A vehicle network channel prediction method and related equipment in complex traffic scenarios

Abstract

The present invention relates to the field of vehicle network communication technology, and discloses a vehicle network channel prediction method and related equipment in complex traffic scenarios. The method includes constructing an urban road scenario of a vehicle network channel, simulating and obtaining V2I communication results under the urban road scenario, and obtaining scene data through V2I communication result collection; pre-processing and forward-noising the scene data, and then embedding side information to obtain channel data with spatiotemporal characteristics and key dependencies; inputting the channel data into a trained reverse denoising model to generate a predicted optimal channel index, and using the predicted optimal channel index to predict the vehicle network channel in a complex traffic scenario. The present invention adopts a reverse denoising model to effectively solve the channel prediction problem through implicit state evolution, overcomes the limitations of relying on lengthy formula derivation or basic nonlinear models, and can accurately capture the complex nonlinear mapping relationship between channel state and communication channel.

Description

Technical Field

[0001] The present invention relates to the field of vehicle networking communication technology, and specifically to a vehicle networking channel prediction method and related equipment in complex traffic scenarios. Background Art

[0002] As a key driver of smart transportation, vehicle-to-infrastructure (V2I) communication technology significantly enhances the overall intelligence of transportation systems through efficient connectivity and collaboration between vehicles, roads, and the cloud. Vehicle-to-infrastructure (V2I) technology, a core component of the IoV, enables seamless collaboration between vehicles and infrastructure, providing rich perception information and intelligent decision-making support. This optimizes road resource allocation and ensures smooth traffic flow.

[0003] Channel prediction technology plays a crucial role in V2I communications. By predicting the channel state at the next moment in real time, it effectively addresses the challenges posed by high vehicle mobility and meets the stringent requirements of low latency and high reliability for connected vehicle communications. However, despite channel prediction technology becoming a research hotspot and achieving significant progress, existing research still has many shortcomings, as reflected in the following aspects:

[0004] First, while formula-based channel prediction methods perform well in static or low-speed scenarios, their accuracy is severely limited in complex and changing traffic environments. These methods rely on complex mathematical modeling and statistical analysis, but fail to fully account for real-world factors such as lane-changing behavior, multi-vehicle occlusion, and vegetation interference, resulting in insufficient prediction capabilities for communication link stability.

[0005] Secondly, while machine learning-based channel prediction methods simplify the prediction process and improve efficiency, they exhibit certain limitations when dealing with nonlinear mappings in complex scenarios. These methods train models to learn the patterns of channel changes from historical data. However, faced with the complex changes in channel conditions in traffic systems, existing models struggle to fully capture these nonlinear characteristics, thus reducing prediction accuracy.

[0006] Furthermore, deep learning-based channel prediction methods demonstrate high accuracy in some scenarios, particularly adept at handling complex nonlinear relationships. However, existing technologies often rely on simplified V2I communication assumptions and fail to effectively incorporate the diverse influencing factors of real-world scenarios, such as the impact of varying weather conditions and traffic density on communication links. This further limits the applicability of deep learning methods in diverse environments.

[0007] In summary, as traffic scenarios become increasingly dynamic and complex, channel quality fluctuates frequently, severely restricting the reliability and real-time nature of communications. Therefore, it is urgent to design a channel prediction method with greater robustness and environmental adaptability to cope with complex and changing traffic environments and improve the stability and reliability of V2I communications. Summary of the Invention

[0008] In order to overcome the defects of the above-mentioned prior art, the purpose of the present invention is to provide a vehicle network channel prediction method and related equipment in complex traffic scenarios, so as to solve the technical problem of insufficient prediction accuracy in the prior art under high dynamics, multipath effects and occlusion scenarios.

[0009] The present invention is achieved through the following technical solutions:

[0010] In a first aspect, the present invention provides a method for predicting a vehicle network channel in a complex traffic scenario, comprising:

[0011] Constructing an urban road scenario for a vehicle-to-vehicle network channel, simulating and obtaining vehicle-to-infrastructure (V2I) communication results in the urban road scenario, and collecting scenario data through the V2I communication results;

[0012] The scene data is preprocessed and forward-noised, and side information is embedded into the data to obtain channel data with spatiotemporal features and key dependencies;

[0013] The channel data is input into the trained inverse denoising model to generate the predicted optimal channel index, which is used to predict the vehicle network channel in complex traffic scenarios.

