Whole vehicle wireless communication performance detection method, device, system, medium and equipment

Abstract

The invention discloses a whole vehicle wireless communication performance detection method, device and system, a medium and equipment. The method comprises the following steps: establishing a test environment; collecting wireless communication data of a whole vehicle to be tested in the test environment; the collected wireless communication data is preprocessed; carrying out dynamic analysis on the preprocessed wireless communication data; a whole vehicle wireless communication performance detection model is established and trained; and inputting the analyzed wireless communication data of the whole vehicle to be detected into a trained wireless communication performance detection model of the whole vehicle so as to detect the wireless communication performance of the whole vehicle to be detected. According to the invention, efficient, dynamic and accurate detection of the wireless communication performance of the whole vehicle can be realized.

Description

Technical Field

[0001] The present application belongs to the field of automobile communication technology, and specifically relates to a method, device, system, medium and equipment for detecting the wireless communication performance of a whole vehicle. Background Art

[0002] Traditional vehicle wireless communication performance testing methods mainly focus on testing communication performance in static environments. They lack effective simulation of complex dynamic road conditions, environmental changes and network interference, making it difficult to comprehensively evaluate the vehicle's communication reliability in actual driving. Summary of the invention

[0003] In response to the deficiencies in the prior art, the purpose of this application is to provide a method, device, system, medium and equipment for detecting the wireless communication performance of a whole vehicle. This application uses machine learning to establish a prediction model, combined with high-precision real-time environmental simulation technology, to achieve dynamic and efficient detection of the wireless communication performance of the whole vehicle.

[0004] To achieve the above objectives, this application provides the following technical solutions: A method for testing the wireless communication performance of a whole vehicle, the method comprising: setting up a test environment; collecting wireless communication data of the whole vehicle to be tested under the test environment; preprocessing the collected wireless communication data; Dynamically analyze the preprocessed wireless communication data; establish and train a whole vehicle wireless communication performance detection model; input the analyzed whole vehicle wireless communication data to be tested into the trained whole vehicle wireless communication performance detection model to detect the wireless communication performance of the whole vehicle to be tested.

[0005] Optionally, the preprocessing of the collected wireless communication data includes: performing data cleaning on the collected wireless communication data; and performing denoising on the wireless communication data after the data cleaning.

[0006] Optionally, the whole vehicle wireless communication performance detection model includes: a backbone network, used to extract features of the wireless communication data; a feature fusion network, used to fuse the features extracted by the backbone network; and an output network, used to classify the wireless communication performance of the whole vehicle to be tested based on the features fused by the feature fusion network.

[0007] Optionally, the vehicle wireless communication performance detection model is trained through the following steps: collecting wireless communication data during vehicle operation or through a simulated environment, and preprocessing the collected data; dividing the sorted data set into a training set and a verification set; setting training parameters, and using the training set data to train the model until the maximum number of training times is reached; using the verification set to verify the model, and during the verification process, evaluating the model performance based on accuracy, recall and precision. When the accuracy, recall and precision are all greater than a threshold, the model verification passes; otherwise, adjusting the training parameters, and retraining the model until the model verification passes.

[0008] In addition, the present application also provides a whole vehicle wireless communication performance test device, the device comprising: a construction module, used to construct a test environment; a collection module, used to collect wireless communication data of the whole vehicle to be tested under the test environment; a preprocessing module, used to preprocess the collected wireless communication data; an analysis module, used to dynamically analyze the preprocessed wireless communication data; a model building and training module, used to establish and train a whole vehicle wireless communication performance detection model; a detection module, used to input the analyzed whole vehicle wireless communication data of the vehicle to be tested into the trained whole vehicle wireless communication performance detection model to detect the wireless communication performance of the whole vehicle to be tested.

[0009] In addition, the present application also provides a whole vehicle wireless communication performance testing system, which includes the device as described above.

[0010] In addition, the present application also provides a whole vehicle wireless communication performance testing system, and the system applies the method described in any of the preceding items.

[0011] In addition, the present application also provides a storage medium, which includes instructions, and when the instructions are executed on a computer, the computer executes the method as described in any of the preceding items.

[0012] In addition, the present application also provides an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the methods described above when executing the program.

[0013] Compared with the prior art, the present application has the following beneficial effects: (1) By introducing high-precision real-time environmental simulation technology, the present application can effectively reproduce complex dynamic road conditions, environmental changes and network interference, thereby evaluating communication reliability under close-to-real driving conditions and improving the comprehensiveness and accuracy of detection; (2) By constructing a whole-vehicle wireless performance detection model, the present application can extract and fuse features from multiple dimensions and multiple granularities, thereby enhancing the model's adaptability and understanding of changes in complex communication environments; (3) By effectively integrating features of different scales and sources, the present application can improve the model's sensitivity to subtle changes in communication performance and its grasp of overall communication trends, thereby achieving efficient, dynamic and accurate detection of the whole-vehicle wireless communication performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a flow chart of a method for detecting the wireless communication performance of a whole vehicle provided in one embodiment of the present application.

[0015] Figure 2 It is a structural diagram of a vehicle wireless communication performance detection model provided by another embodiment of the present application.

[0016] Figure 3 It is a structural diagram of a pyramid pooling module provided in another embodiment of the present application.

[0017] Figure 4 It is a structural diagram of an adaptive fusion module provided in another embodiment of the present application.

[0018] Figure 5 A schematic diagram of the structure of a whole vehicle wireless communication performance detection device provided in another embodiment of the present application.

[0019] Figure 6 It is a structural schematic diagram of a storage medium provided in another embodiment of the present application.

[0020] Figure 7 It is a structural schematic diagram of an electronic device provided in another embodiment of the present application. DETAILED DESCRIPTION

[0021] The specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Although the specific embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0022] It should be noted that certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that technicians may use different nouns to refer to the same component. This specification and claims do not use the difference in nouns as a way to distinguish components, but use the difference in the functions of the components as the criterion for distinction. As mentioned throughout the specification and claims, "including" or "comprising" is an open term, so it should be interpreted as "including but not limited to". The subsequent description of the specification is a preferred embodiment of the implementation of the present application, but the description is based on the general principles of the specification and is not used to limit the scope of the present application. The scope of protection of the present application shall be determined by the attached claims.

[0023] To facilitate understanding of the embodiments of the present application, further explanation will be given below using specific embodiments as examples in conjunction with the accompanying drawings, and the various drawings do not constitute a limitation on the embodiments of the present application.

[0024] Figure 1 FIG. 1 is a flow chart of a vehicle wireless communication performance test method provided by an exemplary embodiment of the present application. The execution subject of the method is an electronic device, which may include a computer, a server, etc. Figure 1 As shown, the method includes: S100: Build a test environment; S200: Collecting wireless communication data of the vehicle to be tested under the test environment; S300: pre-processing the collected wireless communication data; S400: dynamically analyzing the pre-processed wireless communication data; S500: Establish a vehicle wireless communication performance detection model and train the model; S600: Inputting the parsed wireless communication data of the whole vehicle to be tested into a trained whole vehicle wireless communication performance detection model to detect the wireless communication performance of the whole vehicle to be tested.

[0025] This embodiment, by introducing high-precision real-time environmental simulation technology, can effectively reproduce complex dynamic road conditions, environmental changes and network interference, thereby overcoming the limitations of traditional static testing, and then being able to evaluate the reliability of vehicle communication under close-to-real driving conditions, thereby improving the comprehensiveness and accuracy of vehicle wireless communication performance testing. In addition, by constructing a vehicle wireless performance detection model, this application can extract and fuse features from multiple dimensions and granularities, thereby achieving efficient, dynamic and accurate detection of the vehicle's wireless communication performance.

