyperparameter of the first neural network, the second CSI at the first position, and the multi-view image data at the first position, training the first image feature extraction parameter in the first neural network, the first pilot pattern, and the first CSI reconstruction parameter in the second neural network to obtain the second image feature extraction parameter, the second pilot pattern, and the second CSI reconstruction parameter;

[0014] Predicting a third CSI at the second position according to the second CSI at the second position, the second pilot pattern, the second image feature extraction parameter, the first network hyperparameter, the multi-view image data at the second position, and the second CSI reconstruction parameter;

[0015] Acquire first error information and first similarity information between the second CSI and the third CSI;

[0016] When the first error information and the first similarity information do not satisfy a first preset condition, adjusting the first network hyperparameter;

[0017] Returning the step of training the first image feature extraction parameter in the first neural network, the first pilot pattern and the first CSI reconstruction parameter in the second neural network according to the first network hyperparameter of the first neural network, the second CSI at the first position and the multi-view image data at the first position, to obtain the second image feature extraction parameter, the second pilot pattern and the second CSI reconstruction parameter, until the first error information and the first similarity information meet the first preset condition, obtaining the image feature extraction neural network according to the first network hyperparameter adjusted for the last time and the second image feature extraction parameter corresponding to the first network hyperparameter adjusted for the last time, and taking the second pilot pattern corresponding to the first network hyperparameter adjusted for the last time as the target pilot pattern, and obtaining the CSI reconstruction neural network according to the first CSI reconstruction parameter in the second neural network corresponding to the first network hyperparameter adjusted for the last time.

[0018] Optionally, according to the first network hyperparameter of the first neural network, the second CSI at the first position, and the multi-view image data at the first position, training the first image feature extraction parameter in the first neural network, the first pilot pattern, and the first CSI reconstruction parameter in the second neural network to obtain the second image feature extraction parameter, the second pilot pattern, and the second CSI reconstruction parameter, including:

[0019] Using the first neural network, and according to the first image feature extraction parameter, performing feature extraction on the multi-view image data at the first position to obtain an environmental feature map;

[0020] Obtaining, according to the first pilot pattern and the second CSI, a portion of fourth CSI at the first position;

[0021] Predicting a complete fifth CSI in a simulation environment of the first wireless scenario according to the first CSI reconstruction parameter in the second neural network, the environmental characteristic map, and the fourth CSI;

[0022] acquiring loss information between the fifth CSI and the second CSI;

[0023] adjusting the first image feature extraction parameter according to the loss information, and adjusting the first pilot pattern according to the loss information and the depth probability sampling algorithm, and adjusting the first CSI reconstruction parameter according to the loss information, and returning to the step of using the first neural network to extract features from the multi-view image data at the first position according to the first image feature extraction parameter to obtain an environmental feature map, until the number of iterations exceeds a preset number;

[0024] The first image feature extraction parameter adjusted last time is used as the second image feature extraction parameter, the first pilot pattern adjusted last time is used as the second pilot pattern, and the first CSI reconstruction parameter adjusted last time is used as the second CSI reconstruction parameter.

[0025] Optionally, predicting a complete fifth CSI in a simulation environment of the first wireless scenario according to the first CSI reconstruction parameter in the second neural network, the environmental characteristic map, and the fourth CSI includes:

[0026] The fifth CSI is predicted and obtained according to the first CSI reconstruction parameter, the environmental characteristic map and the fourth CSI by using the second neural network and the proximal gradient iteration algorithm.

[0027] Optionally, the first preset condition includes:

[0028] The first error information is smaller than the second error information corresponding to the first channel prediction model, and the first similarity information is larger than the second similarity information corresponding to the first channel prediction model;

[0029] The first channel prediction model includes at least one of the following:

[0030] A first sub-channel prediction model, the first sub-channel prediction model comprising: a randomly generated second pilot pattern and a CSI reconstruction neural network;

[0031] A second sub-channel prediction model, wherein the first sub-channel prediction model includes: a randomly generated second pilot pattern, the image feature extraction neural network and the CSI reconstruction neural network;

[0032] The third sub-channel prediction model, the first sub-channel prediction model includes: the target pilot pattern and the CSI reconstruction neural network.

[0033] Optionally, the method further comprises:

[0034] Acquire environmental information of the first wireless scene;

[0035] A modeling tool is used to construct a simulation environment of the first wireless scenario according to the environmental information.

[0036] Optionally, obtaining the second CSI of the receiver at a target position in the simulation environment of the first wireless scenario includes:

[0037] The second CSI of the receiver at the target position in the simulation environment is generated by using a ray tracing channel simulation tool.

[0038] Optionally, obtaining multi-view image data of the receiver at the target position in the simulation environment of the first wireless scene includes:

[0039] The autonomous driving simulation platform is used to obtain multi-view image data of the receiver at the target position in the simulation environment of the first wireless scene.

[0040] Optionally, the target position further includes a third position;

[0041] The method further comprises:

[0042] Obtaining a sixth CSI of a portion of the third position according to the second CSI at the third position and the target pilot pattern;

[0043] Extracting image features of the multi-view image data at the third position using the image feature extraction neural network;

[0044] Predicting a complete seventh CSI at the third position according to the image features and the sixth CSI through the CSI reconstruction neural network;

[0045] The accuracy of the target channel prediction model is obtained according to the third error information and the third similarity information between the seventh CSI and the second CSI at the third position.

[0046] Optionally, the method further comprises:

[0047] Acquire the complete eighth CSI in the simulation environment of the second wireless scenario and multi-view image data of the receiver in the simulation environment of the second wireless scenario;

[0048] Obtaining a ninth CSI of a portion of the simulation environment of the second wireless scenario according to the eighth CSI and the target pilot pattern;

[0049] Extracting image features of multi-view image data of a receiver in a simulation environment of the second wireless scenario using the image feature extraction neural network;

[0050] By means of the CSI reconstruction neural network, according to the image features and the ninth CSI, a complete tenth CSI in the simulation environment of the second wireless scenario is predicted;

[0051] A generalized applicability result of the target channel prediction model is obtained according to the fourth error information and the fourth similarity information between the tenth CSI and the eighth CSI.

[0052] In a second aspect, an embodiment of the present invention further provides a channel prediction device, including:

[0053] A first acquisition module, configured to acquire part of first channel state information CSI in a first wireless scenario according to a target pilot pattern and acquire multi-view image data of the first wireless scenario;

[0054] A first processing module, configured to predict and obtain a complete CSI in the first wireless scenario according to the first CSI, the multi-view image data and a target channel prediction model;

[0055] Among them, the target channel prediction model includes an image feature extraction neural network, the target pilot pattern and a CSI reconstruction neural network, the image feature extraction neural network is used to extract image features of multi-view image data, and the CSI reconstruction neural network is used to obtain complete CSI based on image features and partial CSI prediction, and the image feature extraction neural network, the target pilot pattern and the CSI reconstruction neural network are trained based on the complete second CSI in the simulation environment of the first wireless scenario and the multi-view image data in the simulation environment of the first wireless scenario.

[0056] In a third aspect, an embodiment of the present invention further provides a channel prediction device, comprising: a transceiver, a processor, a memory, and a program or instruction stored on the memory and executable on the processor; when the processor executes the program or instruction, it implements the steps in the channel prediction method described in any one of the first aspects.

[0057] In a fourth aspect, an embodiment of the present invention further provides a readable storage medium having a program or instruction stored thereon, which, when executed by a processor, implements the steps in the channel prediction method as described in any one of the first aspects.