[0014] Preferably, the urban road scene includes vehicle density, RSU distribution and weather scene.

[0015] Preferably, the scene data includes an optimal channel index value, communication signal time data, vehicle location information, and environmental information, wherein the environmental information includes road topology, traffic flow, and obstruction distribution.

[0016] Furthermore, the preprocessing of the scene data includes:

[0017] Performing data division and data processing on the scene data;

[0018] The data partitioning process includes:

[0019] The collected optimal channel index values are collected into a data set, and divided into a real known part, a pretended unknown part, and a real unknown part; the real known part and the pretended unknown part are used for model training, and the real unknown part is used for model application;

[0020] The data processing process includes:

[0021] The communication signal time data is aligned, the vehicle location information is cleaned and normalized, and the corresponding environmental information is recorded.

[0022] Furthermore, the forward noisy process is to perform forward noisy processing on the pretended unknown part and the real unknown part to obtain forward noisy data.

[0023] Furthermore, side information embedding is to embed side information into the real known part and the forward noise data, where the side information includes communication signal time data and vehicle location information;

[0024] The specific process of embedding the side information includes:

[0025] Input the true known part and the forward noise data into the convolution layer for preliminary feature extraction, and embed the time information into the preliminary features through the fully connected layer and the convolution layer;

[0026] The time-dependent features in the data are extracted through the time series extraction layer, and the law of channel state changes over time is captured based on the time-dependent features;

[0027] The feature extraction layer extracts time-critical features through multi-layer convolution and feature fusion;

[0028] The vehicle’s location information is embedded into the preliminary features through dilated convolution, and high-order spatial features are extracted;

[0029] The time-critical features and high-order spatial features are fused and input into the gated activation unit to output channel data with spatiotemporal features and key dependencies.

[0030] Furthermore, the channel data is fed into the trained inverse denoising model to generate the predicted optimal channel index, including:

[0031] The channel data after side information embedding is input into the reverse denoising model, the noise is gradually removed, and the predicted optimal channel index of the pretend unknown part is generated. The trained reverse denoising model is obtained by calculating the error between the denoised data and the real data.

[0032] The pure noise data is input into the trained inverse denoising model to generate the predicted optimal channel index of the true unknown part, which is used to indicate the channel state at the next moment.

[0033] In a second aspect, the present invention further provides a vehicle network channel prediction system in complex traffic scenarios, comprising:

[0034] An urban road scene simulation and acquisition module is used to construct an urban road scene for a vehicle network channel, simulate and obtain V2I communication results in the urban road scene, and obtain scene data through the V2I communication results;

[0035] A scene data prediction processing module, configured to perform side information embedding on the scene data after preprocessing and forward noise addition to obtain channel data with spatiotemporal features and key dependencies;

[0036] The channel index prediction module is used to input channel data into the trained inverse denoising model to generate the predicted optimal channel index, and use the predicted optimal channel index to predict the vehicle network channel in complex traffic scenarios.

[0037] In a third aspect, the present invention also provides a mobile terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the vehicle network channel prediction method in a complex traffic scenario as described above are implemented.

[0038] In a fourth aspect, the present invention further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the vehicle network channel prediction method in a complex traffic scenario as described above.

[0039] Compared with the prior art, the present invention has the following beneficial technical effects:

[0040] The present invention provides a method for predicting vehicle network channels in complex traffic scenarios. By constructing an urban road scenario of the vehicle network channel and simulating the V2I communication results in this scenario, scene data that is highly relevant to the complex traffic scenario is effectively obtained. The collected scene data is preprocessed and forward-noised, which can eliminate feature scale differences and improve data quality and model convergence efficiency. Forward noise addition provides a unified and controllable initial distribution condition for the reverse denoising process. By embedding side information, channel data with spatiotemporal characteristics and key dependencies are extracted. In network slicing, the present invention solves the dynamic modeling problem of wireless channels at the access network AN level. By constructing an urban road scenario, simulating V2I communication and optimizing channel prediction, the channel state changes can be more accurately described by predicting the optimal channel index, thereby significantly improving the prediction accuracy. The present invention and access network security authentication are connected front and back, and together serve the integrity of the end-to-end architecture of the slice, ensuring the stability and security of the network slice in the access link.