[0026] In another exemplary embodiment, in step S100, the vehicle test environment can be established by executing the following steps: S101: Deploy high-precision wireless communication test equipment (such as 5G test terminals, Wi-Fi 6 test modules) to receive and analyze wireless communication signals emitted by vehicles, as well as deploy GPS simulators, environmental variable simulation devices (such as adjustable obstacles, weather simulators), etc. to simulate different environmental conditions. At the same time, deploy high-speed data acquisition systems to ensure that all relevant communication data can be recorded in real time and accurately.

[0027] S102: Constructing a dynamic simulation scene, which may specifically include: 1. Physical environment simulation: It can be divided into indoor and outdoor environment simulation. Indoor environment simulation can be achieved by establishing an adjustable electromagnetic compatibility (EMC) darkroom, which uses absorbing materials to reduce external interference and is equipped with movable obstacles, simulated buildings and other models.

[0028] For outdoor environment simulation, a test site with various terrain features can be selected or built. The test site should be able to support the simulation of different environments and traffic conditions, including urban streets, highways, complex intersections, etc. In addition, variable obstacles (such as moving vehicle models, rockery, and tree models) can be further arranged.

[0029] In addition, you can use sprayers, fans and other equipment to simulate weather conditions such as rain and wind, or use a dedicated weather simulation cabin to simulate various weather changes.

[0030] 2. Traffic flow simulation: It can specifically include vehicle dynamics simulation and pedestrian simulation. Vehicle dynamics simulation refers to the use of automatic driving or remote control vehicles, combined with GPS and motion control systems, to simulate traffic flows of different speeds and densities.

[0031] Pedestrian simulation refers to the use of human models or volunteers wearing wearable devices to simulate pedestrian traffic to examine the impact on vehicle communication links.

[0032] In addition, interaction scenarios between vehicles can be further designed, such as overtaking, meeting, emergency braking, etc., to test the response and stability of the vehicle communication system.

[0033] 3. Wireless environment simulation: specifically, it may include signal interference source simulation, multipath effect simulation, and dynamic channel simulation, etc.

[0034] Among them, the signal interference source simulation refers to deploying a programmable interference signal source to simulate the interference of other wireless devices, radars, satellite communications, etc. on vehicle wireless communications.

[0035] The multipath effect simulation refers to simulating a complex multipath propagation environment using reflectors, scatterers, etc., to examine the reflection and diffraction effects of vehicle wireless signals on surfaces of different materials.

[0036] The dynamic channel simulation refers to developing or purchasing software that can dynamically adjust parameters based on the channel model of 3GPP or other communication standards to simulate changes in channel characteristics in different scenarios.

[0037] By simulating the above wireless environment, the test environment can evaluate the robustness and adaptability of the vehicle communication system in the face of interference.

[0038] 4. Communication protocol simulation: Install and configure software to simulate various communication protocols and network conditions, including different wireless communication standards and network topologies.

[0039] By simulating different communication protocols (such as LTE, 5G, Wi-Fi, DSRC, etc.), the test environment can reproduce wireless communication standards from old to the latest, ensuring that the wireless communication system of the vehicle under test can work normally under various protocols, thereby improving the comprehensiveness and applicability of the test.

[0040] In summary, by building a test environment, this application can effectively reproduce various complex dynamic road conditions, environmental changes and network interference that a vehicle may encounter during actual driving, and can perform real-time evaluation of the wireless communication performance of the entire vehicle, thereby making up for the limitations of traditional static testing and improving the authenticity and comprehensiveness of the vehicle's wireless communication performance testing.

[0041] In another exemplary embodiment, in step S300, the preprocessing of the collected wireless communication data includes: S301: Cleaning the collected wireless communication data; In this step, cleaning the collected wireless communication data specifically includes the following aspects: 1. Missing value processing: Check whether there are missing values ​​in the data, and choose to fill the missing values ​​(for example, you can use the mean, median, forward filling or backward filling methods) or delete the records with missing values ​​according to the specific situation.

[0042] 2. Outlier detection and processing: Identify and process outliers in the data. Specifically, outlier detection can be achieved through statistical methods (the mean and standard deviation can be used to identify data points far away from the mean. For example, any data point that exceeds the mean plus or minus three times the standard deviation may be considered an anomaly), box plots (the boundaries of the data are determined by the interquartile range, and values ​​beyond the upper and lower boundaries are considered anomalies) or time series analysis (for continuous communication data streams, time series models (such as ARIMA) can be used to predict expected values ​​and compare them with actual values ​​to find anomalies).

[0043] S302: De-noising the wireless communication data after the data cleaning; In this step, the wireless communication data can be smoothed by smoothing technology (such as moving average method) to reduce random noise. Further, the smoothed wireless communication data can be filtered in the frequency domain, that is, the signal is converted to the frequency domain by Fourier transform, so that different frequency components are separated, and then low-pass, high-pass and band-pass filters are applied in sequence in the frequency domain for filtering (low-pass filter can remove high-frequency noise in the data and maintain the basic trend of the signal; high-pass filter can remove low-frequency interference and highlight the transient part of the signal; band-pass filter can filter out signals in a specific frequency band and remove noise of other frequencies), and then inversely transformed back to the time domain to remove noise of specific frequencies. Furthermore, the wireless communication data after frequency domain filtering can be more finely denoised by wavelet transform to deal with the sudden noise contained therein.

[0044] The multi-level denoising strategy mentioned above can effectively separate useful signals from the data, which helps the model to better understand the structure and content of the data, thereby reducing misleading information in the learning and detection process of the model, and further improving the accuracy of the model in predicting and classifying the wireless communication performance of the entire vehicle.

[0045] Furthermore, preprocessing the collected wireless communication data may also include: S303: performing data fusion on the denoised wireless communication data; In this step, the data fusion may specifically include the following steps: 1. Data standardization: It is necessary to standardize data from different sources, such as vehicle status information, location data, driving behavior data, etc. provided by various vehicle terminals, roadside units (RSU), vehicle sensors (such as inertial navigation system INS), vehicle-to-vehicle communication (V2V), etc., to ensure that all data are comparable on the same scale. For example, minimum-maximum normalization or Z-Score normalization can be used to map all value ranges to between 0 and 1.

[0046] 2. Time series alignment: Use Precision Time Protocol (PTP) or Network Time Protocol (NTP) to ensure time synchronization between data from the above different sources.

[0047] For example, for the network time protocol, you first need to configure the NTP server, and specifically you can select or set one or more reliable NTP servers as time sources. These servers can obtain time from a public NTP pool on the Internet, or connect to a high-precision clock source such as GPS, atomic clocks, etc. to ensure high-precision time.

[0048] Secondly, install and configure the NTP client software on all client devices that require time synchronization (such as computers, servers, IoT devices, etc.), edit the NTP configuration file (such as / etc / ntp.conf in Linux system), and specify the upstream NTP server address.

[0049] Finally, synchronize the time. The NTP client will automatically synchronize with the configured NTP server periodically and adjust the local system time to match the server time.

[0050] By implementing the above steps, effective time synchronization can be achieved between different data sources, providing a consistent time benchmark for subsequent data analysis and application.

[0051] 3. Correlation analysis: Use a data association algorithm (such as nearest neighbor, joint probabilistic data association JPDA) to associate the above time-synchronized data.

[0052] For example, take the nearest neighbor (NN) method as an example: 1) Ensure that all data is sorted by timestamp and contains a unique identifier (such as a timestamp or sequence number) 2) Define a distance metric to measure the "distance" between two data points (such as positions or feature vectors observed by two sensors that may belong to the same object). This can be Euclidean distance, Mahalanobis distance, or other applicable distance metrics.