[0058] In a fifth aspect, an embodiment of the present invention further provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the steps in the channel prediction method as described in any one of the first aspects.

[0059] The beneficial effects of the above technical solution of the present invention are as follows:

[0060] The channel prediction method provided by the scheme of the present invention obtains part of the first CSI in the first wireless scenario according to the target pilot pattern, uses the image feature extraction neural network in the target channel model to extract the image features of the multi-view image data of the first wireless scenario, and uses the CSI reconstruction neural network in the target channel model to predict the complete CSI in the first wireless scenario based on the image features and part of the first CSI, that is, under the premise of considering the dynamic environment scenario, high-precision and low-overhead channel state information prediction is achieved to meet the needs of future communication systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 A flow chart of a channel prediction method provided by an embodiment of the present invention;

[0062] Figure 2 A schematic diagram of error information comparison results provided by an embodiment of the present invention;

[0063] Figure 3 A schematic diagram of similarity comparison results provided by an embodiment of the present invention;

[0064] Figure 4 A schematic diagram of predicting complete CSI provided by an embodiment of the present invention;

[0065] Figure 5 A schematic diagram of the structure of a neural network in a target channel prediction model provided by an embodiment of the present invention;

[0066] Figure 6 A schematic diagram of the structure of a channel prediction device provided by an embodiment of the present invention;

[0067] Figure 7 A schematic diagram of the structure of a channel prediction device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0068] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0069] In order to solve the problem that the existing channel prediction algorithm cannot achieve high-precision and low-overhead channel prediction in dynamic environment scenarios, the embodiments of the present invention provide a channel prediction method, device and program based on wireless environment information.

[0070] like Figure 1 As shown, an embodiment of the present invention provides a channel prediction method, including:

[0071] Step 101: Acquire part of first channel state information CSI in a first wireless scenario according to a target pilot pattern, and acquire multi-view image data of the first wireless scenario.

[0072] The target pilot pattern is obtained by training based on the complete second CSI in the simulation environment of the first wireless scenario and multi-view image data in the simulation environment of the first wireless scenario.

[0073] It should be noted that there is an optimal pilot pattern for each scenario to facilitate the selection of the pilot position. The target pilot pattern is obtained by training based on the complete second CSI in the simulation environment of the first wireless scenario and the multi-view image data in the simulation environment of the first wireless scenario. The target pilot pattern is the optimal pilot pattern in the first wireless scenario.

[0074] After the target pilot pattern is determined, the transmitter and the receiver know the target pilot pattern and acquire part of the CSI, that is, the first CSI, according to the target pilot pattern.

[0075] Among them, the first wireless scene is any wireless communication scene, and the first wireless scene includes scene size information, scatterer positions in the scene, scatterer size information, transmitter (transmitter, Tx) position, receiver (receiver, Rx) position and other information.

[0076] In this step, the multi-perspective image data in the first wireless scene can be understood as the multi-perspective image data of the receiver in the first wireless scene, and the multi-perspective image data includes full-perspective image data. For example, the full-perspective image data includes image data of the east, west, south, and north of the receiver. Then the multi-perspective image data includes image data in at least one direction of east, west, south, and north, that is, the multi-perspective image data can also be understood as image data in different directions.

[0077] Step 102: predicting a complete CSI in the first wireless scenario according to the first CSI, the multi-view image data and the target channel prediction model.

[0078] Among them, the target channel prediction model includes an image feature extraction neural network, the target pilot pattern and a CSI reconstruction neural network, the image feature extraction neural network is used to extract image features of multi-view image data, and the CSI reconstruction neural network is used to obtain complete CSI based on image features and partial CSI prediction, and the image feature extraction neural network and the CSI reconstruction neural network are trained based on the complete second CSI in the simulation environment of the first wireless scenario and the multi-view image data in the simulation environment of the first wireless scenario.

[0079] Specifically, in this step, an image feature extraction neural network is used to extract image features of the multi-view image data, and a CSI reconstruction neural network is used to predict the complete CSI in the first wireless scenario based on the first CSI and the environmental features.

[0080] In this step, the image feature extraction neural network in the target channel prediction model is used to extract image features (also referred to as environmental features) in the first wireless communication scenario, and the image features are obtained by constructing a relationship between the environment and the channel.

[0081] In this step, the environmental features of the multi-view image data are extracted through an image feature extraction neural network. The first CSI of this part, the environmental features and the CSI reconstruction neural network are used to predict the complete CSI (or the overall CSI). In this embodiment of the present invention, environmental information is introduced into the prediction, which significantly improves the accuracy of the overall CSI prediction and reduces the pilot overhead.

[0082] Moreover, this method is more suitable for the future mobile communication system's demand for a wide variety of application scenarios and is more adaptable to different scenarios.

[0083] Furthermore, before training the target channel prediction model, the method further includes:

[0084] Acquire the second CSI of the receiver at the target position in the simulation environment of the first wireless scenario, and acquire multi-perspective image data of the receiver at the target position in the simulation environment of the first wireless scenario, wherein the target position includes a first position and a second position.

[0085] Among them, one receiver corresponds to one target position, or it can be understood that the position of the receiver in the first wireless scenario is the target position.

[0086] Specifically, in this step, for the first wireless scenario, the corresponding channel parameters of each receiver at the target location are the second CSI. For the first wireless scenario, the RGB images of each receiver in the four directions of east, west, south and north of the target location are the multi-view image data of the target location.

[0087] The CSI and multi-view image data are aligned according to the target position to obtain a channel-multi-view image data set, which includes the second CSI and multi-view image data corresponding to each target position. The target position includes a first position and a second position, that is, the channel-multi-view image data set includes the second CSI and multi-view image data corresponding to the first position and the second CSI and multi-view image data corresponding to the second position. It should also be noted that the target position may also include a third position, that is, the channel-multi-view image data set may also include the second CSI and multi-view image data corresponding to the third position.

[0088] This channel-multi-view image dataset provides a data basis for the training of subsequent target channel prediction models.

[0089] Optionally, the second CSI and multi-view image data corresponding to the first position can be used as a training data set, the second CSI and multi-view image data corresponding to the second position can be used as a verification data set, and the second CSI and multi-view image data at the third position can be used as a test data set, respectively for training, verification and testing of the target channel prediction model. In the target position, the ratio of the first position, the second position and the third position can be 0.7:0.1:0.2 respectively.

[0090] In an optional embodiment, the process of building a simulation environment includes:

[0091] Obtain environmental information of the first wireless scene, specifically, collect the three-dimensional wireless propagation environment of the first wireless scene, and automatically collect environmental information of the three-dimensional wireless propagation environment through actual measurement, map-assisted acquisition, and computer vision related technologies. The environmental information includes the overall layout of the wireless scene, scatterer information, transceiver information, etc., such as scene size information, scatterer positions in the scene, scatterer size information, transmitter position, receiver position and other information.

[0092] Using modeling tools, a simulation environment of the first wireless scene is constructed according to the environmental information. Specifically, using 3D modeling tools such as Blender, Google SketchUp, etc., according to the environmental information, a wireless propagation environment dataset with scatterer distribution is constructed in a coordinate system of the same size as the actual scene, that is, a simulation environment of the first wireless scene.

[0093] In an optional embodiment, obtaining the second CSI of the receiver at a target position in the simulation environment of the first wireless scenario includes:

[0094] The second CSI of the receiver at the target position in the simulation environment is generated by using a ray tracing channel simulation tool.