[0041] Furthermore, the real known portion and the pretended unknown portion are used for model training. The pretended unknown portion simulates data loss scenarios that the model may encounter in real-world applications, while the pretended unknown portion also retains real results to facilitate performance evaluation. During training, the model uses self-supervised learning to deeply analyze and learn from the real results of the real known and pretended unknown portions, gradually mastering the ability to accurately infer unknown information from partially known information. This mechanism not only mitigates the common data loss issues in real-world scenarios but also provides a scientific and reliable foundation for model training and evaluation, significantly improving the model's adaptability and generalization capabilities in complex traffic scenarios.

[0042] Furthermore, the pseudo-unknown portion simulates unknown conditions during model training. After forward noise processing, the data exhibits more diverse patterns of variation. During model training, faced with this noisy pseudo-unknown data, the model needs to deeply explore the data's inherent stability characteristics and underlying patterns to accurately distinguish between noise interference and true channel variation signals. This enables the model to learn more essential channel characteristics, avoiding merely memorizing superficial patterns in the training data. This improves the model's ability to learn complex channel characteristics and more accurately respond to dynamic changes in the channel environment.

[0043] Furthermore, the present invention introduces a side information module, which significantly enhances the model's time series prediction capabilities by embedding time and location information. Traditional channel prediction methods usually ignore the dynamics of time changes and driving behavior, resulting in a significant decrease in their prediction performance in changing traffic environments. By inputting time and location information into the model as side information, the present invention fully exploits the temporal correlation of the channel, enabling the model to adaptively capture the dynamic changes in driving behavior. This strategy significantly improves the model's prediction performance in complex traffic scenarios, enhances the comprehensiveness and accuracy of channel prediction, and performs particularly well in environments with high uncertainty. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of a method for predicting vehicle network channels in a complex traffic scenario according to an embodiment of the present invention;

[0045] Figure 2 This is a schematic diagram of an urban road scene of an Internet of Vehicles channel in an embodiment of the present invention;

[0046] Figure 3 Schematic diagram of a V2I communication prediction algorithm model in an embodiment of the present invention;

[0047] Figure 4 This is a bar chart comparing CRPS values of different experimental schemes under different weather conditions and traffic volumes in the embodiment of the present invention;

[0048] Figure 5 This is a schematic diagram of a vehicle network channel prediction system in a complex traffic scenario according to an embodiment of the present invention;

[0049] In the figure: 1. Urban road scene simulation and acquisition module; 2. Scene data prediction and processing module; 3. Channel index prediction module. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0051] The purpose of the present invention is to provide a vehicle network channel prediction method and related equipment in complex traffic scenarios to solve the technical problem of insufficient prediction accuracy in high dynamic, multipath effect and occlusion scenarios in the prior art.

[0052] The terms involved in this invention are explained as follows:

[0053] RSU: Road Side Unit;

[0054] Nheads: N attention heads;

[0055] Air-Sim: Air-Sim is an open-source, cross-platform simulator based on a game engine. It can be used for physical and visual simulation of robots such as drones and autonomous vehicles.

[0056] WaveFarer: WaveFarer is a high-fidelity radar simulator that accounts for multipath and scattering from structures and vehicles in the radar system's surrounding environment, as well as critical atmospheric and scattering effects at frequencies up to and beyond 100 GHz. Applications include simulating automotive driving scenarios, indoor sensors, and far-field radar cross-section.

[0057] Wireless InSite: Wireless InSite is a prediction tool for understanding wireless coverage, channel multipath, and data throughput for 5G, 6G, and WiFi networks. Using advanced accelerated 3D ray tracing and fast ray-based surrogacy methods, along with empirical models, it efficiently and accurately predicts multipath channel characteristics in indoor, urban, and rural environments. Dynamic scenario modeling of vehicles and pedestrians captures time-varying attenuation, while frequency sweeps account for broadband effects. Communication analysis capabilities apply multi-input, multi-output algorithms to channel prediction to estimate wireless network coverage and throughput.

[0058] The present invention is described in further detail below with reference to the accompanying drawings:

[0059] Example 1

[0060] See also Figure 1 In one embodiment of the present invention, a method for predicting a vehicle network channel in a complex traffic scenario is provided, comprising:

[0061] Step 1: construct an urban road scenario for a vehicle-to-vehicle network channel, simulate and obtain a V2I communication result in the urban road scenario, and obtain scenario data through the V2I communication result collection;

[0062] Specifically, the urban road scene includes vehicle density, RSU distribution and weather scene.