[0053] 3) Matching data points: For each new observation at a moment, calculate its distance from all existing trajectories or observations at the previous moment, and select the observation with the closest distance as its associated target.

[0054] 4) Threshold judgment: Set a maximum distance threshold. If the distance of the nearest neighbor exceeds this threshold, it is considered that the new observation has no direct correlation with the existing trajectory and may represent a new target or noise. Only when the time difference between the two data points is less than the threshold and the distance metric meets the preset conditions, the match is confirmed to be successful. For example, set a time difference threshold, such as 0.1 seconds. Only when the time difference between the two data points is less than this threshold, they are considered whether they can be associated. In addition, in addition to the time window, a threshold for the distance metric needs to be set. For example, if the Euclidean distance exceeds a certain preset value, then even if they are very close in time, it may not be considered a successful match. Only when two data points are both within the time window and meet the criteria of the distance metric, they are confirmed to be successfully matched and data association can be performed.

[0055] 4. Data Fusion: Methods such as weighted averaging, Kalman filtering, and Bayesian fusion can be selected to merge the wireless communication data after association analysis into a new data set, which includes the key information of all source data while reducing redundancy and noise.

[0056] In summary, by fusing the wireless communication data from different sources after denoising, firstly, different data can be complemented to obtain a more comprehensive wireless signal coverage map and richer scene information, which helps the model learn diverse wireless communication signal characteristics, so as to more accurately classify the types (such as Wi-Fi, LTE, 5G, etc.) and performance levels of the whole vehicle wireless communication data; secondly, the fusion of multi-source data can increase the model's resistance to outliers or erroneous data. Even if a part of the data is biased or erroneous, the information from other data sources can still support the correct classification decision, making the model more robust; thirdly, by continuously collecting and fusing wireless communication data, a complete picture of the signal changing over time and space can be constructed. This spatiotemporal continuity is particularly important for understanding complex communication environments, and helps the model capture the dynamic changes of signal characteristics, so as to more accurately classify wireless communication data. In short, by fusing vehicle wireless communication data from different sources, the model's understanding of the depth and breadth of wireless communication signal characteristics can be significantly enhanced, thereby improving its classification ability.

[0057] S304: Establish a data warehouse to store the integrated wireless communication data.

[0058] Specifically, the establishment of the data warehouse may refer to executing the following steps: 1. Data storage design: Choose a suitable data warehouse solution based on the characteristics of the data, such as a relational database or a NoSQL database. Specifically, it may include open source Apache Hive, Hadoop + HBase, Greenplum, ClickHouse, or commercial solutions such as Amazon Redshift, Google BigQuery, Snowflake, Teradata, etc.

[0059] 2. Data conversion: Convert the integrated wireless communication data into the format required by the data warehouse, which may include data type conversion, data splitting, calculation of derived fields, etc.

[0060] 3. Data loading: Use ETL (Extract, Transform, Load) tools or custom scripts to load the cleaned and integrated wireless communication data into the data warehouse 4. Implement data encryption, access control and audit logs to ensure data security.

[0061] In summary, by establishing a data warehouse, we can first effectively cope with the massive growth of vehicle wireless communication data and ensure that model training can be performed on a sufficiently large data set, thereby improving the generalization ability and stability of the model; secondly, the data warehouse can store a large amount of data, which is convenient for time series analysis and identification of signal change trends, periodic patterns or abnormal situations, which is very important for predicting future communication conditions and improving the timeliness of classification models. Finally, data warehouses usually use specific data models (such as star or snowflake models) to store data in a highly organized manner, improve the quality of the data, and ensure the accuracy and consistency of the data used for model training.

[0062] In another exemplary embodiment, in step S400, dynamically parsing the pre-processed wireless communication data includes: S401: define data format and protocol; In this step, it is first necessary to clarify the structure and communication protocol specifications of the wireless communication data to be processed, including the header information, payload structure, control field, etc. of the data packet.

[0063] By defining data formats and protocols, seamless communication between different systems, devices, and services can be allowed. For example, in vehicle communications, whether it is Wi-Fi, LTE or 5G, each communication technology has its own specific data packet structure and transmission rules. By defining unified data formats and protocols, devices from different manufacturers can recognize and understand each other's data, thus ensuring interoperability in the entire ecosystem.

[0064] S402: Capturing and segmenting data packets based on the defined format and protocol; When data arrives, it is first processed by the physical layer and link layer to ensure that the data is received correctly, and then the data stream is segmented to separate the complete data packet from the continuous bit stream.

[0065] S403: Dynamically identify the data packet type; In this step, the type of each data packet is dynamically identified by parsing the header information of the data packet, such as the protocol type identifier.

[0066] S404: Dynamically apply parsing rules according to the data packet type; In this step, the corresponding parsing logic is applied according to the data packet type, including decompression, decoding, field splitting, etc. Dynamic parsing means that the parsing rules are not statically fixed, but dynamically adjusted according to the content of the data packet.

[0067] S405: Data verification and error handling: In this step, during the parsing process, data integrity check, error detection and correction mechanisms are implemented to ensure data accuracy and reliability. For invalid or damaged data packets, appropriate error handling strategies should be adopted, such as discarding, requesting retransmission or recording error logs.

[0068] In order to better understand the above parsing steps, this embodiment takes a data packet in a Wi-Fi-based wireless network as an example for specific description, the purpose is to extract the source MAC address, destination MAC address and signal strength of each data packet. Specifically, the parsing steps for the data packet are as follows: 1. Define data formats and protocols Wi-Fi packets follow the 802.11 standard and include a frame header, frame body, and frame trailer. The frame header contains the source and destination MAC addresses, and the signal strength can be found in the Wi-Fi signal report.

[0069] 2. Packet Capture and Segmentation You can set up a sniffer at a suitable location (such as a network interface, serial port) or use the API provided by the communication module to capture data packets. Alternatively, you can use the network programming interface (such as socket) of the corresponding language to receive data.

[0070] Exemplarily, in this embodiment, the data packets sent by the vehicle communication module are captured by executing the following procedures: import socket # Set the server address and port HOST = 'localhost' # or use '0.0.0.0' to listen on all network interfaces PORT = 12345 # Select an unused port # Create a socket object with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: # Bind port s.bind((HOST, PORT)) # Start monitoring, and hang up to 5 connection requests s.listen(5) print(f"Server listening on {HOST}:{PORT}...") while True: # Accept a client connection conn, addr = s.accept() with conn: print(f"Connected by {addr}") while True: # Receive data. Here we use 4096 bytes as the buffer size. Adjust it according to the actual situation. data = conn.recv(4096) # If no data is received, it means the client has closed the connection if not data: break # Parse and process the received data process_data(data) def process_data(data): """ Example function: Processing received data """ print(f"Received data: {data.decode('utf-8')}") # Assume the data is text encoded in UTF-8 3. Dynamically identify data packet types Check the frame type field (located in the frame header) of the data packet, such as management frame, control frame or data frame, to determine the type of data packet and the corresponding parsing method. For example, the scapy library in Python can be used to dynamically identify the type of 802.11 wireless LAN data packet. The executed program example is as follows: from scapy.all import * def analyze_wifi_packet(pkt): # First check if it is an 802.11 packet if Dot11 in pkt: # Check the frame control field to determine the packet type frame_type = orb(pkt[Dot11].FCfield) # Parse the frame control field to get type information subtype = (frame_type>>4)&0b00111111 # Determine the packet type based on the subtype if subtype == 0: print("Management frame") elif subtype<= 15: # The subtype range of the control frame is 1-15 print("Control frame") else: print("data frame") # Print more information about the packet print("Source MAC:", pkt[Dot11].addr1) print("Destination MAC:", pkt[Dot11].addr2) if Dot11QoS in pkt: print("QoS Packet") else: print("Non-QoS Packet") else: print("Non-802.11 packet") # Use Scapy to capture packets and analyze sniff(iface="wlan0", prn=analyze_wifi_packet, lfilter=lambda x: Dot11in x) The above program first imports the Scapy library and defines a function analyze_wifi_packet to analyze each captured data packet. The program first checks whether the packet contains the 802.11 layer (i.e. whether it is a Wi-Fi packet), then reads the frame control field (FCfield), and determines whether the packet is a management frame, control frame, or data frame based on the subtype bit of the field. It then prints out the frame type, source MAC address, destination MAC address, and whether it is a QoS packet.