[0095] A ray tracing channel simulation tool is used to generate the second CSI of the receiver at the target position in the simulation environment. Specifically, a ray tracing (RT) channel simulation tool, such as WirelessInsite, is used to generate the corresponding channel parameters of each Rx position, that is, to generate the second CSI of the target position.

[0096] In an optional embodiment, obtaining multi-view image data of the receiver at the target location in the simulation environment of the first wireless scene includes:

[0097] An autonomous driving simulation platform is used to obtain multi-perspective image data of the receiver at the target position in the simulation environment of the first wireless scene. Specifically, an autonomous driving simulation platform such as carla is used to collect RGB images in the east, west, south and north directions of each Rx position.

[0098] The following uses a specific example to illustrate the construction process of the channel-multi-view image dataset:

[0099] A large outdoor urban scene with a length of 200m and a width of 200m was constructed in roadrunner, and a base station was set up to cover the scene. The scene contains four building complexes and four roads. Different types of vehicles are randomly placed on the road on one side of the road to ensure that the scenes of different roads have sufficient diversity. Among them, the building complexes, roads, and vehicles can all be considered as scatterers.

[0100] The scene is imported into carla, and multi-view image data at different Rx positions are obtained by deploying cameras at intervals of 0.26m on the road and capturing multi-view images in sequence.

[0101] Since the lack of surface details has limited impact on channel data, the entire scene was imported into blender software for simplification, where buildings and vehicles were replaced by simple cubes. The simplified scene model was imported into Wirelessinsite for RT simulation, channels at different Rx positions were generated, and CSI data was extracted. The number of transmitting antennas M for CSI data generated by simulation t =128, number of receiving antennas N r =1, the number of subcarriers N c =69, number of time domain orthogonal frequency division multiplexing (OFDM) symbols N T =3, the detailed RT simulation data are shown in Table 1 below.

[0102] Table 1

[0103]

[0104] The following describes the training process of the target channel prediction model in detail:

[0105] According to the first network hyperparameters of the first neural network, the second CSI at the first position and the multi-view image data at the first position, the first image feature extraction parameters in the first neural network, the first pilot pattern and the first CSI reconstruction parameters in the second neural network are trained to obtain the second image feature extraction parameters, the second pilot pattern and the second CSI reconstruction parameters.

[0106] The first network hyperparameter is a randomly generated initial network hyperparameter, the first image feature extraction parameter is a randomly generated initial image feature extraction parameter, and the first CSI reconstruction parameter is a randomly generated CSI reconstruction parameter.

[0107] The first neural network can be understood as an image feature extraction neural network before training and optimization, and the CSI reconstruction neural network includes a second neural network that has not been optimized and trained. The type of the first neural network should be consistent with the type of the image feature extraction neural network, and the CSI reconstruction neural network should be consistent with the type of the second neural network.

[0108] Among them, the first neural network can be a convolutional neural network (CNN), a self-attention layer, etc. It can be understood that the image feature extraction neural network is of the same type as the first neural network, and the image feature extraction neural network is obtained by training the first neural network.

[0109] The network hyperparameters are variables that determine the network structure (e.g., the number of hidden units) and variables that determine how the network is trained (e.g., the learning rate). In the context of machine learning, network hyperparameters are parameters whose values ​​are set before the learning process begins, rather than parameter data obtained through training. Typically, network hyperparameters need to be optimized to select a set of optimal hyperparameters for the learning machine to improve learning performance and results.

[0110] The first image feature extraction parameters are parameters mainly related to feature extraction and include weight parameters, bias parameters, and the like.

[0111] The target image feature extraction parameters are obtained by training and optimizing the first image feature extraction parameters using the above-mentioned training data, namely, including optimal weight parameters, bias parameters, etc.

[0112] The second neural network includes CNN, Recurrent Neural Network (RNN), etc.

[0113] The CSI reconstruction parameters include parameters related to CSI estimation, design parameters of the reconstruction algorithm, etc.

[0114] According to the first network hyperparameter of the first neural network, the second CSI at the first position, and the multi-view image data at the first position, the first image feature extraction parameter in the first neural network, the first pilot pattern, and the first CSI reconstruction parameter in the second neural network are trained to obtain the second image feature extraction parameter, the second pilot pattern, and the second CSI reconstruction parameter. Specifically, according to the second CSI at the second position, the CSI of the portion corresponding to the second position is obtained, and according to the first neural network and the first network hyperparameter in the first neural network, and the second image feature extraction parameter, the image features of the multi-view image data at the second position are extracted, and according to the image features, the CSI of the above portion, the second neural network, and the first CSI reconstruction parameter of the second neural network, a complete third CSI at the second position is predicted;

[0115] First error information and first similarity information between the second CSI and the third CSI are obtained, wherein the first error information is a normalized mean square error (Normalized Mean Square Error, NMSE), and the first similarity information is a cosine similarity (Cosine Similarity).

[0116] Determining whether the first error information and the first similarity information meet a first preset condition, wherein the first preset condition includes: the first error information is smaller than the second error information corresponding to the first channel prediction model, and the first similarity information is larger than the second similarity information corresponding to the first channel prediction model;

[0117] The first channel prediction model includes at least one of the following:

[0118] a first sub-channel prediction model, the first sub-channel prediction model comprising: a randomly generated second pilot pattern and a CSI reconstruction neural network, that is, the first sub-channel prediction model does not perform pilot pattern optimization and does not include an image feature extraction neural network for extracting image features of multi-view image data (abbreviated as non-ambient random sampling);

[0119] a second sub-channel prediction model, wherein the second sub-channel prediction model includes: a randomly generated second pilot pattern, the image feature extraction neural network and the CSI reconstruction neural network, that is, the first sub-channel prediction model does not perform pilot pattern optimization, but includes an image feature extraction neural network (abbreviated as environmental random sampling);

[0120] The third sub-channel prediction model includes: the target pilot pattern and the CSI reconstruction neural network, that is, the third sub-channel prediction model performs pilot pattern optimization but does not include an image feature extraction neural network (abbreviated as non-environmental DPS) for extracting image features of multi-view image data.

[0121] The target channel prediction model is referred to as the contextual DPS.

[0122] All the above models were trained for 200 epochs and used the same mean-square error (MSE) loss function.

[0123] Specifically, the first error information is compared with the second error information corresponding to the first sub-channel prediction model, the second sub-channel prediction model, and the third sub-channel prediction model, respectively. The comparison result is as follows: Figure 2 As shown, the first similarity information is compared with the second similarity information corresponding to the first sub-channel prediction model, the second sub-channel prediction model, and the third sub-channel prediction model, respectively. The comparison result is as shown in FIG. Figure 3 As shown, Figure 2 and Figure 3 In the table, “random+withoutenvironment” means “random sampling without environment”, “random+environment” means “random sampling with environment”, “DPS+without environment” means “DPS without environment”, and “DPS+withenvironment” means “DPS with environment”.

[0124] according to Figure 2 and Figure 3 , it can be determined whether the first error information and the first similarity information meet the first preset condition.

[0125] In the case that the first error information and the first similarity information do not meet the first preset condition, the first network hyperparameter is adjusted, and the first image feature extraction parameter and the first pilot pattern in the first neural network are trained according to the first network hyperparameter of the first neural network, the second CSI at the first position and the multi-view image data at the first position to obtain the second image feature extraction parameter and the second pilot pattern. It is judged again whether the adjusted first error information and the first similarity information meet the first preset condition. If the first preset condition is not met, the return is continued. If the condition is met, the training is terminated, and the image feature extraction neural network is obtained according to the first network hyperparameter after the last adjustment and the second image feature extraction parameter corresponding to the first network hyperparameter after the last adjustment, and the second pilot pattern corresponding to the first network hyperparameter after the last adjustment is used as the target pilot pattern.