[0063] The scene data includes the optimal channel index value, communication signal time data, vehicle location information and environmental information, wherein the environmental information includes road topology, traffic flow and obstruction distribution.

[0064] Among them, according to Figure 2 As shown in the figure, in the medium-density scenario, the scene contains 10 cars, 3 buses, and 11 RSUs. In the high-density scenario, the number of vehicles increases to 15 cars and 6 buses, while the number of RSUs remains at 11. Weather scenarios include sunny, rainy, and snowy days. Raw data is collected for 1000 time frames for each scenario, including intact time frames (75%) and lost data (25%). Each intact time frame contains the top three channel index values of the V2I communication signal-to-noise ratio, the communication signal time frame number, and the vehicle and RSU location coordinates. Each lost time frame only contains the communication signal time frame number and the vehicle and RSU location coordinates.

[0065] Step 2: Preprocessing and forward-noising the scene data and then embedding side information to obtain channel data with spatiotemporal features and key dependencies;

[0066] Specifically, according to Figure 3 As shown, the preprocessing of the scene data includes:

[0067] Performing data division and data processing on the scene data;

[0068] The data partitioning process includes:

[0069] The collected best channel index values are collected as a data set and divided into the real known part , pretend to be unknown part and the real unknown part; the real known part and pretend unknown parts It is used for model training, and the real unknown part is used for model application;

[0070] The data processing process includes:

[0071] The communication signal time data is aligned, the vehicle location information is cleaned and normalized, and the corresponding environmental information is recorded.

[0072] Among them, the non-lost time frame is randomly divided into real known parts and pretend unknown parts , the lost time frame is the real unknown part, the real known part For training (80%) and validation (10%), pretending to be unknown Used for testing (10%), the missing time frames are used to apply the trained denoising model. The communication signal time frames are aligned, and the vehicle position coordinates are aligned with the RSU position coordinates and normalized.

[0073] Specifically, according to Figure 3 As shown, the forward noise is to pretend that the unknown part The forward noise addition process is performed on the real unknown part to obtain the forward noise addition data.

[0074] Among them, the unknown part after preprocessing Perform 50 forward noise addition processes to obtain forward noise data. The real known part No noise addition is performed.

[0075] The expression of forward noise processing is as follows:

[0076]

[0077] in, is a function of the noise intensity constant, .

[0078] The noise intensity is interpolated from 0.0001 to 0.5 at each time step according to the law of a quadratic function. The current state depends only on the state at the previous moment, forming a Markov chain process on the time series. Through multiple steps of noise addition, the channel state data gradually transforms from the true distribution to the noise-perturbed distribution.

[0079] Step 3: Input the channel data into the trained reverse denoising model to generate the predicted optimal channel index, and use the predicted optimal channel index to predict the vehicle network channel in complex traffic scenarios.

[0080] Specifically, according to Figure 3 As shown, the side information embedding is for the real known part and forward noise data Embed side information, where the side information includes communication signal time data and vehicle location information. The specific process of embedding side information is as follows:

[0081] First, the real known part and forward noise data Input to the convolution layer for preliminary feature extraction, and the convolution kernel size is 1.

[0082] Then, the temporal information is embedded into the preliminary features with 128 dimensions through a fully connected layer and a convolutional layer.

[0083] The time series extraction layer then further extracts time-dependent features from the data, capturing patterns in channel state changes over time. The time series extraction layer uses a multi-head attention mechanism (nheads parameter is 8). The feature extraction layer extracts time-critical features relevant to channel prediction through multi-layer convolution and feature fusion. Vehicle location information is embedded into the preliminary features in 16 dimensions through dilated convolution operations, further extracting high-order spatial features from the data. Feature fusion combines temporal and spatial features to produce fused data.

[0084] Finally, the gated activation unit is used to control the flow and activation of features, retaining the most important features for channel prediction and suppressing irrelevant or redundant information. The fused data contains rich spatiotemporal features and key dependencies, providing high-quality input for subsequent denoising and channel prediction.