[0071] 4. Dynamically apply parsing rules based on data type In this step, the data types include physical layer data (such as signal strength, frequency, modulation mode, etc.), link layer data packets (including 802.11 Wi-Fi data packets (management frames, control frames, data frames), network layer data packets (such as IPv4, IPv6 data packets, used for addressing and routing), transport layer data packets (TCP, UDP data packets, providing end-to-end reliable or connectionless services), and application layer protocol data (HTTP, HTTPS, SMTP, FTP, etc., to implement specific network service functions). The parsing rules include: header parsing rules (defining how to parse and verify fixed format information in the header of the data packet, such as version number, length, checksum, etc.), field extraction rules (for different data packet types, specifying which fields need to be extracted, such as source IP, destination IP, port number, payload data, etc.), error handling rules (specifying how to handle errors in the parsing process, such as checksum mismatch, incomplete data packet, etc.), data conversion rules (converting raw data into a format suitable for further processing or analysis, such as timestamp conversion, numerical unit conversion, etc.), and logical decision rules (making decisions based on the parsed information, such as adjusting the processing priority of the data packet according to the signal strength).

[0072] Below, this embodiment describes in detail how to parse a data packet through the following exemplary procedures: According to the protocol, it is necessary to start from the beginning of the data packet and parse each field one by one. For example, for a TCP data packet, first parse the 16-bit source port and destination port, followed by the 32-bit sequence number and confirmation number, etc.

[0073] Assume there is a simple TCP protocol, whose data format is specifically defined as follows: 16-bit source port number (Source Port): identifies the port number of the sending end, used for multiplexing.

[0074] 16-bit destination port number (Destination Port): identifies the port number of the receiving end, also used for multiplexing.

[0075] 32-bit sequence number (Sequence Number): The sequence number of the first byte in the sent data segment, used for data sorting and confirmation.

[0076] 32-bit Acknowledgment Number: The sequence number of the next byte expected to be received from the other party, used to confirm data reception.

[0077] 4-bit Data Offset (i.e. header length): indicates the length of the TCP header in 32-bit words (4 bytes), used to determine the starting point of the data portion.

[0078] 6 reserved bits (Reserved, usually set to 0): Usually set to 0 and reserved for future protocol expansion.

[0079] 6-bit flags (such as ACK, SYN, etc.): Contains multiple control flags, such as ACK (acknowledgement), SYN (synchronization), FIN (end), etc., which are used to control the establishment, maintenance and removal of the connection.

[0080] 16-bit Window Size: Indicates the current size of the receive buffer and is used for flow control.

[0081] Furthermore, the definition of the protocol includes not only the data format, but also the rules for how the data is encapsulated, transmitted, error handled, and session managed. For the TCP header definition, the protocol definition may include: Encapsulation rules: The TCP header is followed by the IP header, and the TCP data segment is followed by the TCP header.

[0082] Checksum: defines the method of calculating the checksum of the entire TCP header and data segment to ensure the integrity of data transmission.

[0083] Retransmission mechanism: defines how to trigger the retransmission of a data packet when the expected confirmation is not received.

[0084] Congestion control: defines how to dynamically adjust the data sending rate according to network conditions to avoid network congestion.

[0085] Connection establishment and termination: Detailed description of the process and rules of TCP three-way handshake to establish a connection and four-way handshake to disconnect.

[0086] Exemplarily, this embodiment takes TCP / IP protocol header parsing as an example, and uses Python language to illustrate how to parse a received TCP data packet according to a defined TCP header data format: def parse_tcp_header(data): # Ensure that the data length at least includes the TCP header if len(data)<20: # Assume the minimum length of the TCP header is 20 bytes raise ValueError("Data too short to contain a TCP header.") # Parsing header fields src_port = int.from_bytes(data[:2], byteorder='big') # 16-bit source port dst_port = int.from_bytes(data[2:4], byteorder='big') # 16-bit destination port seq_num = int.from_bytes(data[4:8], byteorder='big') # 32-bit sequence number ack_num = int.from_bytes(data[8:12], byteorder='big') # 32-bit confirmation number # Data offset (unit is 4 bytes), needs to be shifted left 2 bits (because each unit is 4 bytes) data_offset = (data

[12] &0xf0)>>2 # Reserved bits and flag bits (here we simplify the process and only read some bits in the flag bit) flags = data

[13] flag_ACK = (flags&1<<4) != 0 # ACK flag flag_SYN = (flags&1<<6) != 0 # SYN flag # 16-bit window size window_size = int.from_bytes(data[14:16], byteorder='big') # Return parsing results return { "Source Port": src_port, "Destination Port": dst_port, "Sequence Number": seq_num, "Acknowledgment Number": ack_num, "Data Offset": data_offset * 4, # Convert to bytes "ACK Flag": flag_ACK, "SYN Flag": flag_SYN, "Window Size": window_size } # Assume that data is the original data received from the network, including the TCP header raw_data = b'\x12\x34\x56\x78\x90\xAB\xCD\xEF\x00\x00\x00\x00\x00\x10\x02\x00\x00\x00\x00\x00' parsed_header = parse_tcp_header(raw_data) print(parsed_header) In this example, the parse_tcp_header function receives raw data as input and then uses bit manipulation and byte order conversion to extract the value of each field according to the structure of the TCP header.

[0087] Processing load data: Decode according to the payload type. For example, if it is in JSON format, use the JSON parser; if it is binary data, decode according to the specific protocol definition.

[0088] For example, let's take the TCP payload as an example (assuming the payload is in JSON format). import socket import json # Define the function to handle TCP connection def handle_tcp_connection(conn): """Handle TCP connection, receive and parse JSON payload""" # Use the with statement to ensure that the connection is properly closed at the end of the function with conn: while True: # Use the recv method to receive data. 4096 is the buffer size, the maximum amount of data that can be received data = conn.recv(4096) # If no data is received (an empty byte string is returned), it means that the client has closed the connection if not data: break # Assume that the received data is UTF-8 encoded text and decode it into a string payload = data.decode('utf-8') # Try to convert the decoded string into a JSON object try: # Use the loads method of the json library to parse the JSON string json_payload = json.loads(payload) print("Received JSON Payload:", json_payload) # Call another function to further process the parsed JSON data process_json_payload(json_payload) except json.JSONDecodeError: # If an error occurs while parsing JSON, print the error message print("Error decoding JSON payload") # Define a function to process the parsed JSON data def process_json_payload(payload): """Process the parsed JSON data""" # Example operation: Print the 'message' field in the received JSON data print("Processing JSON data...") print("Message:", payload.get("message")) # Main function if __name__ == "__main__": # Set the server listening address and port host = 'localhost' port = 12345 # Create a TCP socket based on IPv4 with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: # Bind the server address and port to the socket s.bind((host, port)) # Start listening, the maximum number of pending connections allowed is 1 s.listen() print(f"Server listening on {host}:{port}...") # Infinite loop waiting for client connection while True: # Accept a client connection request and return a new socket and client address conn, addr = s.accept() print(f"Connected by {addr}") # Call handle_tcp_connection function for each connection handle_tcp_connection(conn) The above program implements a TCP server that listens for connection requests on a specified port. Once a client connects, it receives the data sent by the client and tries to parse it into a JSON object. Finally, it executes the corresponding processing logic based on the parsing result.