[0126] Specifically, after the training is completed, the network hyperparameters in the first neural network are set to the first network hyperparameters adjusted last time, and the image feature extraction parameters in the first neural network are set to the second image feature extraction parameters corresponding to the first network hyperparameters adjusted last time, so as to obtain the trained image feature extraction neural network. The first network hyperparameters adjusted last time are the optimal network hyperparameters, and the second image feature extraction parameters corresponding to the first network hyperparameters adjusted last time are the optimal image feature extraction parameters.

[0127] like Figure 2 and Figure 3 As shown, the proposed method (target channel prediction model) obtains the highest correlation coefficient and the lowest NMSE in network model training. Comparing the results with and without the pilot optimization module (such as the green and dark black curves or the blue and red curves), under the same environment, the optimal pilot pattern is beneficial to improve the CSI prediction accuracy, and the NMSE is reduced by about 76.4% and 64.7%. In addition, under the same pilot optimization scheme, without the environment (such as the green-blue curve or the dark black-red curve), it can be seen that the environmental information brings gains, resulting in a reduction of NMSE by about 64.2% and 46.4%. In addition, compared with the 1 / 5 pilot without environmental random sampling (light black curve), the CSI prediction performance of the 1 / 8 pilot with environmental DPS is better. The pilot overhead is reduced by at least 37.5%.

[0128] Specifically, the overall network of the target neural channel prediction model is composed of three modules: image feature extraction module, pilot optimization module and CSI reconstruction neural network.

[0129] According to the first network hyperparameter of the first neural network, the second CSI at the first position, and the multi-view image data at the first position, the first image feature extraction parameter in the first neural network, the first pilot pattern, and the first CSI reconstruction parameter in the second neural network are trained to obtain the second image feature extraction parameter, the second pilot pattern, and the second CSI reconstruction parameter, including:

[0130] Using the first neural network, feature extraction is performed on the multi-view image data at the first position according to the first image feature extraction parameters to obtain an environmental feature map. Specifically, a multi-view image is acquired and input into an image feature extraction module, which includes an image feature extraction neural network. Image features of the multi-view image data are extracted through the image feature extraction neural network.

[0131] In the image feature extraction module, CNN can extract hierarchical features from multi-view image data and performs well in visual tasks. The three convolutional layers and one pooling layer of CNN are used to extract environmental feature maps from the panoramic image, which can reflect the scene environment information around the receiver.

[0132] In the pilot optimization module, the fourth CSI of the part under the first position is obtained according to the first pilot pattern and the second CSI, and the formula is as follows: H partial =H m,n A, where H partial Indicates the fourth CSI, H m,n represents the second CSI, and A represents the first pilot pattern;

[0133] Predicting a complete fifth CSI in a simulation environment of the first wireless scenario according to the first CSI reconstruction parameter in the second neural network, the environmental characteristic map, and the fourth CSI;

[0134] acquiring loss information between the fifth CSI and the second CSI;

[0135] adjusting the first image feature extraction parameter according to the loss information, and adjusting the first pilot pattern according to the loss information and the depth probability sampling algorithm, and adjusting the first CSI reconstruction parameter according to the loss information, and returning to the step of using the first neural network to extract features from the multi-view image data at the first position according to the first image feature extraction parameter to obtain an environmental feature map, until the number of iterations exceeds a preset number;

[0136] The first image feature extraction parameter adjusted last time is used as the second image feature extraction parameter, the first pilot pattern adjusted last time is used as the second pilot pattern, and the first CSI reconstruction parameter adjusted last time is used as the second CSI reconstruction parameter.

[0137] Specifically, the low-overhead optimal pilot pattern design can be described as a task-adaptive compressed sensing problem, whose goal is to select the optimal subset of signal samples to achieve end-to-end optimization. For each channel state information H corresponding to the mth transmit antenna and the nth receive antenna m,n (i.e., the second CSI), the deep probabilistic subsampling algorithm (DPS) generates an optimized pilot pattern in the time-frequency dimension based on the loss information (ie the adjusted first pilot pattern). is the CSI (partial CSI) in the time-frequency dimension under a given pilot pattern, which can be expressed as H partial =H m,n A. The number of elements with 1 in the optimized sampling matrix A is equal to the number of pilots N p The optimal pilot position is obtained during the training process of the neural network forward propagation. The pilot optimization module improves the channel prediction accuracy while minimizing the pilot overhead. The pilot sampling ratio is set to 1 / 8, that is, 1 / 8 (N c *N T / 8=26) elements are 1, and the rest are 0. c Indicates the number of subcarriers, N T represents the number of OFDM symbols in the time domain. After a preset number of iterations (ie, a preset number of times), the final optimized pilot pattern is the target pilot pattern. The preset number of times is 200.

[0138] In the CSI reconstruction neural network, the environment feature map and the CSI of this part are taken as input, and a complete CSI prediction value, namely, the fifth CSI, is obtained according to the first CSI reconstruction parameter.

[0139] The loss information between the fifth CSI and the second CSI is obtained, and the loss information can be obtained through a preset loss function.

[0140] After each acquisition of loss information, the first image feature extraction parameter, the first pilot pattern and the first CSI reconstruction parameter are adjusted according to the loss information, and then the next iteration is performed until the number of iterations exceeds 200, and the iteration process is terminated.

[0141] Further, according to the first CSI reconstruction parameter in the second neural network, the environmental characteristic map and the fourth CSI, predicting a complete fifth CSI in the simulation environment of the first wireless scenario, including:

[0142] The fifth CSI is predicted and obtained according to the first CSI reconstruction parameter, the environmental characteristic map and the fourth CSI by using the second neural network and the proximal gradient iteration algorithm.

[0143] Specifically, in the CSI reconstruction neural network, the environment feature map and the fourth CSI are used as input, and the second neural network (wherein the second neural network includes the first CSI reconstruction parameter) is combined with the proximal gradient iteration algorithm to predict the complete channel matrix {H} (ie, the fifth CSI), thereby accelerating the convergence of the model. The iterative steps are expressed as:

[0144]

[0145] X k+1 :=prox k (S k+1 ).

[0146] Among them, X k ,prox k , α k are the forward value, proximal gradient operator and channel prediction parameter of the kth iteration respectively. k+1 and S k+1 are the backward value and intermediate variable of the (k+1)th iteration respectively. The proximal operator prox is executed using the second neural network k and channel prediction parameter α k The number of algorithm iterations is set to 2 (i.e., k is equal to 0, 1, and 2), and the CSI data obtained after the iteration is combined with the channel number and the image feature map, and the complete CSI prediction value (i.e., the fifth CSI) is obtained through the second neural network.

[0147] The overall process of predicting the complete fifth CSI is as follows Figure 4 The schematic diagram of the structure of the neural network in the target channel prediction model is shown in Figure 5 shown. Figure 5 The real CSI in is the second CSI mentioned above, and the partial CSI is the fourth CSI.

