Inter-vehicle communication system signal-to-noise ratio intelligent monitoring method for automobile intelligent network connection

Abstract

The invention discloses an inter-vehicle communication system signal-to-noise ratio intelligent monitoring method for an automobile intelligent network, and the method comprises the steps: sequentially carrying out the operation of constellation diagram coordinate axis enhancement, quadrant normalization, constellation point diameter selection and the like on an ACO-OFDM signal carrying in-vehicle sensor data under the receiving of a vehicle-mounted sensor, and carrying out the detection of the signal-to-noise ratio of the ACO-OFDM signal. According to the method, image features such as density and position of the enhanced constellation diagram are improved, and intelligent estimation of the received signal-to-noise ratio of the inter-vehicle communication system corresponding to the enhanced constellation diagram is realized by adopting a classical ResNet-18 deep learning network. A simulation result shows that the intelligent signal-to-noise ratio monitoring method provided by the invention can improve the transmission quality of automobile data network connection in a complex driving road condition channel compared with a traditional scheme without constellation diagram enhancement.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent automobile networking, and specifically relates to an intelligent monitoring method for optical communication transmission quality based on deep learning. Background Art

[0002] With the rapid development of China's manufacturing and Industry 4.0, new energy vehicles, as an important alternative to traditional fuel vehicles, have attracted widespread attention from scholars and companies around the world. As an important technology for new energy vehicles, intelligent vehicle networking has ushered in rapid development. Among them, the most representative vehicle networking technology will realize high-speed interconnection between vehicles and ensure driving safety. It has broad application prospects in the fields of intelligent driving, intelligent transportation systems, smart cities, etc. The vehicle networking environment is a combination of wireless network environment and road conditions. It has the ability to communicate with the inside and outside environment of the vehicle. The interaction of the vehicle networking includes vehicle-to-vehicle communication, vehicle-to-Internet communication and vehicle-to-infrastructure communication. The vehicle networking is a dynamic mobile communication network with functions such as data collection, data exchange, and data processing. In order to better upload the massive data generated by in-vehicle sensors such as vehicle computers, on-board cameras and smart cockpits in real time, and provide support for intelligent driving and vehicle intelligence, high-speed optical communication technology based on vehicle-mounted LED headlights and on-board photoelectric detectors has become a potential means of in-vehicle and vehicle-to-vehicle communication (V2V). In order to ensure the rate and stability requirements of vehicle data networking, real-time monitoring of communication quality for communication reliability has become extremely necessary. Most of the current research on communication quality monitoring methods based on deep learning focuses on deep learning models and massive data acquisition, and has not yet considered the intrinsic physical characteristics and data structure of communication data. To improve this problem, we propose an intelligent signal-to-noise ratio estimation scheme based on the symmetry of the constellation diagram, adding coordinate position information to the constellation diagram, and changing the diameter of the constellation points. This scheme can achieve fast and accurate communication quality monitoring of the Internet of Vehicles using small batches of received data. Summary of the invention

[0003] In order to solve the above problems, the present invention proposes an intelligent signal-to-noise ratio estimation method for asymmetrically clipped optical orthogonal frequency division multiplexing (ACO-OFDM) modulation:

[0004] An intelligent monitoring method for signal-to-noise ratio of an ACO-OFDM communication system for automobile data networking, comprising:

[0005] 1. Data Processing in ACO-OFDM System

[0006] 1. Input terminal

[0007] 1) Massive data collected in real time by in-vehicle sensors, such as road conditions, vehicle status information, and driver information in the smart cockpit;

[0008] 2) Selecting the ACO-OFDM system based on QAM modulation for data modulation, calculating the power normalization factor according to the QAM order, performing QAM mapping on the collected data in step 1), and dividing the QAM mapped signal by the power normalization factor;

[0009] 3) Perform Hermitian conjugation operation on the complex signal mapped by QAM. The ACO-OFDM signal uses odd subcarriers to transmit data, and the even subcarriers are set to zero.