[0089] 5. Convert the decoded data into appropriate data structures (such as dictionaries, lists, etc.) and perform further processing such as data validation, formatting, or storage.

[0090] The conversion of the decoded data into an appropriate data structure (such as dictionary, list, etc.) mainly depends on the format of the data.

[0091] The following is an example. When the decoded data is in JSON format, you can use the JSON library in most programming languages ​​to convert it into a structure such as a dictionary or a list.

[0092] The following uses Python as an example to illustrate how to convert the decoded data into an appropriate data structure: import json # Assume `json_string` is a string containing JSON data json_string = '{"name": "Alice", "age": 30, "skills": ["Python", "JavaScript"]}' # Use the `json.loads()` method to convert a JSON string to a Python dictionary data_dict = json.loads(json_string) print(data_dict) # Output: {'name': 'Alice', 'age': 30, 'skills': ['Python', 'JavaScript']} In this example, the JSON string is converted into a dictionary containing the key-value pair structure of the original data.

[0093] 6. Format the parsed data according to the predefined data model to ensure that all sensor data is mapped to the same fields and data types. For example, timestamps are uniformly converted to ISO 8601 format, and numerical types are kept consistent. The following steps may be included: 1. Define the data model Clarify the data model, define the name, data type, and any necessary conversion rules for each field.

[0094] For example, a sensor data model can be defined as follows: class SensorDataModel: timestamp: str # Timestamp in ISO 8601 format sensor_id: int reading_type: str value: float 2. For each parsed data element, clean and format it according to the data model: Timestamp conversion: Make sure all timestamps are converted to ISO 8601 format. If the timestamp format of the original data is different, you can use the corresponding library (such as Python's datetime module) to convert it: from datetime import datetime def convert_to_isoformat(timestamp_str, input_format): dt_object = datetime.strptime(timestamp_str, input_format) return dt_object.isoformat() Unify numeric types: Check and convert numeric types to ensure that all numeric data is of the expected data type (such as float). For non-standard numeric representations (such as numbers in string form), you can try to convert: Python def convert_to_float(value_str): try: return float(value_str) except ValueError: return None # or handle exceptions as needed 3. Map the cleaned and transformed data into the pre-defined data model.

[0095] The cleansed and transformed data may be mapped into a predefined data model by executing the following exemplary procedure.

[0096] def map_data_to_model(raw_data): # Assume raw_data is a dictionary containing the raw sensor data cleaned_data = { 'timestamp': convert_to_isoformat(raw_data['timestamp'], '%Y-%m-%d %H:%M:%S'), # Assume the original timestamp format 'sensor_id': int(raw_data['sensor_id']), 'reading_type': raw_data['type'].lower(), # Assume that the original data contains type information 'value': convert_to_float(raw_data['value']) } return SensorDataModel(**cleaned_data) # Assume that SensorDataModel supports keyword parameter initialization In summary, the dynamic analysis of the pre-processed wireless communication data can, firstly, be able to perform instant analysis according to the real-time changes of the data packet without the need to fix the analysis rules in advance. This means that even if a new or mutated data format is encountered during the communication process, the system can quickly adapt and correctly analyze it, thereby improving the flexibility and real-time performance of processing complex wireless communication data; secondly, by dynamically identifying the type of data packet and analyzing it according to the specific rules of different types of data packets, analysis errors can be reduced to ensure that the extracted data information is accurate; thirdly, dynamic analysis can allocate computing resources according to the importance of data and analysis requirements, prioritize the analysis of key data, and quickly process non-critical or known format data, thereby optimizing the use of system resources and improving overall efficiency; fourthly, in the whole vehicle wireless communication performance detection, dynamic analysis can ensure the efficient processing of various types of wireless communication data, including but not limited to Wi-Fi, LTE, 5G and other communication standards, so that the detection model can be based on a more accurate and comprehensive data set for performance evaluation, thereby improving the accuracy and precision of model detection. In short, by dynamically analyzing the pre-processed wireless communication data, not only the efficiency and accuracy of model detection can be improved, but also the adaptability of the model to the changing environment can be enhanced, so that the model can achieve dynamic, efficient and accurate detection of the whole vehicle wireless communication performance.

[0097] In another exemplary embodiment, Figure 2 As shown, the whole vehicle wireless communication performance detection model includes: a backbone network, a feature fusion network and an output network.

[0098] Wherein, the backbone network includes: Input layer, embedding layer, 1D convolution layer, normalization layer, dilated convolution layer, normalization layer, pooling layer, 1D convolution layer, normalization layer, dilated convolution layer, normalization layer, pooling layer, and pyramid pooling module (SPPF).

[0099] Among them, Figure 3 As shown, the pyramid pooling module includes a ConvBNSiLU layer (composed of a Conv2d layer, a BatchNorm2d layer and a SiLU activation function), and the ConvBNSiLU layer connects two branches, wherein the first branch includes a Concat function and a ConvBNSiLU layer; the second branch includes three sequentially connected two-dimensional maximum pooling layers MaxPool2d, and the three sequentially connected two-dimensional maximum pooling layers MaxPool2d are respectively connected to the Concat function, and the two-dimensional maximum pooling layer MaxPool2d selects the maximum value in the window of input data as output, In addition, the backbone network also adopts a residual structure, that is, the output of the first pooling layer and the output of the second pooling layer, in addition to being input into the feature fusion network respectively, need to be superimposed with the input and the output of the pyramid pooling module before being input into the feature fusion network.

[0100] The working principle of the backbone network with the above structure is described below in this embodiment: The input layer is used to receive wireless communication data after preprocessing and analysis, including signal strength, spectrum characteristics, communication delay, packet loss rate, multipath effect, etc.

[0101] The embedding layer is used to encode the input data into a higher-dimensional feature vector so that subsequent layers of the model can better capture complex patterns.

[0102] The first one-dimensional convolution layer is responsible for capturing the most intuitive and basic features from the input data, such as the fluctuation pattern of signal strength, changes in frequency components, preliminary patterns of communication delays, and basic laws of packet loss rates. Since it is the initial layer, it tends to focus on local features in the input data, such as short-term signal patterns and repetitive sequences. In addition, the first one-dimensional convolution layer can also identify the boundaries of signal changes in time series data, such as mutation points in signal strength, which helps to distinguish different types of communication events or anomalies; further, the first one-dimensional convolution layer can also extract features within a specific frequency band, such as spectrum utilization and spectrum occupancy, to provide basic information for subsequent processing.

[0103] The first normalization layer is used to normalize the basic features initially extracted by the first one-dimensional convolutional layer to ensure that these basic features have good distribution properties, thereby reducing the impact of input data scale changes on model training and helping the model converge faster.

[0104] The first dilated convolutional layer focuses on extracting local but larger-view features, such as subtle changes in signal strength or fluctuations in specific frequencies, allowing the model to capture broader spatial context information.

[0105] The second normalization layer is located in the middle of the backbone network, which can continue to maintain the stable distribution of intermediate activation values, help the gradient to better back propagate, and further accelerate training.

[0106] The first pooling layer reduces the temporal dimension of the input data through downsampling operations, which can reduce the computational burden of the model while extracting local features.