[0085] Specifically, the channel data after the side information is embedded is subjected to reverse denoising through the denoising function. The denoising function learns to recover the original signal from the noisy data, and the total number of output channels is 128. The channel data after the side information is embedded is input into the denoising model, and the noise is gradually removed 50 times to generate the denoised data. The number of output channels is 1, and the denoised data is calculated. With real data The error between them is used to optimize the denoising model.

[0086] Specifically, based on the trained denoising model, the optimal channel index is generated from pure noise data for the actual unknown location. The pure noise data has a shape of (16, 3, 1000), where 16 is the batch size, 3 is the index of the top three channels in terms of signal-to-noise ratio, and 1000 is the time frame length. The pure noise data is input into the trained denoising model to generate a channel index, which indicates the channel direction at the next moment. The channel index specifically divides the 180-degree plane horizontal to the RSU and the road into 128 evenly spaced sections, with the channel index representing one of these directions. For example, index values of 0 and 127 represent the leftmost and rightmost directions, respectively. The channel index accurately indicates the communication direction, further improving the performance of the communication system.

[0087] In order to verify the beneficial effect of the proposed inverse denoising model in the channel prediction task, based on the public dataset A series of simulation experiments were conducted to compare the present invention with the Support Vector Machine (SVM), a typical representative algorithm in machine learning, and the Long Short-Term Memory (LSTM), a typical representative algorithm in deep learning.

[0088] Data source: The dataset uses Air-Sim and WaveFarer to collect multimodal perception data, and Wireless InSite to collect communication data. The dataset realizes a communication perception integrated system by deeply integrating and precisely aligning Air-Sim, WaveFarer, and WirelessInSite. The dataset also covers various weather conditions, multiplex frequency bands, and times of the day.

[0089] Hardware environment and software platform: The hardware environment uses Intel Xeon W-3235 12-core processor and RTX 3090 GPU, and the programming language uses Python 3.9.

[0090] To ensure the reliability of the experimental results and the scientific nature of the model evaluation, the dataset was divided using five-fold cross-validation. In each fold, the non-missing data was randomly divided into a training set (80% × 70%), a validation set (80% × 30%), and a test set (20%). For each sample, 100 probability points were generated, and the median was taken as the final prediction result to enhance the stability and accuracy of the prediction, effectively reducing the impact of outliers and ensuring that the model can provide more reliable predictions when faced with uncertainty or extreme data. Considering the potential for real missing values in actual channel state data, the experiment introduced a self-supervised learning mechanism. By setting the missing probability to 0.1 and 0.3, the impact of different degrees of data missingness on model performance was simulated, thereby more comprehensively evaluating the robustness of the model.

[0091] To simulate real-world connected vehicle scenarios, different vehicle densities and RSU distributions were set to reflect typical urban road communication environments. In the medium-density scenario, the vehicle population consisted of 10 cars, 3 buses, and 11 RSUs. In the high-density scenario, the vehicle population increased to 15 cars and 6 buses, while maintaining the same 11 RSUs. These two density scenarios enabled a comprehensive evaluation of the model's adaptability to varying traffic conditions. Furthermore, to simulate realistic road conditions, two weather scenarios, rainy and snowy, were introduced. In the rainy scenario, the rainfall was set to 50 mm / hr and the road humidity was 100%. In the snowy scenario, the snowfall was set to 10 mm / hr and the road humidity was also 100%. Furthermore, to more accurately reflect the impact of inclement weather on channel propagation, the experiment also modeled the physical and electromagnetic space parameters under different weather conditions. For example, in the rainy scenario, the ambient temperature was set to 22.2°C and the humidity was 100%. In the snowy scenario, the temperature was set to -10°C and the humidity was 20%.

[0092] In the communication parameter setting, the experiment adopted an integrated design of communication and perception to ensure the accuracy of channel prediction and the efficiency of model training. Specifically, each RSU and vehicle is equipped with 128 and 32 antenna units, respectively, to support high-precision beamforming and channel prediction. In terms of frequency band selection, the experiment simulated the communication performance of the Sub-6 GHz band and the millimeter wave band at the same time. In the Sub-6 GHz band, the carrier frequency is set to 5.9 GHz and the communication bandwidth is 20 MHz; in the millimeter wave band, the carrier frequency is set to 28 GHz and the communication bandwidth is 2 GHz. Through a comprehensive comparison of the communication performance of different frequency bands, the experiment further verified the adaptability and performance of the model in a multi-band environment.