[0010] , n=0, 1, …, N-1;

[0011] in, Represents the frequency domain data symbol modulated onto the nth subcarrier after mapping; the signal is converted into a bipolar real-valued OFDM signal through serial-to-parallel conversion and inverse fast Fourier transform;

[0012] Then, negative clipping operation is performed, that is, the signal amplitude greater than zero in the time domain x is retained, and the signal amplitude less than zero in x is set to zero; then the signal is sent through the LED drive circuit to control the car headlights to send optical signals carrying the sensor data in the car, and through the complex time-varying driving road condition channel, the optical signal is transmitted to the photoelectric sensors of other vehicles on the road for real-time reception;

[0013]

[0014] Among them, y(n) is the optical receiving signal carrying the in-vehicle sensor data, h(n) is the channel response of the driving road conditions, x(n) is the transmitted signal carrying the in-vehicle sensor data in the time domain, and w(n) represents the noise introduced in the channel. Represents the convolution operation;

[0015] 2. Receiving End

[0016] 1) Other vehicles on the road receive signals y(n) in real time through on-board photoelectric detectors, and perform fast Fourier transform on them to obtain:

[0017] ;

[0018] Convert the received signal from time domain to frequency domain; frequency domain signal , extract the odd subcarriers to get ;

[0019] 2) Perform parallel-to-serial conversion on the odd-numbered subcarrier QAM signal containing valid data, and perform QAM demapping to obtain its complex signal;

[0020] 3) Draw a two-dimensional QAM constellation diagram, enhance the coordinate axis information and perform quadrant normalization operations on the original constellation diagram

[0021] That is, add horizontal and vertical axes to the drawn constellation diagram; then add the complex Take the absolute values ​​of the real and imaginary parts of respectively, and normalize the complex signal demodulated by QAM to the first quadrant;

[0022] 4) Selection of the constellation point drawing diameter

[0023] According to the number of received signal samples, different diameters of constellation points are selected, including 5, 10 and 20, to avoid overlapping of high-density signals in the constellation diagram, so that the final enhanced constellation diagram containing coordinate axis position information, quadrant normalization and constellation point diameter has richer density and position features;

[0024] 5) Use the on-board computer to estimate the signal-to-noise ratio of the received signal to obtain real-time road condition information for inter-vehicle communication. Based on the estimated signal-to-noise ratio at the receiving end, the LED drive circuit module is fed back to adjust the instantaneous power of the vehicle headlights, thereby improving the optical signal quality of inter-vehicle communication in real time and transmitting road condition information back to other vehicles on the road.

[0025] 2. Constructing a Constellation Map Dataset Enhanced by Deep Learning

[0026] (1) For QAM modulation, based on the bit error rate curve of the ACO-OFDM system, the signal-to-noise ratio range is selected as [1-40] dB, with an interval of 1 dB.

[0027] (2) The number of received signals was selected as 32, 64, 128, 256, and 512. Then, 100 enhanced constellation diagrams were randomly generated for each number of received signals with the same signal-to-noise ratio. The enhancement method was the superposition of coordinate axis position information, quadrant normalization, and constellation point diameter selection. For the same order of QAM modulation, a total of 500 constellation diagram data with the same signal-to-noise ratio were generated. Considering all 40 signal-to-noise ratio intervals, a total of 20,000 constellation diagram data sets were generated for the same order of QAM modulation.

[0028] (3) Of these, 80% is used for training and 20% is used for testing and validation.

[0029] 3. Deep Learning Model

[0030] The ResNet-18 framework is used to conduct supervised training on the constellation diagram containing the signal-to-noise ratio label to estimate the signal-to-noise ratio, a communication quality indicator.

[0031] The ResNet-18 framework is a ResNet18 residual network, with a total of 18 trainable convolutional layers in the input and output layers; the convolutional layer is followed by a 3×3 maximum pooling layer with a stride of 2. After the last residual block, a global average pooling layer is used to convert the feature map into a fixed-length vector. The output of the global average pooling (GAP) layer is fed into a fully connected layer, and the output of the fully connected layer is fed into the Softmax layer.

[0032] The power normalization factor calculation formula is as follows:

[0033]

[0034] Where M represents the QAM modulation order, and Respectively represent the real part and imaginary part of the i-th signal after QAM modulation.

[0035] Make the horizontal axis coordinate of each constellation point of the QAM modulation constellation diagram under the M-ary system fall within [ ] within the range of ].

[0036] The coordinate positioning is to first find the center point, i.e., the coordinate origin, in the QAM constellation diagram, and then draw mutually orthogonal horizontal and vertical coordinate axes, i.e., two straight lines without arrows, x=0 and y=0, according to the coordinate origin;

[0037] Further quadrant normalization is performed, that is, the four quadrants of the original constellation diagram are normalized to the first quadrant:

[0038] The constellation image quadrant normalization operation is implemented as follows: the real part and the imaginary part of the complex signal demodulated by QAM are both subjected to absolute value operation, so that the complex constellation points originally belonging to the second, third and fourth quadrants are redrawn into the first quadrant;

[0039] Under the same number of signal points, if the number of constellation points in the constellation diagram is small, such as 32, the diameter of the constellation diagram should be 20; conversely, if the number of constellation points in the constellation diagram is large, such as 512, the diameter of the constellation diagram should be 5 to reduce the occlusion between the constellation points.