[0148] Furthermore, the method further comprises:

[0149] Obtaining a sixth CSI of a portion of the third position according to the second CSI at the third position and the target pilot pattern;

[0150] Extracting image features of the multi-view image data at the third position using the image feature extraction neural network;

[0151] Predicting a complete seventh CSI at the third position according to the image features and the sixth CSI through the CSI reconstruction neural network;

[0152] The accuracy of the target channel prediction model is obtained according to the third error information and the third similarity information between the seventh CSI and the second CSI at the third position.

[0153] In the embodiment of the present invention, the accuracy of the target channel prediction model is tested using the above test data, wherein the third error information is the normalized mean square error, and the third similarity information is the cosine similarity information.

[0154] That is, using the test data, according to the fourth CSI acquisition process in the above part, the sixth CSI is obtained, and according to the fifth CSI acquisition process in the above part, the seventh CSI is obtained. The specific process will not be repeated here.

[0155] Furthermore, the method further comprises:

[0156] Acquire the complete eighth CSI in the simulation environment of the second wireless scenario and multi-view image data of the receiver in the simulation environment of the second wireless scenario;

[0157] The second wireless scene can be understood as a new scene different from the first wireless scene. For example, the building and vehicle layout of the original scene are modified to create a new scene. The data set (simulation environment) of the new scene includes 1601 Rxs of the two middle roads.

[0158] Obtaining a ninth CSI of a portion of the simulation environment of the second wireless scenario according to the eighth CSI and the target pilot pattern;

[0159] Extracting image features of multi-view image data of a receiver in a simulation environment of the second wireless scenario using the image feature extraction neural network;

[0160] By means of the CSI reconstruction neural network, according to the image features and the ninth CSI, a complete tenth CSI in the simulation environment of the second wireless scenario is predicted;

[0161] A generalized applicability result of the target channel prediction model is obtained according to the fourth error information and the fourth similarity information between the tenth CSI and the eighth CSI.

[0162] Among them, according to the fourth CSI acquisition process in the above part, the ninth CSI is obtained, and according to the fifth CSI acquisition process in the above part, the tenth CSI is obtained, and the specific process is not repeated here.

[0163] The fourth error information is a normalized mean square error, and the fourth similarity information is cosine similarity information.

[0164] Overall, the cosine similarity and NMSE in the new scenario are similar to those in the original scenario (the first wireless scenario), indicating that the target channel prediction model generalizes well.

[0165] Furthermore, the cosine similarity and NMSE of the original scene and the new scene under the target channel prediction model, the first sub-channel prediction model, the second sub-channel prediction model and the third sub-channel prediction model can be compared to obtain the target channel prediction model. The comparison and results are shown in Table 2 below.

[0166] Table 2

[0167]

[0168] In summary, the present invention not only uses the powerful feature extraction and nonlinear mapping capabilities of AI, but also combines the environmental feature information provided by the image. It can save pilot overhead while improving CSI prediction accuracy, solve the problem of high CSI acquisition overhead in large-scale MIMO systems, and optimize the pilot arrangement scheme. Compared with the random pilot scheme, it can increase the prediction accuracy, and can perform efficient and accurate CSI prediction for specific actual application scenarios. It introduces environmental information into the CSI prediction process, which increases the adaptability of the AI ​​model to the environment. It is more suitable for the future mobile communication system's needs for a wide variety of application scenarios, and is more scenario-universal.

[0169] like Figure 6 As shown, the embodiment of the present invention further provides a channel prediction device, comprising:

[0170] A first acquisition module 601 is configured to acquire a portion of first channel state information CSI in a first wireless scenario according to a target pilot pattern, and acquire multi-view image data of the first wireless scenario;

[0171] A first processing module 602 is configured to predict and obtain a complete CSI in the first wireless scenario according to the first CSI, the multi-view image data and a target channel prediction model;

[0172] Among them, the target channel prediction model includes an image feature extraction neural network, the target pilot pattern and a CSI reconstruction neural network, the image feature extraction neural network is used to extract image features of multi-view image data, and the CSI reconstruction neural network is used to obtain complete CSI based on image features and partial CSI prediction, and the image feature extraction neural network, the target pilot pattern and the CSI reconstruction neural network are trained based on the complete second CSI in the simulation environment of the first wireless scenario and the multi-view image data in the simulation environment of the first wireless scenario.

[0173] Optionally, the device further comprises:

[0174] A second acquisition module is used to acquire the second CSI of the receiver at a target position in the simulation environment of the first wireless scenario, and acquire multi-view image data of the receiver at the target position in the simulation environment of the first wireless scenario, wherein the target position includes a first position and a second position;

[0175] a second processing module, configured to train a first image feature extraction parameter in the first neural network, a first pilot pattern, and a first CSI reconstruction parameter in the second neural network according to a first network hyperparameter of the first neural network, the second CSI at the first position, and the multi-view image data at the first position, so as to obtain a second image feature extraction parameter, a second pilot pattern, and a second CSI reconstruction parameter;

[0176] a third processing module, configured to predict a third CSI at the second position according to the second CSI at the second position, the second pilot pattern, the second image feature extraction parameter, the first network hyperparameter, the multi-view image data at the second position, and the second CSI reconstruction parameter;

[0177] A third acquisition module, configured to acquire first error information and first similarity information between the second CSI and the third CSI;

[0178] a fourth processing module, configured to adjust the first network hyperparameter when the first error information and the first similarity information do not satisfy a first preset condition;

[0179] a fifth processing module, for returning the step of training the first image feature extraction parameter in the first neural network, the first pilot pattern and the first CSI reconstruction parameter in the second neural network according to the first network hyperparameter of the first neural network, the second CSI at the first position and the multi-view image data at the first position, and obtaining the second image feature extraction parameter, the second pilot pattern and the second CSI reconstruction parameter, until the first error information and the first similarity information meet the first preset condition, obtaining the image feature extraction neural network according to the first network hyperparameter adjusted for the last time and the second image feature extraction parameter corresponding to the first network hyperparameter adjusted for the last time, and using the second pilot pattern corresponding to the first network hyperparameter adjusted for the last time as the target pilot pattern, and obtaining the CSI reconstruction neural network according to the first CSI reconstruction parameter in the second neural network corresponding to the first network hyperparameter adjusted for the last time.

[0180] Optionally, the second processing module includes:

[0181] A first processing unit is used to use the first neural network to extract features from the multi-view image data at the first position according to the first image feature extraction parameter to obtain an environment feature map;

[0182] A second processing unit, configured to obtain a fourth CSI of a portion of the first position according to the first pilot pattern and the second CSI;

[0183] A third processing unit is configured to predict a complete fifth CSI in a simulation environment of the first wireless scenario according to the first CSI reconstruction parameter in the second neural network, the environmental characteristic map and the fourth CSI;

[0184] A first acquiring unit, configured to acquire loss information between the fifth CSI and the second CSI;

[0185] a fourth processing unit, configured to adjust the first image feature extraction parameter according to the loss information, adjust the first pilot pattern according to the loss information and the depth probability sampling algorithm, and adjust the first CSI reconstruction parameter according to the loss information, and return to the step of using the first neural network to extract features from the multi-view image data at the first position according to the first image feature extraction parameter to obtain an environmental feature map, until the number of iterations exceeds a preset number;

[0186] The fifth processing unit is used to use the first image feature extraction parameter adjusted for the last time as the second image feature extraction parameter, and to use the first pilot pattern adjusted for the last time as the second pilot pattern, and to use the first CSI reconstruction parameter adjusted for the last time as the second CSI reconstruction parameter.

[0187] Optionally, the third processing unit is specifically configured to:

[0188] The fifth CSI is predicted and obtained according to the first CSI reconstruction parameter, the environmental characteristic map and the fourth CSI by using the second neural network and the proximal gradient iteration algorithm.