[0093] The simulation experiment parameter settings are summarized in Table 1:

[0094] Table 1 Parameter settings for channel prediction simulation experiment based on the reverse denoising model

[0095]

[0096] according to Figure 4As shown in Figure 2, under medium traffic density on a sunny day, the continuous distribution score of CSDM (Continuous Ranked Probability The CRPS of the CSDM is 0.184, which is significantly lower than 0.675 of SVM and 0.445 of LSTM, and has increased by about 72.8% and 58.7% respectively. Under high-density traffic on sunny days, CSDM still performs outstandingly, with a CRPS value of 0.232, which is about 66.2% and 48.5% higher than SVM (0.683) and LSTM (0.448), respectively. Under medium traffic density on rainy days, the CRPS of CSDM is 0.258, which is significantly better than SVM (0.689) and LSTM (0.490), and has increased by about 62.5% and 47.4% respectively. Under medium traffic density on snowy days, the CRPS of CSDM is 0.392, which is still lower than SVM (0.681) and LSTM (0.535), and has increased by about 42.5% and 26.7% respectively. These results show that CSDM can effectively improve the accuracy of channel prediction, outperforming SVM and LSTM methods, regardless of sunny, rainy or snowy environments.

[0097] The time performance of a model is an important indicator of its practicality. Real-time performance is a core requirement of channel prediction models, especially in connected vehicle scenarios. To this end, the experiment recorded the training time for a single sample and the prediction time for a single channel. The model's time performance was quantitatively analyzed based on the size and type of input data. The results are shown in Table 2. The experimental results show that the model's input data includes a timestamp, the historical best channel index, and location information, with each piece of data sized at 25 bytes. During the training phase, the training time for a single sample was 2.411 ms, demonstrating that the model maintains high efficiency when training each piece of data. During the inference phase, the prediction time for a single channel was only 0.098 ms. The model's ability to complete predictions in a short time demonstrates its high computational efficiency and application potential. This result demonstrates that the proposed CSDM-based channel prediction model can achieve both efficient training and rapid real-time channel prediction, laying a solid foundation for practical deployment.

[0098] Table 2 Time performance of the channel prediction model based on the inverse denoising model

[0099]

[0100] In summary, the present invention provides a method for predicting IoV channels in complex traffic scenarios. By incorporating historically known channel direction indices as conditional inputs, it can more efficiently capture the nonlinear relationship between channel states and spatiotemporal factors, overcoming the limitations of traditional methods in highly dynamic scenarios. Specifically, the channel prediction problem can be transformed into a time series prediction problem. This involves inferring the channel direction indices for future moments by learning the implicit patterns and spatiotemporal variations in historical time series data.

[0101] The introduction of the side information module in the present invention provides the reverse denoising model with more accurate and rich additional information, especially showing significant advantages in mining time correlation and dynamic driving behavior. First, the time information embedding integrates information such as timestamps and time intervals into the model through the fully connected layer and the convolutional layer, which enhances the modeling ability of time dependency and enables the model to more accurately capture the law of channel state changes over time. Secondly, the spatial information embedding integrates spatial information such as vehicle position and road topology into the model through the convolutional layer and the expansion layer, which improves the adaptability of the model to dynamic driving behavior. Finally, feature fusion integrates the time features and spatial features and then passes through the gated activation unit to generate high-quality side information embedding data, providing more comprehensive additional information for channel prediction.

[0102] The present invention demonstrates excellent robustness across multiple scenarios, primarily due to the model's precise prediction of nonlinear relationships and the selection of embedded information in the side modules. In highly dynamic scenarios, the inverse denoising model can capture rapid changes in channel states, ensuring the stability of prediction results. Timestamps accurately record the temporal evolution of channel states, providing the model with critical temporal information. Compared to other time-related information (such as time intervals or time windows), timestamps more directly reflect the specific time point of the channel state, thereby more effectively capturing the transient characteristics of channel state changes. Vehicle position information directly reflects the dynamic changes of the vehicle in space, providing the model with key spatial characteristics. Weather-related environmental factors may interfere with the performance of certain sensors or the collected raw data. However, vehicle position information, which relies primarily on satellite signal transmission, has high environmental adaptability and maintains stable performance under most weather conditions (including rain and snow). Furthermore, by combining vehicle position and time data, the model can further map other important spatially relevant information, such as vehicle speed, acceleration, and direction of travel.