[0040] The present invention provides an intelligent monitoring method for the signal-to-noise ratio of an inter-vehicle communication system for intelligent automobile networking. By receiving the ACO-OFDM signal carrying the in-vehicle sensor data, the operations such as constellation coordinate axis enhancement, quadrant normalization and constellation point diameter selection are sequentially performed to improve the density, position and other image features of the enhanced constellation, and the classic ResNet-18 deep learning network is used to realize the intelligent estimation of the received signal-to-noise ratio of the inter-vehicle communication system corresponding to the enhanced constellation. The simulation results show that under complex driving road conditions, compared with the traditional scheme without constellation enhancement, the intelligent signal-to-noise ratio monitoring method proposed in the present invention can improve the transmission quality of automobile data networking. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flowchart illustrating the system model of the present invention;

[0042] Figure 2 It is a data enhancement display diagram of the present invention;

[0043] Figure 3 It is a quadrant normalization and coordinate information enhancement result diagram of the present invention;

[0044] Figure 4 It is a graph showing the results of the density-diameter selection method of the present invention. DETAILED DESCRIPTION

[0045] Embodiment 1

[0046] Figure 1 This is a flow chart of the intelligent signal-to-noise ratio (SNR) estimation method for the V2V vehicle networking environment. A moving vehicle will generate a large amount of sensor data in real time, including air conditioning, camera, infrared lidar and other data. These data are carried by ACO-OFDM signals with high spectrum efficiency, and the lights are controlled by the on-board LED driver module to transmit the data to the receiving vehicle in the form of optical signals. The on-board photodetector carried by the receiving vehicle can convert the received optical signal into an electrical signal, and send it to the intelligent SNR estimation module for processing and adjusting the instantaneous power setting of the lights. At the same time, it is fed back to the sending vehicle using a two-way communication network. The following is the implementation process of the intelligent SNR estimation module:

[0047] 1. ACO-OFDM signal constellation diagram construction method:

[0048] (1) Sending end:

[0049] Calculate the power normalization factor based on the selected QAM order. Perform QAM mapping on the vehicle data. After QAM mapping, the transmission data is converted from real numbers to complex numbers. Divide the signal at this time by the power normalization factor. Perform Hermitian conjugation operation on the complex signal after QAM mapping to make the transmitted data into a Hermitian symmetric structure. Generally, ACO-OFDM uses odd subcarriers to transmit data, and even subcarriers are set to zero.

[0050] , n=0, 1, …, N-1;

[0051] in, Represents the frequency domain data symbol modulated onto the nth subcarrier after mapping. The signal is converted into a bipolar real-valued OFDM signal through serial-to-parallel conversion and inverse fast Fourier transform. Then a negative clipping operation is performed, that is, the signal amplitude greater than zero in the time domain x is retained, and the signal amplitude less than zero in x is set to zero. The transmitted signal is then transmitted to the receiving vehicle through a complex time-varying road channel.

[0052]

[0053] Among them, y(n) is the received signal after passing through the channel, h(n) is the discrete impulse response of the transmission channel, x(n) is the time domain transmitted signal, and w(n) represents the noise introduced in the channel. Represents a convolution operation.

[0054] (2) Receiving end:

[0055] The signal y(n) is received by the on-board photoelectric detector and subjected to fast Fourier transform to obtain:

[0056] ;

[0057] Convert the received signal from time domain to frequency domain. , extract the odd subcarriers to get , then the odd subcarrier QAM signal containing valid data is converted into serial and parallel, and the QAM demapping is performed to obtain the complex number. Then, according to the central symmetry of the constellation diagram, the horizontal and vertical axes are added to the drawn constellation diagram. The real and imaginary parts of are taken as absolute values ​​respectively, and the complex signal demodulated by QAM is normalized to the first quadrant. Finally, according to the number of received signal samples, the appropriate constellation point diameter is selected so that the normalized constellation diagram containing the coordinate axis is finally drawn with rich density and position information.