[0189] Optionally, the first preset condition includes:

[0190] The first error information is smaller than the second error information corresponding to the first channel prediction model, and the first similarity information is larger than the second similarity information corresponding to the first channel prediction model;

[0191] The first channel prediction model includes at least one of the following:

[0192] A first sub-channel prediction model, the first sub-channel prediction model comprising: a randomly generated second pilot pattern and a CSI reconstruction neural network;

[0193] A second sub-channel prediction model, the second sub-channel prediction model comprising: a randomly generated second pilot pattern, the image feature extraction neural network and the CSI reconstruction neural network;

[0194] A third sub-channel prediction model, wherein the third sub-channel prediction model includes: the target pilot pattern and a CSI reconstruction neural network.

[0195] Optionally, the device further comprises:

[0196] A fourth acquisition module, used to acquire environmental information of the first wireless scene;

[0197] The sixth processing module is used to use a modeling tool to construct a simulation environment of the first wireless scenario according to the environmental information.

[0198] Optionally, the second acquisition module includes:

[0199] The second acquisition unit is configured to generate the second CSI of the receiver at the target position in the simulation environment by using a ray tracing channel simulation tool.

[0200] Optionally, the second acquisition module includes:

[0201] The third acquisition unit is used to acquire multi-view image data of the receiver at the target position in the simulation environment of the first wireless scene by using the autonomous driving simulation platform.

[0202] Optionally, the target position further includes a third position;

[0203] The device also includes:

[0204] a seventh processing module, configured to obtain a sixth CSI of a portion of the third position according to the second CSI and the target pilot pattern at the third position;

[0205] An eighth processing module, configured to extract image features of the multi-view image data at the third position by using the image feature extraction neural network;

[0206] A ninth processing module, configured to reconstruct a neural network through the CSI, and predict a complete seventh CSI at the third position according to the image features and the sixth CSI;

[0207] The tenth processing module is used to obtain the accuracy of the target channel prediction model according to the third error information and the third similarity information between the seventh CSI and the second CSI at the third position.

[0208] Optionally, the device further comprises:

[0209] A fifth acquisition module, used to acquire the complete eighth CSI in the simulation environment of the second wireless scenario and multi-view image data of the receiver in the simulation environment of the second wireless scenario;

[0210] An eleventh processing module, configured to obtain a ninth CSI in a simulation environment of the second wireless scenario according to the eighth CSI and the target pilot pattern;

[0211] A twelfth processing module, configured to extract image features of multi-view image data of a receiver in a simulation environment of the second wireless scene by using the image feature extraction neural network;

[0212] A thirteenth processing module is used to reconstruct a neural network through the CSI, and predict a complete tenth CSI in a simulation environment of the second wireless scenario according to the image features and the ninth CSI;

[0213] The fourteenth processing module is used to obtain a generalized applicability result of the target channel prediction model according to the fourth error information and the fourth similarity information between the tenth CSI and the eighth CSI.

[0214] It should be noted that the channel prediction device provided in the embodiment of the present invention is a device capable of executing the above-mentioned channel prediction method, and all embodiments of the above-mentioned channel prediction method are applicable to the device and can achieve the same or similar technical effects.

[0215] like Figure 7 As shown, an embodiment of the present invention also provides a channel prediction device, including: a processor 701; and a memory 703 connected to the processor 701 through a bus interface 702, the memory 703 is used to store programs and data used by the processor 701 when performing operations, and the processor 701 calls and executes the programs and data stored in the memory 703.

[0216] The transceiver 704 is connected to the bus interface 702 and is used to receive and send data under the control of the processor 701. Specifically, the processor 701 is used to read the program in the memory 703 and execute the following process:

[0217] Acquire a portion of first channel state information CSI in a first wireless scenario according to a target pilot pattern, and acquire multi-view image data of the first wireless scenario;

[0218] Predicting a complete CSI in the first wireless scenario according to the first CSI, the multi-view image data and the target channel prediction model;

[0219] Among them, the target channel prediction model includes an image feature extraction neural network, the target pilot pattern and a CSI reconstruction neural network, the image feature extraction neural network is used to extract image features of multi-view image data, and the CSI reconstruction neural network is used to obtain complete CSI based on image features and partial CSI prediction, and the image feature extraction neural network, the target pilot pattern and the CSI reconstruction neural network are trained based on the complete second CSI in the simulation environment of the first wireless scenario and the multi-view image data in the simulation environment of the first wireless scenario.

[0220] Optionally, the processor 701 is further configured to:

[0221] Acquire the second CSI of the receiver at a target position in the simulation environment of the first wireless scenario, and acquire multi-view image data of the receiver at the target position in the simulation environment of the first wireless scenario, wherein the target position includes a first position and a second position;

[0222] According to the first network hyperparameter of the first neural network, the second CSI at the first position, and the multi-view image data at the first position, training the first image feature extraction parameter in the first neural network, the first pilot pattern, and the first CSI reconstruction parameter in the second neural network to obtain the second image feature extraction parameter, the second pilot pattern, and the second CSI reconstruction parameter;

[0223] Predicting a third CSI at the second position according to the second CSI at the second position, the second pilot pattern, the second image feature extraction parameter, the first network hyperparameter, the multi-view image data at the second position, and the second CSI reconstruction parameter;

[0224] Acquire first error information and first similarity information between the second CSI and the third CSI;

[0225] When the first error information and the first similarity information do not satisfy a first preset condition, adjusting the first network hyperparameter;

[0226] Returning the step of training the first image feature extraction parameter in the first neural network, the first pilot pattern and the first CSI reconstruction parameter in the second neural network according to the first network hyperparameter of the first neural network, the second CSI at the first position and the multi-view image data at the first position, to obtain the second image feature extraction parameter, the second pilot pattern and the second CSI reconstruction parameter, until the first error information and the first similarity information meet the first preset condition, obtaining the image feature extraction neural network according to the first network hyperparameter adjusted for the last time and the second image feature extraction parameter corresponding to the first network hyperparameter adjusted for the last time, and taking the second pilot pattern corresponding to the first network hyperparameter adjusted for the last time as the target pilot pattern, and obtaining the CSI reconstruction neural network according to the first CSI reconstruction parameter in the second neural network corresponding to the first network hyperparameter adjusted for the last time.

[0227] Optionally, the processor 701 is specifically configured to:

[0228] Using the first neural network, and according to the first image feature extraction parameter, performing feature extraction on the multi-view image data at the first position to obtain an environmental feature map;

[0229] Obtaining, according to the first pilot pattern and the second CSI, a portion of fourth CSI at the first position;

[0230] Predicting a complete fifth CSI in a simulation environment of the first wireless scenario according to the first CSI reconstruction parameter in the second neural network, the environmental characteristic map, and the fourth CSI;

[0231] acquiring loss information between the fifth CSI and the second CSI;

[0232] adjusting the first image feature extraction parameter according to the loss information, and adjusting the first pilot pattern according to the loss information and the depth probability sampling algorithm, and adjusting the first CSI reconstruction parameter according to the loss information, and returning to the step of using the first neural network to extract features from the multi-view image data at the first position according to the first image feature extraction parameter to obtain an environmental feature map, until the number of iterations exceeds a preset number;

[0233] The first image feature extraction parameter adjusted last time is used as the second image feature extraction parameter, the first pilot pattern adjusted last time is used as the second pilot pattern, and the first CSI reconstruction parameter adjusted last time is used as the second CSI reconstruction parameter.