[0103] This paper uses an inverse denoising model to effectively solve the channel prediction problem through implicit state evolution. This overcomes the limitations of relying on lengthy formula derivations or basic nonlinear models and accurately captures the complex nonlinear mapping relationship between the channel state and the communication channel. The inverse denoising model offers significant advantages in dynamic modeling, enabling more accurate simulation of channel state changes based on input conditions, significantly improving prediction accuracy.

[0104] This paper systematically experiments the proposed channel prediction method under various weather conditions and traffic densities for the first time. Traditional channel prediction models are typically tested in idealized environments, ignoring the variability of real-world application scenarios, resulting in significantly reduced performance in complex environments. This paper designed an experimental environment encompassing different weather conditions (e.g., sunny, rainy, snowy) and traffic densities (e.g., medium density, high density), comprehensively validating the model's robustness and stability under various real-world conditions. The experimental results demonstrate that the proposed method maintains excellent prediction performance across multiple scenarios, fully demonstrating its broad adaptability and reliability in complex real-world environments.

[0105] Example 2

[0106] according to Figure 5 As shown, the present invention also provides a vehicle network channel prediction system in complex traffic scenarios, including:

[0107] Urban road scene simulation and acquisition module 1, used to construct an urban road scene for a vehicle network channel, simulate and obtain V2I communication results in the urban road scene, and obtain scene data through the V2I communication results;

[0108] A scene data prediction processing module 2 is configured to perform side information embedding on the scene data after preprocessing and forward noise addition to obtain channel data with spatiotemporal features and key dependencies;

[0109] The channel index prediction module 3 is used to input the channel data into the trained reverse denoising model to generate the predicted optimal channel index, and use the predicted optimal channel index to predict the vehicle network channel in complex traffic scenarios.

[0110] Example 3

[0111] The present invention also provides a mobile terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, such as a vehicle network channel prediction program in complex traffic scenarios.

[0112] When the processor executes the computer program, the steps of the method for predicting a vehicle network channel in a complex traffic scenario are implemented, for example:

[0113] Constructing an urban road scenario for a vehicle-to-vehicle network channel, simulating and obtaining vehicle-to-infrastructure (V2I) communication results in the urban road scenario, and collecting scenario data through the V2I communication results;

[0114] The scene data is preprocessed and forward-noised, and side information is embedded into the data to obtain channel data with spatiotemporal features and key dependencies;

[0115] The channel data is input into the trained inverse denoising model to generate the predicted optimal channel index, which is used to predict the vehicle network channel in complex traffic scenarios.

[0116] Alternatively, when the processor executes the computer program, the functions of each module in the above system are realized, for example:

[0117] Urban road scene simulation and acquisition module 1, used to construct an urban road scene for a vehicle network channel, simulate and obtain V2I communication results in the urban road scene, and obtain scene data through the V2I communication results;

[0118] A scene data prediction processing module 2 is configured to perform side information embedding on the scene data after preprocessing and forward noise addition to obtain channel data with spatiotemporal features and key dependencies;

[0119] The channel index prediction module 3 is used to input the channel data into the trained reverse denoising model to generate the predicted optimal channel index, and use the predicted optimal channel index to predict the vehicle network channel in complex traffic scenarios.

[0120] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program in the mobile terminal.

[0121] For example, the computer program may be divided into an urban road scene simulation acquisition module 1, a scene data prediction processing module 2, and a channel index prediction module 3;

[0122] The specific functions of each module are as follows:

[0123] Urban road scene simulation and acquisition module 1, used to construct an urban road scene for a vehicle network channel, simulate and obtain V2I communication results in the urban road scene, and obtain scene data through the V2I communication results;

[0124] A scene data prediction processing module 2 is configured to perform side information embedding on the scene data after preprocessing and forward noise addition to obtain channel data with spatiotemporal features and key dependencies;

[0125] The channel index prediction module 3 is used to input the channel data into the trained reverse denoising model to generate the predicted optimal channel index, and use the predicted optimal channel index to predict the vehicle network channel in complex traffic scenarios.

[0126] The mobile terminal may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The mobile terminal may include, but is not limited to, a processor and a memory.

[0127] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the mobile terminal and connects various parts of the entire mobile terminal using various interfaces and lines.