[0058] (3) Deep learning model:

[0059] The classic ResNet-18 framework in convolutional neural networks is used to perform supervised training on constellations containing signal-to-noise ratio labels. To meet the low latency requirements of high-speed communications, a lightweight ResNet18 residual network is used, with a total of 18 trainable convolutional layers in the input and output layers. The convolutional layer is followed by a 3×3 maximum pooling layer with a stride of 2 to reduce the size of the feature map. After the last residual block, a global average pooling (GAP) layer is used to convert the feature map into a fixed-length vector, usually 512 dimensions. The output of the GAP layer is fed into a fully connected layer with the number of nodes equal to the number of categories of the classification task. The output of the fully connected layer is fed into the Softmax layer to generate a probability prediction for each category. The classification of the present invention is the predicted signal-to-noise ratio value range of the ACO-OFDM system [1, 40] dB, where the classification step is 1 dB. In the constellation data set containing signal-to-noise ratio labels, 80% of the constellation data is used for training, and 20% of the constellation data is used for testing and verification. The hardware environment for deep learning model training is: GPU is a single RTX4090 graphics card with 24G video memory, Epoch is 80, Batch size is 300, and learning rate is .

[0060] 2. The data enhancement method proposed in the present invention:

[0061] 1) Power normalization method:

[0062] By using the normalization factor (1), the horizontal axis coordinate of each constellation point in the M-ary QAM modulation constellation diagram falls within [ ], and does not change with the modulation order M, to meet the fixed-size image input requirements of the deep learning model. If power normalization is not performed, taking 16QAM as an example, the coordinate range of the constellation point without noise is [ , and the constellation points after the proposed power normalization are approximately in the range of [ , all did not exceed the scheduled [ ] range. Power normalization ensures that the coordinate positions of the constellation points of the QAM signal constellation diagrams under different orders M do not exceed the upper limit of 1.5, which well fixes the size of the drawn constellation diagram. Generating constellations of the same scale can reduce the impact of model prediction caused by input image scaling, and also help improve the stability of the training process, accelerate convergence, and ensure that the model focuses more on the image content rather than the border size, thereby improving the prediction accuracy. The power normalization factor calculation formula is as follows:

[0063] (1);

[0064] Where M represents the QAM modulation order, and Respectively represent the real part and imaginary part of the i-th signal after QAM modulation.

[0065] 2) Coordinate axis enhancement method:

[0066] The coordinate axis enhancement method of the present invention is specifically to first find the center point, i.e., the coordinate origin, in the QAM constellation diagram according to the symmetry of its four quadrants, and then draw mutually orthogonal horizontal and vertical coordinate axes according to the coordinate origin, i.e. Two straight lines without arrows. This operation can reveal the implicit position information in the constellation diagram and provide distance features for the signal-to-noise ratio estimation scheme based on the constellation diagram. The increase in feature dimensions helps the model learn the complexity and details of the data. For constellations without coordinate axes, deep learning models will not be able to fully capture the distribution pattern or category boundaries of the data. The proposed constellation diagram with coordinate position information provides more image feature information, allowing the model to learn more information that is more helpful for classification. At the same time, adding coordinate axes can change the distribution characteristics of the data, making the boundaries between different categories more obvious. For deep neural networks, the added features can make it easier for the model to capture the structure and pattern of the data during training. This improves the estimation performance of the signal-to-noise ratio estimation algorithm.

[0067] 3) Quadrant normalization method:

[0068] The constellation diagram with coordinate axes is further normalized by quadrants. Specifically, the real and imaginary parts of the demodulated complex signal are both subjected to absolute value operations, so that the complex constellation points originally belonging to the second, third and fourth quadrants are redrawn into the first quadrant. This step can further enrich the density distribution feature information of the constellation diagram under the premise of receiving data in small batches.

[0069] 4) Selection of constellation point size:

[0070] Given that the batch size of the received data has an impact on the density distribution of the drawn constellation diagram. This scheme also studies the impact of different constellation point diameters on the accuracy of signal-to-noise ratio estimation. Under the same number of signal points, if the density of constellation points in the constellation diagram is sparse, a larger constellation point diameter should be selected to draw the constellation diagram. On the contrary, if the constellation points in the constellation diagram are dense, some effective information will be lost due to overlapping coverage. At this time, it is chosen to reduce the diameter of the constellation points to reduce the occlusion between the constellation points. This method helps to reveal the actual distribution of the data and avoid visual information loss due to excessive overlap of points. Therefore, according to the size of the received signal sample, changing the diameter of the drawn constellation points can improve the density characteristics of the constellation diagram after quadrant normalization, which can further improve the signal-to-noise ratio recognition accuracy of the deep learning framework.