[0107] The second one-dimensional convolutional layer can further refine and combine the basic features extracted by the first one-dimensional convolutional layer to extract higher-level feature representations, such as more complex signal patterns, timing dependencies, potential communication anomaly patterns, changes in communication patterns over a long period of time, complex manifestations of multipath effects, or patterns of interaction between signals, etc., allowing the model to focus more on key features that have a decisive impact on communication performance.

[0108] As the network depth increases, the gradient problem becomes more prominent. Setting a third normalization layer can help the model control the scale of the gradient and avoid gradient disappearance or explosion. In addition, in the deeper part of the network, setting a third normalization layer can help the model learn more abstract and meaningful features because it can ensure that each layer receives well-normalized input.

[0109] The second dilated convolutional layer can further expand the receptive field based on the first dilated convolutional layer, and can capture more global and abstract features, such as periodic changes in communication channels or multipath effect characteristics in complex environments, so that the model can capture long-distance dependency information in the data without omission.

[0110] The fourth normalization layer can directly limit the maximum and minimum values ​​of the model output value, avoiding the problem of abnormally large or small output values ​​caused by improper adjustment of network weights, thereby ensuring that the output of the model is within a reasonable range.

[0111] The second pooling layer can further reduce the time dimension of the input data and extract more abstract and high-level features, which helps the model learn higher-level patterns of the data, such as the trend of communication performance over time. In addition, the second pooling layer can also help the model see a wider range of input information, thereby capturing longer-term dependencies.

[0112] The pyramid pooling module can fuse the features obtained by the two pooling layers to form a feature representation with rich information. This fusion not only retains the feature details of each scale, but also promotes the complementarity between features, thereby enhancing the model's ability to understand complex communication scenarios. In addition, in wireless communication data, the characteristics of the signal may show different scale characteristics due to factors such as distance and environmental changes. The pyramid pooling module can effectively cope with such changes, thereby improving the robustness of the model.

[0113] In summary, this embodiment can increase the depth of the model by setting two one-dimensional convolutional layers in the backbone network, so that the model is capable of learning deeper, nonlinear data representations. This deep structure helps to capture long-distance dependencies and complex interactions in the data. For wireless communication performance detection, it can more accurately identify and predict changes in communication quality in dynamic environments; in addition, through two convolution operations, the model can generate more diverse feature maps, which can describe input data from different angles and scales. This diversity helps the model to more comprehensively understand all aspects of wireless communication performance and improve the accuracy and robustness of detection. Furthermore, the setting of two one-dimensional convolutional layers can enhance the model's ability to express data, especially when dealing with highly dynamic and complex communication scenarios, such as multipath effects in urban centers, channel changes in rapid movement, etc. By extracting richer features, the model can better adapt to and distinguish different communication performance states.

[0114] By setting four normalization layers in the backbone network, the stability of data distribution can be maintained in multiple stages of model training, the training speed can be accelerated, and the stability and generalization ability of the model can be improved. At the same time, considering that the backbone network adopts a residual structure, the normalization layer can ensure that the original data input through the input layer maintains a suitable distribution before being added to the deep layer, making residual learning more effective and avoiding information loss or conflict caused by differences in data distribution.

[0115] By setting two dilated convolutional layers in the backbone network, not only can the model's feature extraction capability in wireless communication performance detection be improved, but the model can also be made more robust to noise and anomalies in the data. By extracting features at different scales, the model can more stably identify and respond to challenges in various communication environments, such as dynamically changing channel conditions and different types of interference.

[0116] By setting two pooling layers in the backbone network, it is possible to gradually reduce the data dimension, extract and integrate key features, and enhance the stability and generalization ability of the model.

[0117] By setting up a pyramid pooling module in the backbone network and its ability to extract and fuse multi-scale features, the model's adaptability to complex communication environments can be enhanced, thereby improving the accuracy and efficiency of detection.

[0118] In addition, it should be noted that the use of residual structures to superimpose input and pooling layers and the output of the pyramid pooling module can establish a direct connection between features at different levels. This design ensures that low-level features can be combined with high-level features, which helps the model to utilize more levels of contextual information and improve the ability to understand complex communication signals while reducing overfitting. In addition, by adopting a residual structure, the feature fusion network can receive not only high-level features that have undergone multi-layer transformations in the backbone network, but also low-level features that have not been transformed or have undergone a small amount of transformation, as well as multi-scale features provided by the pyramid pooling module. The superposition of all these features makes the feature fusion more comprehensive, which is conducive to the model generating more accurate and robust communication performance predictions.

[0119] The feature fusion network includes a first branch, a second branch and a third branch with the same structure, and each branch includes an adaptive fusion module, an attention pooling layer and a long short-term memory network LSTM connected in sequence. Among them, the adaptive fusion module of the first branch is connected to the first pooling layer in the backbone network, that is, the output of the first pooling layer of the backbone network is used as the input of the adaptive fusion module of the first branch; the adaptive fusion module of the second branch is connected to the second pooling layer in the backbone network, that is, the output of the second pooling layer of the backbone network is used as the input of the adaptive fusion module of the second branch; the adaptive fusion module of the third branch is connected to the pyramid pooling module in the backbone network, that is, the output of the pyramid pooling module of the backbone network is used as the input of the adaptive fusion module of the third branch.

[0120] The above feature fusion strategy enables the model to integrate feature information from different abstract levels by sending the outputs of different levels in the backbone network to different branches of the feature fusion network. Among them, the output of the first pooling layer represents more basic and local features, the output of the second pooling layer represents higher-level and more abstract features, and the output of the pyramid pooling module contains multi-scale feature information. This design ensures that the model can comprehensively consider features of different complexities and granularities, thereby comprehensively analyzing wireless communication performance. In addition, combining the outputs of different pooling layers and pyramid pooling modules enables the model to capture and express key information in the data from multiple angles, including not only the basic characteristics of communication signals, but also deeper time series patterns and long-term dependencies, thereby enhancing the model's ability to identify and predict changes in wireless communication performance. In short, this feature fusion strategy can not only enhance the accuracy and robustness of the model for wireless communication performance detection through multi-level and multi-angle feature integration, but also optimize the model's adaptability and generalization performance in complex environments, providing technical support for efficient and accurate communication performance detection.

[0121] It should be noted that the attention pooling layer in the first branch of the feature fusion network is also connected to the long short-term memory network LSTM in the second branch and the long short-term memory network LSTM in the third branch; the attention pooling layer in the second branch is also connected to the long short-term memory network LSTM in the first branch and the long short-term memory network LSTM in the third branch; the attention pooling layer in the third branch is also connected to the long short-term memory network LSTM in the first branch and the long short-term memory network LSTM in the second branch.

[0122] In the structure shown in the feature fusion network, by connecting the attention pooling layer and the LSTM of different branches across branches, the model can promote the interaction and information exchange between different feature branches. This not only enables the model to better fuse the features extracted by each branch, but also ensures the transmission and complementarity of key information between different feature dimensions, and enhances the model's ability to understand and model complex communication scenarios. In addition, as a network structure that is good at processing sequence data, LSTM can not only capture the temporal dependencies within each branch through cross-branch connections, but also use the temporal information of other branches for supplementation, making the model more accurate and comprehensive when learning the pattern of wireless communication performance changing over time. Furthermore, the attention mechanism given by the attention pooling layer allows the model to dynamically assign different weights to different features, and the cross-branch connection can further promote the globality of this dynamic adjustment, so that the model can adjust the attention to each feature according to the content of the entire input sequence, thereby improving the efficiency of feature utilization. In summary, the cross-branch attention pooling layer connection mechanism described above can significantly enhance the performance of the feature fusion network by promoting cross-interaction of features, integration of temporal dependencies, and adaptive adjustment of global feature importance, enabling it to show stronger adaptability, robustness, and accuracy when processing complex tasks such as wireless communication performance detection.