[0128] The memory may be used to store the computer programs and / or modules, and the processor implements various functions of the mobile terminal by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory.

[0129] The memory may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as sound playback or image playback); the data storage area may store data generated based on the use of the mobile phone (such as audio data and a phone book). Furthermore, the memory may include high-speed random access memory (RAM) and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0130] Example 4

[0131] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the vehicle network channel prediction method in a complex traffic scenario.

[0132] If the module / unit integrated in the mobile terminal is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0133] Based on this understanding, the present invention can implement all or part of the processes in the above-mentioned method by means of a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of the above-mentioned method for scheduling aggregated reinforcement learning resources. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form.

[0134] The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0135] It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practices in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practices, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0136] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the present invention.

Claims

1. A method for predicting vehicle network channels in complex traffic scenarios, characterized by: include: Constructing an urban road scenario for a vehicle-to-vehicle network channel, simulating and obtaining vehicle-to-infrastructure (V2I) communication results in the urban road scenario, and collecting scenario data through the V2I communication results; The scene data includes the optimal channel index value, communication signal time data, vehicle location information and environmental information, wherein the environmental information includes road topology, traffic flow and obstruction distribution; The scene data is preprocessed and forward-noised, and side information is embedded into the data to obtain channel data with spatiotemporal features and key dependencies; The preprocessing of the scene data includes: Performing data division and data processing on the scene data; The data partitioning process includes: The collected optimal channel index values are collected into a data set, and divided into a real known part, a pretended unknown part, and a real unknown part; the real known part and the pretended unknown part are used for model training, and the real unknown part is used for model application; The data processing process includes: Align the communication signal time data, clean and normalize the vehicle's location information, and record the corresponding environmental information; The forward noising is to perform forward noising processing on the pretended unknown part and the real unknown part to obtain forward noisy data; The side information embedding is to embed side information into the real known part and the forward noise data, wherein the side information includes communication signal time data and vehicle location information; The specific process of embedding the side information includes: Input the true known part and the forward noise data into the convolution layer for preliminary feature extraction, and embed the time information into the preliminary features through the fully connected layer and the convolution layer; The time-dependent features in the data are extracted through the time series extraction layer, and the law of channel state changes over time is captured based on the time-dependent features; The feature extraction layer extracts time-critical features through multi-layer convolution and feature fusion; The vehicle’s location information is embedded into the preliminary features through dilated convolution, and high-order spatial features are extracted; The time-critical features and high-order spatial features are fused and input into the gated activation unit to output channel data with spatiotemporal features and key dependencies; Input the channel data into the trained inverse denoising model to generate the predicted optimal channel index, and use the predicted optimal channel index to predict the IoV channel in complex traffic scenarios. Inputting the channel data into the trained inverse denoising model to generate a predicted optimal channel index includes: The channel data after side information embedding is input into the reverse denoising model, the noise is gradually removed, and the predicted optimal channel index of the pretend unknown part is generated. The trained reverse denoising model is obtained by calculating the error between the denoised data and the real data. The pure noise data is input into the trained inverse denoising model to generate the predicted optimal channel index of the true unknown part, which is used to indicate the channel state at the next moment.

2. The method for predicting vehicle network channels in complex traffic scenarios according to claim 1, characterized in that: The urban road scene includes vehicle density, RSU distribution and weather scene.

3. A vehicle network channel prediction system in complex traffic scenarios, characterized by: A method for predicting a vehicle network channel in a complex traffic scenario according to any one of claims 1 to 2, comprising: An urban road scene simulation and acquisition module is used to construct an urban road scene for a vehicle network channel, simulate and obtain V2I communication results in the urban road scene, and obtain scene data through the V2I communication results; A scene data prediction processing module, configured to perform side information embedding on the scene data after preprocessing and forward noise addition to obtain channel data with spatiotemporal features and key dependencies; The channel index prediction module is used to input channel data into the trained inverse denoising model to generate the predicted optimal channel index, and use the predicted optimal channel index to predict the vehicle network channel in complex traffic scenarios.

4. A mobile terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the vehicle network channel prediction method in a complex traffic scenario as described in any one of claims 1-2 are implemented.

5. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the vehicle network channel prediction method in a complex traffic scenario as described in any one of claims 1-2 are implemented.

Images