[0234] Optionally, the processor 701 is specifically configured to:

[0235] The fifth CSI is predicted and obtained according to the first CSI reconstruction parameter, the environmental characteristic map and the fourth CSI by using the second neural network and the proximal gradient iteration algorithm.

[0236] Optionally, the first preset condition includes:

[0237] The first error information is smaller than the second error information corresponding to the first channel prediction model, and the first similarity information is larger than the second similarity information corresponding to the first channel prediction model;

[0238] The first channel prediction model includes at least one of the following:

[0239] A first sub-channel prediction model, the first sub-channel prediction model comprising: a randomly generated second pilot pattern and a CSI reconstruction neural network;

[0240] A second sub-channel prediction model, the second sub-channel prediction model comprising: a randomly generated second pilot pattern, the image feature extraction neural network and the CSI reconstruction neural network;

[0241] A third sub-channel prediction model, wherein the third sub-channel prediction model includes: the target pilot pattern and a CSI reconstruction neural network.

[0242] Optionally, the processor 701 is further configured to:

[0243] Acquire environmental information of the first wireless scene;

[0244] A modeling tool is used to construct a simulation environment of the first wireless scenario according to the environmental information.

[0245] Optionally, the processor 701 is specifically configured to:

[0246] The second CSI of the receiver at the target position in the simulation environment is generated by using a ray tracing channel simulation tool.

[0247] Optionally, the processor 701 is specifically configured to:

[0248] The autonomous driving simulation platform is used to obtain multi-view image data of the receiver at the target position in the simulation environment of the first wireless scene.

[0249] Optionally, the target position further includes a third position;

[0250] The processor 701 is further configured to:

[0251] Obtaining a sixth CSI of a portion of the third position according to the second CSI at the third position and the target pilot pattern;

[0252] Extracting image features of the multi-view image data at the third position using the image feature extraction neural network;

[0253] Predicting a complete seventh CSI at the third position according to the image features and the sixth CSI through the CSI reconstruction neural network;

[0254] The accuracy of the target channel prediction model is obtained according to the third error information and the third similarity information between the seventh CSI and the second CSI at the third position.

[0255] Optionally, the processor 701 is further configured to:

[0256] Acquire the complete eighth CSI in the simulation environment of the second wireless scenario and multi-view image data of the receiver in the simulation environment of the second wireless scenario;

[0257] Obtaining a ninth CSI of a portion of the simulation environment of the second wireless scenario according to the eighth CSI and the target pilot pattern;

[0258] Extracting image features of multi-view image data of a receiver in a simulation environment of the second wireless scenario using the image feature extraction neural network;

[0259] By means of the CSI reconstruction neural network, according to the image features and the ninth CSI, a complete tenth CSI in the simulation environment of the second wireless scenario is predicted;

[0260] A generalized applicability result of the target channel prediction model is obtained according to the fourth error information and the fourth similarity information between the tenth CSI and the eighth CSI.

[0261] Among them, Figure 7 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by processor 701 and memory represented by memory 703. The bus architecture may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides a user interface 705. The transceiver 704 may be a plurality of components, i.e., including a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium. The processor 701 is responsible for managing the bus architecture and general processing, and the memory 703 may store data used by the processor 701 when performing operations.

[0262] A readable storage medium according to an embodiment of the present invention stores a program or instruction thereon. When the program or instruction is executed by a processor, the steps in the channel prediction method as described above are implemented and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0263] In the embodiment of the present invention, module can be implemented with software so that it can be executed by various types of processors. For example, an executable code module of an identification can include one or more physical or logical blocks of computer instructions, for example, it can be constructed as an object, process or function. Nevertheless, the executable code of the identified module does not need to be physically located together, but can include different instructions stored in different positions, and when these instructions are logically combined together, it constitutes a module and realizes the specified purpose of the module.

[0264] In fact, executable code module can be a single instruction or many instructions, and can even be distributed on a plurality of different code segments, distributed among different programs, and distributed across a plurality of memory devices. Similarly, operating data can be identified in the module, and can be implemented and organized in the data structure of any appropriate type according to any appropriate form. The operating data can be collected as a single data set, or can be distributed in different locations (including on different storage devices), and can only be present on a system or network as an electronic signal at least in part.

[0265] When a module can be implemented by software, considering the level of existing hardware technology, a person skilled in the art can build a corresponding hardware circuit to implement the corresponding function of the module that can be implemented by software without considering the cost. The hardware circuit includes a conventional very large scale integration (VLSI) circuit or gate array and existing semiconductors such as logic chips, transistors, or other discrete components. The module can also be implemented by a programmable hardware device, such as a field programmable gate array, a programmable array logic, a programmable logic device, etc.

[0266] The above exemplary embodiments are described with reference to the accompanying drawings, and many different forms and embodiments are feasible without departing from the spirit and teachings of the present invention. Therefore, the present invention should not be constructed as a limitation of the exemplary embodiments proposed herein. More specifically, these exemplary embodiments are provided so that the present invention will be perfect and complete, and the scope of the present invention will be conveyed to those who are familiar with the technology. In these figures, the component sizes and relative sizes may be exaggerated for clarity. The terms used here are only based on the purpose of describing specific exemplary embodiments and are not intended to be limiting. As used herein, unless the text clearly indicates otherwise, the singular forms "one", "an" and "the" are intended to include these multiple forms. It will be further understood that the terms "including" and / or "comprising" when used in this specification indicate the presence of the features, integers, steps, operations, components and / or components, but do not exclude the presence or increase of one or more other features, integers, steps, operations, components, components and / or their groups. Unless otherwise indicated, when stated, a range of values ​​includes the upper and lower limits of that range and any subranges therebetween.

[0267] A specific embodiment of the present invention further provides a computer program product, including computer instructions, which, when executed by a processor, implement the above Figure 1 The various processes of the method embodiment shown can achieve the same technical effect, and will not be described again here to avoid repetition.

[0268] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A channel prediction method, characterized in that: include: Acquire a portion of first channel state information CSI in a first wireless scenario according to a target pilot pattern, and acquire multi-view image data of the first wireless scenario; Predicting a complete CSI in the first wireless scenario according to the first CSI, the multi-view image data and the target channel prediction model; Among them, the target channel prediction model includes an image feature extraction neural network, the target pilot pattern and a CSI reconstruction neural network, the image feature extraction neural network is used to extract image features of multi-view image data, and the CSI reconstruction neural network is used to obtain complete CSI based on image features and partial CSI prediction, and the image feature extraction neural network, the target pilot pattern and the CSI reconstruction neural network are trained based on the complete second CSI in the simulation environment of the first wireless scenario and the multi-view image data in the simulation environment of the first wireless scenario.