[0123] Further, such as Figure 4 As shown, the adaptive fusion module adopts a bottom-up input method, specifically including three branches, wherein the first branch includes the sequentially connected input layer Input1, 2×2 convolution layer Conv, and 1×1 convolution layer Conv; the second branch includes the sequentially connected input layer Input2 and 1×1 convolution layer Conv; the third branch includes the sequentially connected input layer Input3, 1×1 convolution layer Conv, upsampling layer Upsample, and 1×1 convolution layer Conv. Furthermore, the adaptive fusion module also includes sequentially connected 3×3 convolution layers, activation function softmax, 3×3 convolution layers, and output layers output, wherein the outputs of the first to third branches are added together, in addition to being input into the first 3×3 convolution layer, and are multiplied by the output of the activation function softmax before being input into the second 3×3 convolution layer.

[0124] In such Figure 4 In the adaptive fusion module shown, the input is first upsampled or downsampled, where upsampling is achieved using a 1×1 convolution layer and a bilinear interpolation upsampling module, and downsampling is achieved through a 2×2 convolution layer with the stride set to 2. By combining upsampling and downsampling, the dimensions of different levels can be made the same, and then feature fusion can be performed. Then, using the residual structure, the weights of different levels are obtained by using a 1×1 convolution layer and a softmax function, the weights are multiplied by the features of each level, and finally the fused feature information is output through a 3×3 convolution layer.

[0125] The adaptive fusion module described in this embodiment has a three-branch structure, each branch processes features with different inputs and processing paths, including convolution layers of different sizes (such as 2×2 convolution layers, 1×1 convolution layers), and upsampling layers (Upsample), which shows that the adaptive fusion module can process and fuse feature information from different scales. This design helps the model capture features at different resolutions or granularities, and has good sensitivity to subtle changes in wireless communication performance and global trends. In addition, the adaptive fusion module also includes an activation function softmax. Through the operation of this activation function, the adaptive fusion module can assign different weights to features from different branches, which is equivalent to adaptively adjusting the importance of the features, so that the model can automatically identify which features are more critical in the current task, thereby optimizing the utilization efficiency of the features and the prediction ability of the model. In addition, the upsampling layer and downsampling operation (implemented by a 2×2 convolution layer) ensure that the outputs of different branches can match in dimension to facilitate feature fusion. This operation ensures the preparation before information integration, so that the model can effectively integrate features across different scales. Furthermore, two consecutive 3×3 convolutional layers can not only deepen the nonlinear expression ability of the model, but also further refine and enhance these features after feature fusion, which helps the model learn more complex patterns and improve the recognition and prediction accuracy of communication performance changes.

[0126] In summary, the adaptive fusion module with the above structure can significantly improve the accuracy, robustness and learning efficiency of the model in wireless communication performance detection tasks through multi-branch feature fusion strategy, adaptive adjustment of feature importance and deep feature enhancement mechanisms.

[0127] Please continue to refer to Figure 2The output network includes a fully connected layer, a fully connected layer, a fully connected layer, a Dropout layer and an activation function Softmax layer connected in sequence, wherein the first fully connected layer is connected to the long short-term memory network LSTM in the first branch of the feature fusion network, the second fully connected layer is connected to the long short-term memory network LSTM in the second branch of the feature fusion network, and the third fully connected layer is connected to the long short-term memory network LSTM in the third branch of the feature fusion network.

[0128] The output network can integrate feature information from different branches by respectively connecting the three branches of the feature fusion network (each branch contains LSTM units), thereby ensuring that the model can fully utilize features of different levels and dimensions, including local, high-level and multi-scale fused features, which helps to improve the model's comprehensive understanding and accuracy of wireless communication performance detection.

[0129] In addition, continuous fully connected layers can also play a role in feature refinement. Each fully connected layer can further abstract and refine features based on the output of the previous layer, thereby gradually transforming the fused features into prediction results related to communication performance indicators. This process helps the model to perform more detailed divisions and decisions in high-dimensional feature space.

[0130] Furthermore, the introduction of the Dropout layer is a regularization technique that randomly "discards" the output of some neurons, forcing the model to learn more robust feature representations, reducing the risk of overfitting, and increasing the generalization ability of the model, so that the model can perform well on unseen data.

[0131] The final Softmax layer is responsible for converting the output of the model into a probability distribution, with each category corresponding to a probability value, and the sum is 1, so that the model's prediction result is not only a simple category label, but also reflects the model's confidence in each classification result. In addition, the probability distribution generated by the Softmax layer can intuitively evaluate the performance of the vehicle's wireless communication. For example, if the model predicts that the probability of a performance problem on a certain communication link is high, this suggests that the link may have serious problems and needs to be checked or optimized first.

[0132] In summary, the output network structure improves the prediction accuracy, generalization ability and interpretability of the model in wireless communication performance detection tasks through hierarchical feature integration, feature refinement, regularization strategy and interpretable classification output, so that the model can not only effectively utilize multi-source feature information, but also provide stable and reliable performance evaluation in complex communication environments.

[0133] In summary, the vehicle wireless communication performance detection model constructed in this application can extract features including but not limited to signal strength, packet loss rate, delay time, bit error rate, etc. based on the collected vehicle wireless communication data, and then based on the above-mentioned networks, the wireless communication performance of the vehicle can be detected and performance classification can be performed, for example, including: Excellent: strong signal, low latency, near-zero bit error rate, high throughput, and very low packet loss rate.

[0134] Good: Communication is smooth, with occasional slight delay or packet loss, but it does not affect normal communication.

[0135] Normal: There is a certain signal fluctuation, and the packet loss rate and delay time are within an acceptable range, which may affect some service experience.

[0136] Poor: Frequent packet loss, high latency, and weak signal, which seriously affect communication quality and user experience.

[0137] Very bad: Communication is almost interrupted and effective data transmission is impossible.

[0138] This model can realize real-time evaluation of the communication status of the whole vehicle and classify the communication quality according to the evaluation results, providing a scientific basis for vehicle communication system optimization, fault prediction and maintenance decision-making.

[0139] Furthermore, the present application compares the two classic models, the signal-to-noise ratio threshold model and the packet loss rate and delay analysis model in the field of wireless communication performance detection of the present model, and records the performance of each model in terms of accuracy, recall rate, precision, F1 score and processing time, as shown in Table 1:

[0140] As shown in Table 1, this model is superior to the classic model in accuracy, recall, precision, and F1 score. It also has a faster processing time and strong generalization ability, showing higher detection efficiency and stability. This proves the superiority of this model in vehicle wireless communication performance detection, especially in complex dynamic road conditions, environmental changes, and network interference. It can provide more accurate and efficient detection results.

[0141] In another exemplary embodiment, the present application trains the vehicle wireless communication performance detection model through the following steps: S501: Collect wireless communication data during vehicle operation or through a simulated environment, including but not limited to signal strength, packet loss rate, delay time, bit error rate, data throughput, etc., and pre-process the collected data (for the specific pre-processing process, please refer to the above description); S502: Divide the sorted data set into a training set and a validation set, for example, the division ratio may be 7:3; S503: Setting training parameters, for example, the learning rate can be set to 0.001, the batch size can be set to 32, the number of iterations can be set to 100, and the model is trained using the training set data until the maximum number of training times is reached; S504: Use the validation set to validate the model. During the validation process, the model performance is evaluated based on accuracy, recall, and precision. For example, when the accuracy, recall, and precision are all greater than a threshold (e.g., 90%), the model validation is successful. Otherwise, the training parameters are adjusted. For example, the learning rate can be adjusted to 0.05, the batch size can be adjusted to 64, the number of iterations can be adjusted to 200, and the model is retrained until the model validation is successful.