[0071] Embodiment 2: 64-QAM includes four enhanced signal-to-noise ratio estimation schemes.

[0072] Step 1: If Figure 1 This is an example of the system model process. If you want to output a 64-QAM constellation diagram, you need to complete it in the ACO-OFDM system. The data processing steps are as follows:

[0073] (1) Massive amounts of data collected in real time by in-vehicle sensors, such as road conditions, vehicle status information, and driver information in the smart cockpit.

[0074] (2) The above sensor data are processed through QAM mapping, power normalization, Hermite symmetry, serial-to-parallel conversion, inverse Fourier transform and negative clipping to obtain the ACO-OFDM signal to be transmitted.

[0075] (3) After the transmission signal passes through the time-varying channel, it undergoes parallel-to-serial conversion and Fourier transform at the receiving end to obtain a frequency domain subcarrier containing the transmission data. The complex signal on it is extracted and the data enhancement method proposed in the present invention is applied. First, coordinate axes are added to it, then quadrant normalization is performed, and finally a suitable constellation point diameter is selected to obtain a constellation diagram after data enhancement.

[0076] Step 2: Construct a deep learning dataset. Taking 64-QAM as an example, the selected SNR range is [11-25] dB, with a step of 1 dB. Among them, the batch size of the received signal (i.e. the number of received signals) is selected as 32, 64, 128, 256 and 512 respectively. For each signal-to-noise ratio, 100 constellation diagrams after data enhancement are generated, and 1500 enhanced constellation diagrams are generated for 15 signal-to-noise ratios; combined with five batches of received data, a total of 1500*5=7500 constellation diagrams are generated.

[0077] Step 3: The present invention uses the convolutional neural network ResNet-18 to extract image features and uses the constructed constellation data set to perform end-to-end training to estimate the signal-to-noise ratio of the communication quality indicator. The specific system flow is attached. Figure 1 As shown, the present invention uses the pytorch architecture to construct a specific network framework and training. The constructed constellation diagram training data set is input into the network framework for training. After the model is optimized and trained, the signal-to-noise ratio estimation accuracy of each method is obtained on the test set.

Claims

1. An intelligent monitoring method for signal-to-noise ratio of an AO-OFDM communication system for automobile data networking, comprising:

1. Data Processing in ACO-OFDM System 1. Input terminal 1) Massive data collected in real time by in-vehicle sensors, including road conditions, vehicle status information, and driver information in the smart cockpit; 2) Selecting the ACO-OFDM system based on QAM modulation for data modulation, calculating the power normalization factor according to the QAM order, performing QAM mapping on the collected data in step 1), and dividing the QAM mapped signal by the power normalization factor; 3) Perform Hermitian conjugation operation on the complex signal mapped by QAM. The ACO-OFDM signal uses odd subcarriers to transmit data, and the even subcarriers are set to zero. , n=0, 1, …, N-1; in, Represents the frequency domain data symbol modulated onto the nth subcarrier after mapping; the signal is converted into a bipolar real-valued OFDM signal through serial-to-parallel conversion and inverse fast Fourier transform; Then, negative clipping operation is performed, that is, the signal amplitude greater than zero in the time domain x is retained, and the signal amplitude less than zero in x is set to zero; then the signal is sent through the LED drive circuit to control the car headlights to send optical signals carrying the sensor data in the car, and through the complex time-varying driving road condition channel, the optical signal is transmitted to the photoelectric sensors of other vehicles on the road for real-time reception; ； Among them, y(n) is the optical receiving signal carrying the in-vehicle sensor data, h(n) is the channel response of the driving road conditions, x(n) is the transmitted signal carrying the in-vehicle sensor data in the time domain, and w(n) represents the noise introduced in the channel. Represents the convolution operation; 2. Receiving End 1) Other vehicles on the road receive signals y(n) in real time through the on-board photoelectric detector, and perform fast Fourier transform on them to obtain: ； Convert the received signal from time domain to frequency domain; frequency domain signal , extract the odd subcarriers to get ; 2) Perform parallel-to-serial conversion on the odd-numbered subcarrier QAM signal containing valid data, and perform QAM demapping to obtain its complex signal; 3) Draw a two-dimensional QAM constellation diagram, and perform coordinate axis information enhancement and quadrant normalization operations on the original constellation diagram; That is, add horizontal and vertical axes to the drawn constellation diagram; then add the complex Take the absolute values ​​of the real and imaginary parts of respectively, and normalize the complex signal demodulated by QAM to the first quadrant; 4) Selection of the constellation point drawing diameter According to the number of received signal samples, different diameters of constellation points are selected, including 5, 10 and 20, to avoid overlapping of high-density signals in the constellation diagram, so that the final enhanced constellation diagram containing coordinate axis position information, quadrant normalization and constellation point diameter has richer density and position features; 5) Use the on-board computer to estimate the signal-to-noise ratio of the received signal to obtain real-time road condition information for inter-vehicle communication. Based on the estimated signal-to-noise ratio at the receiving end, the LED drive circuit module is fed back to adjust the instantaneous power of the vehicle headlights, thereby improving the optical signal quality of inter-vehicle communication in real time and transmitting road condition information back to other vehicles on the road.