2. The method according to claim 1, characterized in that The method further comprises: Acquire the second CSI of the receiver at a target position in the simulation environment of the first wireless scenario, and acquire multi-view image data of the receiver at the target position in the simulation environment of the first wireless scenario, wherein the target position includes a first position and a second position; According to the first network hyperparameter of the first neural network, the second CSI at the first position, and the multi-view image data at the first position, training the first image feature extraction parameter in the first neural network, the first pilot pattern, and the first CSI reconstruction parameter in the second neural network to obtain the second image feature extraction parameter, the second pilot pattern, and the second CSI reconstruction parameter; Predicting a third CSI at the second position according to the second CSI at the second position, the second pilot pattern, the second image feature extraction parameter, the first network hyperparameter, the multi-view image data at the second position, and the second CSI reconstruction parameter; Acquire first error information and first similarity information between the second CSI and the third CSI; When the first error information and the first similarity information do not satisfy a first preset condition, adjusting the first network hyperparameter; Returning the step of training the first image feature extraction parameter in the first neural network, the first pilot pattern and the first CSI reconstruction parameter in the second neural network according to the first network hyperparameter of the first neural network, the second CSI at the first position and the multi-view image data at the first position, to obtain the second image feature extraction parameter, the second pilot pattern and the second CSI reconstruction parameter, until the first error information and the first similarity information meet the first preset condition, obtaining the image feature extraction neural network according to the first network hyperparameter adjusted for the last time and the second image feature extraction parameter corresponding to the first network hyperparameter adjusted for the last time, and taking the second pilot pattern corresponding to the first network hyperparameter adjusted for the last time as the target pilot pattern, and obtaining the CSI reconstruction neural network according to the first CSI reconstruction parameter in the second neural network corresponding to the first network hyperparameter adjusted for the last time.

3. The method according to claim 2, characterized in that According to the first network hyperparameter of the first neural network, the second CSI at the first position, and the multi-view image data at the first position, the first image feature extraction parameter in the first neural network, the first pilot pattern, and the first CSI reconstruction parameter in the second neural network are trained to obtain the second image feature extraction parameter, the second pilot pattern, and the second CSI reconstruction parameter, including: Using the first neural network, and according to the first image feature extraction parameter, performing feature extraction on the multi-view image data at the first position to obtain an environmental feature map; Obtaining, according to the first pilot pattern and the second CSI, a portion of fourth CSI at the first position; Predicting a complete fifth CSI in a simulation environment of the first wireless scenario according to the first CSI reconstruction parameter in the second neural network, the environmental characteristic map, and the fourth CSI; acquiring loss information between the fifth CSI and the second CSI; adjusting the first image feature extraction parameter according to the loss information, and adjusting the first pilot pattern according to the loss information and the depth probability sampling algorithm, and adjusting the first CSI reconstruction parameter according to the loss information, and returning to the step of using the first neural network to extract features from the multi-view image data at the first position according to the first image feature extraction parameter to obtain an environmental feature map, until the number of iterations exceeds a preset number; The first image feature extraction parameter adjusted last time is used as the second image feature extraction parameter, the first pilot pattern adjusted last time is used as the second pilot pattern, and the first CSI reconstruction parameter adjusted last time is used as the second CSI reconstruction parameter.

4. The method according to claim 3, characterized in that Predicting a complete fifth CSI in a simulation environment of the first wireless scenario according to the first CSI reconstruction parameter in the second neural network, the environmental characteristic map, and the fourth CSI, including: The fifth CSI is predicted and obtained according to the first CSI reconstruction parameter, the environmental characteristic map and the fourth CSI by using the second neural network and the proximal gradient iteration algorithm.

5. The method according to claim 2, characterized in that: The first preset condition includes: The first error information is smaller than the second error information corresponding to the first channel prediction model, and the first similarity information is larger than the second similarity information corresponding to the first channel prediction model; The first channel prediction model includes at least one of the following: A first sub-channel prediction model, the first sub-channel prediction model comprising: a randomly generated second pilot pattern and a CSI reconstruction neural network; A second sub-channel prediction model, the second sub-channel prediction model comprising: a randomly generated second pilot pattern, the image feature extraction neural network and the CSI reconstruction neural network; A third sub-channel prediction model, wherein the third sub-channel prediction model includes: the target pilot pattern and a CSI reconstruction neural network.

6. The method according to claim 1 or 2, characterized in that: The method further comprises: Acquire environmental information of the first wireless scene; A modeling tool is used to construct a simulation environment of the first wireless scenario according to the environmental information.

7. The method according to claim 2, characterized in that Acquiring the second CSI of the receiver at a target position in the simulation environment of the first wireless scenario includes: The second CSI of the receiver at the target position in the simulation environment is generated by using a ray tracing channel simulation tool.

8. The method according to claim 2, characterized in that: Acquiring multi-view image data of the receiver at the target location in the simulation environment of the first wireless scene, including: The autonomous driving simulation platform is used to obtain multi-view image data of the receiver at the target position in the simulation environment of the first wireless scene.

9. The method according to claim 2, characterized in that: The target location also includes a third location; The method further comprises: Obtaining a sixth CSI of a portion of the third position according to the second CSI at the third position and the target pilot pattern; Extracting image features of the multi-view image data at the third position using the image feature extraction neural network; Predicting a complete seventh CSI at the third position according to the image features and the sixth CSI through the CSI reconstruction neural network; The accuracy of the target channel prediction model is obtained according to the third error information and the third similarity information between the seventh CSI and the second CSI at the third position.

10. The method according to claim 2, characterized in that The method further comprises: Acquire the complete eighth CSI in the simulation environment of the second wireless scenario and multi-view image data of the receiver in the simulation environment of the second wireless scenario; Obtaining a ninth CSI in a simulation environment of the second wireless scenario according to the eighth CSI and the target pilot pattern; Extracting image features of multi-view image data of a receiver in a simulation environment of the second wireless scenario using the image feature extraction neural network; By means of the CSI reconstruction neural network, according to the image features and the ninth CSI, a complete tenth CSI in the simulation environment of the second wireless scenario is predicted; A generalized applicability result of the target channel prediction model is obtained according to the fourth error information and the fourth similarity information between the tenth CSI and the eighth CSI.

11. A channel prediction device, characterized in that: include: A first acquisition module, configured to acquire part of first channel state information CSI in a first wireless scenario according to a target pilot pattern and acquire multi-view image data of the first wireless scenario; A first processing module, configured to predict and obtain complete CSI in the first wireless scenario according to the first CSI, the multi-view image data and a target channel prediction model; Among them, the target channel prediction model includes an image feature extraction neural network, the target pilot pattern and a CSI reconstruction neural network, the image feature extraction neural network is used to extract image features of multi-view image data, and the CSI reconstruction neural network is used to obtain complete CSI based on image features and partial CSI prediction, and the image feature extraction neural network, the target pilot pattern and the CSI reconstruction neural network are trained based on the complete second CSI in the simulation environment of the first wireless scenario and the multi-view image data in the simulation environment of the first wireless scenario.

12. A channel prediction device, comprising: A transceiver, a processor, a memory, and a program or instruction stored in the memory and executable on the processor; wherein the processor implements the steps of the channel prediction method as described in any one of claims 1 to 10 when executing the program or instruction.

13. A readable storage medium having a program or instruction stored thereon, characterized in that: When the program or instruction is executed by a processor, the steps in the channel prediction method according to any one of claims 1 to 10 are implemented.

14. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implement the steps in the channel prediction method according to any one of claims 1 to 9.

Citation Information

Patent Citations

  • Transmission method and apparatus of channel state information reference signal

    CN108631994A

  • Orthogonal frequency division multiplexing (OFDM) system with channel transfer function prediction

    US20030232601A1

  • Wireless signal information transmission

    US20240056856A1

Cited By

  • 6G dynamic channel map construction method based on machine learning and image processing

    CN119649633A

  • Wireless channel prediction method based on machine learning and geographical environment information

    CN120107613A

  • Wireless channel prediction method based on machine learning and geographical environment information

    CN120107613B

  • Communication method and related apparatus

    WO2026118808A1

  • Hybrid receiver with learnable decider

    WO2026157378A1