[0142] In another exemplary embodiment, Figure 5 As shown, the present application also provides a vehicle wireless communication performance test device, the device comprising: A building module 100 is used to build a test environment; The acquisition module 200 is used to acquire wireless communication data of the vehicle to be tested in the test environment; The preprocessing module 300 is used to preprocess the collected wireless communication data; The parsing module 400 is used to dynamically parse the pre-processed wireless communication data; Model building and training module 500, used to build and train the vehicle wireless communication performance detection model; The detection module 600 is used to input the parsed wireless communication data of the whole vehicle to be tested into a trained whole vehicle wireless communication performance detection model to detect the wireless communication performance of the whole vehicle to be tested.

[0143] In another exemplary embodiment, the present application also provides a vehicle wireless communication performance testing system, the system comprising the device as described above.

[0144] In another exemplary embodiment, the present application also provides a vehicle wireless communication performance testing system, and the system applies the method described in any of the preceding items.

[0145] Based on the above embodiments, Figure 6 , this application also provides a storage medium, please refer to Figure 6, the computer-readable storage medium shown is a CD 40, on which a computer program (i.e., a program product) is stored. When the computer program is executed by the processor, each step recorded in the above method implementation will be implemented, for example, setting up a test environment; collecting wireless communication data of the whole vehicle to be tested under the test environment; preprocessing the collected wireless communication data; dynamically parsing the preprocessed wireless communication data; establishing and training a whole vehicle wireless communication performance detection model; inputting the parsed whole vehicle wireless communication data to be tested into the trained whole vehicle wireless communication performance detection model to detect the wireless communication performance of the whole vehicle to be tested. The specific implementation method of each step will not be repeated here.

[0146] It should be noted that the computer-readable storage medium includes but is not limited to phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0147] Based on the above embodiments, the present application also provides an electronic device, as shown below: Figure 7 An electronic device for downloading a file according to an exemplary embodiment of the present application is described.

[0148] Figure 7 A block diagram of an exemplary electronic device 50 suitable for implementing the embodiments of the present application is shown, and the electronic device 50 may be a computer system or a cloud server. Figure 7 The electronic device 50 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0149] like Figure 7 As shown, the electronic device 50 includes but is not limited to: one or more processors or processing units 501, a system memory 502, and a bus 503 connecting different system components (including the system memory 502 and the processing unit 501).

[0150] The electronic device 50 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 50, including volatile and non-volatile media, removable and non-removable media.

[0151] The system memory 502 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 5021 and / or cache memory 5022. The electronic device 50 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the ROM 5023 may be used to read and write non-removable, non-volatile magnetic media ( Figure 7 is not shown in the Figure 7 As shown in the figure, a disk drive for reading and writing a removable non-volatile disk (such as a "floppy disk") and an optical disk drive for reading and writing a removable non-volatile optical disk (such as a CD-ROM, DVD-ROM or other optical media) can be provided. In these cases, each drive can be connected to the bus 503 through one or more data medium interfaces. The system memory 502 may include at least one program product, which has a set (such as at least one) of program modules, which are configured to perform the functions of each embodiment of the present application.

[0152] A program / utility 5025 having a set (at least one) of program modules 5024 may be stored, for example, in system memory 502, and such program modules 5024 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. Program modules 5024 generally perform the functions and / or methods of the embodiments described herein.

[0153] The electronic device 50 may also communicate with one or more external devices 504 (e.g., a keyboard, a pointing device, a display, etc.). Such communication may be performed via an input / output (I / O) interface 505. Furthermore, the electronic device 50 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 506. Figure 7 As shown, the network adapter 506 communicates with other modules (such as the processing unit 501, etc.) of the electronic device 50 via the bus 503. It should be understood that although Figure 7 Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 50 .

[0154] The processing unit 501 executes various functional applications and data processing by running the program stored in the system memory 502, for example, setting up a test environment; collecting wireless communication data of the whole vehicle to be tested under the test environment; preprocessing the collected wireless communication data; dynamically parsing the preprocessed wireless communication data; establishing and training a whole vehicle wireless communication performance detection model; inputting the parsed whole vehicle wireless communication data to be tested into the trained whole vehicle wireless communication performance detection model to detect the wireless communication performance of the whole vehicle to be tested. The specific implementation of each step is not repeated here. It should be noted that although several units / modules or sub-units / sub-modules of the file concurrent download device are mentioned in the above detailed description, this division is only exemplary and not mandatory. In fact, according to the implementation of the present application, the features and functions of two or more units / modules described above can be concretized in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules to be concretized.

[0155] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0156] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0157] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0158] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0159] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that can be executed by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a cloud server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0160] The above embodiments are only for illustrating the technical concept and features of the present application, and their purpose is to enable people familiar with the technology to understand the content of the present application and implement it accordingly, and they cannot be used to limit the protection scope of the present application. Any equivalent changes or modifications made according to the spirit of the present application should be included in the protection scope of the present application.

Claims

1. A vehicle wireless communication performance testing method, characterized in that: The method comprises: Set up a test environment; Collecting wireless communication data of the vehicle to be tested under the test environment; Preprocessing the collected wireless communication data; Dynamically analyzing the preprocessed wireless communication data; Establish and train the vehicle wireless communication performance detection model; The parsed wireless communication data of the whole vehicle to be tested is input into a trained whole vehicle wireless communication performance detection model to detect the wireless communication performance of the whole vehicle to be tested.

2. The method according to claim 1, characterized in that The preprocessing of the collected wireless communication data includes: Performing data cleaning on the collected wireless communication data; The wireless communication data after the data cleaning is denoised.

3. The method according to claim 1, characterized in that The vehicle wireless communication performance detection model includes: A backbone network, used for extracting features from the wireless communication data; A feature fusion network, used to fuse the features extracted by the backbone network; The output network is used to classify the wireless communication performance of the whole vehicle to be tested based on the features fused by the feature fusion network.

4. The method according to claim 1, characterized in that The vehicle wireless communication performance detection model is trained by the following steps: Collect wireless communication data during vehicle operation or through a simulated environment, and pre-process the collected data; Divide the sorted data set into training set and validation set; Set the training parameters and use the training set data to train the model until the maximum number of training times is reached; The model is verified using the validation set. During the verification process, the model performance is evaluated based on accuracy, recall, and precision. When the accuracy, recall, and precision are all greater than the threshold, the model verification passes; otherwise, the training parameters are adjusted and the model is retrained until the model verification passes.

5. A vehicle wireless communication performance test device, characterized in that: The device comprises: Build modules to build the test environment; A collection module, used to collect wireless communication data of the vehicle to be tested under the test environment; A preprocessing module, used for preprocessing the collected wireless communication data; A parsing module, used for dynamically parsing the pre-processed wireless communication data; Model building and training module, used to build and train the vehicle wireless communication performance detection model; The detection module is used to input the parsed wireless communication data of the whole vehicle to be tested into a trained whole vehicle wireless communication performance detection model to detect the wireless communication performance of the whole vehicle to be tested.

6. A vehicle wireless communication performance test system, characterized in that: The system comprises the apparatus of claim 5.

7. A vehicle wireless communication performance test system, characterized in that: The system applies the method according to any one of claims 1 to 4.

8. A storage medium, characterized in that: The method comprises instructions, which, when executed on a computer, enable the computer to execute the method according to any one of claims 1 to 4.

9. An electronic device, characterized in that: The electronic device comprises: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 4 is implemented.

Images