2. Constructing a Constellation Map Dataset Enhanced by Deep Learning (1) For QAM modulation, based on the bit error rate curve of the ACO-OFDM system, the signal-to-noise ratio range is selected as [1-40] dB, with an interval of 1 dB; (2) Select the number of received signals to be 32, 64, 128, 256, and 512. Then, randomly generate 100 enhanced constellation diagrams for each number of received signals with the same signal-to-noise ratio. The enhancement method is the superposition of coordinate axis position information, quadrant normalization, and constellation point diameter selection. For the same order of QAM modulation, a total of 500 constellation diagram data with the same signal-to-noise ratio are generated. Considering all 40 signal-to-noise ratio intervals, a total of 20,000 constellation diagram data sets are generated for the same order of QAM modulation; (3) Of these, 80% is used for training and 20% is used for testing and validation; 3. Deep Learning Model The ResNet-18 framework is used to perform supervised training on the constellation diagram containing the signal-to-noise ratio label to estimate the communication quality indicator signal-to-noise ratio.

2. The intelligent monitoring method for signal-to-noise ratio of vehicle-to-vehicle communication system for intelligent vehicle networking according to claim 1 is characterized by: The ResNet-18 framework is a ResNet18 residual network, with a total of 18 trainable convolutional layers in the input and output layers; the convolutional layer is followed by a 3×3 maximum pooling layer with a stride of 2. After the last residual block, a global average pooling layer is used to convert the feature map into a fixed-length vector. The output of the global average pooling (GAP) layer is fed to a fully connected layer, and the output of the fully connected layer is fed to the Softmax layer.

3. The intelligent monitoring method for signal-to-noise ratio of vehicle-to-vehicle communication system for intelligent vehicle networking according to claim 1 or 2, characterized in that: The power normalization factor calculation formula is as follows: ； Where M represents the QAM modulation order, and Respectively represent the real and imaginary parts of the i-th signal after QAM modulation; Make the horizontal axis coordinate of each constellation point of the QAM modulation constellation diagram under the M-ary system fall within [ ] within the range of ].

4. The intelligent monitoring method for signal-to-noise ratio of vehicle-to-vehicle communication system for intelligent vehicle networking according to claim 3 is characterized by: The coordinate positioning is to first find the center point, i.e., the coordinate origin, in the QAM constellation diagram, and then draw mutually orthogonal horizontal and vertical coordinate axes, i.e., two straight lines without arrows, x=0 and y=0, based on the coordinate origin.

5. The intelligent monitoring method for signal-to-noise ratio of vehicle-to-vehicle communication system for intelligent vehicle networking according to claim 4 is characterized in that: Further quadrant normalization is performed, that is, the four quadrants of the original constellation diagram are normalized to the first quadrant: The constellation image quadrant normalization operation is implemented as follows: both the real part and the imaginary part of the QAM demodulated complex signal are subjected to absolute value operations, thereby redrawing the complex constellation points originally belonging to the second, third and fourth quadrants into the first quadrant.

6. The intelligent monitoring method for signal-to-noise ratio of vehicle-to-vehicle communication system for intelligent vehicle networking according to claim 5 is characterized by: Under the same number of signal points, if the number of constellation points in the constellation diagram is small, for example, the number of constellation points is 32, then the diameter of the constellation diagram should be 20; conversely, if the number of constellation points in the constellation diagram is large, such as 512, then the diameter of the constellation diagram should be 5 to reduce the occlusion between the constellation points.

Images