Wireless signal active capturing method applied to driver theory examination cheating prevention

Abstract

The invention discloses a wireless signal active capturing method applied to cheating prevention of a driver theory examination, which comprises the following steps: acquiring wireless signal real-time data in an examination room, preprocessing the wireless signal real-time data, and screening out abnormal wireless signals from the preprocessed wireless signal real-time data; performing modulation mode identification on the abnormal wireless signal by adopting a preset modulation identification algorithm to obtain a corresponding modulation type; performing protocol feature analysis on the abnormal wireless signal based on the modulation type to obtain a protocol feature of the abnormal wireless signal; and judging whether the corresponding abnormal wireless signal is a cheating signal based on the protocol feature and a preset cheating signal feature library. According to the method, cheating signals of known types can be identified, the feature library can be continuously expanded through self-learning, and dynamic identification and prevention of novel cheating means are realized. The accuracy and the real-time performance of examination room wireless signal cheating detection are remarkably improved, and a powerful guarantee is provided for maintaining fairness and justice of an examination.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication and signal processing, and in particular relates to a method for actively capturing wireless signals for preventing cheating in a driver's theoretical test. Background Art

[0002] In the method of actively capturing wireless signals to prevent cheating in driver's theoretical test, the signal parsing and analysis system faces many technical challenges. First, the wireless signal environment in the test room is complex, with a large number of legitimate signals and interference signals. How to accurately identify cheating signals from many signals is a difficult problem. Secondly, the characteristics of cheating signals are diverse, and different modulation methods, frequencies and protocols may be used. How to establish a complete cheating signal feature library and achieve accurate identification of various cheating signals is also a major challenge. Furthermore, the cheating methods are constantly upgraded, and the concealment and camouflage of cheating signals are constantly improving. How to achieve rapid detection and response to new cheating signals is also an urgent problem to be solved. Finally, in the actual test scenario, the wireless signal environment may change at any time. How to achieve the adaptability of the system and ensure stable and reliable operation in a dynamically changing environment is also a technical difficulty that cannot be ignored. In summary, how to achieve accurate identification and rapid response to diversified and concealed cheating signals in a complex wireless signal environment is the core technical problem faced by the signal parsing and analysis system. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a wireless signal active capture method for anti-cheating in driver's theoretical test, so as to solve the problems existing in the above prior art.

[0004] To achieve the above object, the present invention provides a wireless signal active capture method for anti-cheating in driver theory test, comprising:

[0005] Acquire the real-time data of wireless signals in the examination room and perform preprocessing, and filter out abnormal wireless signals from the preprocessed real-time data of wireless signals;

[0006] Using a preset modulation identification algorithm to identify the modulation mode of the abnormal wireless signal to obtain a corresponding modulation type;

[0007] Performing protocol feature analysis on the abnormal wireless signal based on the modulation type to obtain the protocol feature of the abnormal wireless signal;

[0008] Based on the protocol feature and a preset cheating signal feature library, it is determined whether the corresponding abnormal wireless signal is a cheating signal.

[0009] Optionally, the process of screening out abnormal wireless signals includes:

[0010] A spectrum analysis algorithm is used to analyze the preprocessed real-time data of wireless signals to obtain the spectrum characteristics of the signals. According to the preset normal signal spectrum characteristic model, it is judged whether the spectrum characteristics are abnormal. If the spectrum characteristics are abnormal, it is determined as a suspected abnormal signal and marked. A signal strength detection algorithm is used to analyze the suspected abnormal signals, and abnormal wireless signals are screened out based on the signal strength analysis results and the spectrum characteristic judgment results.

[0011] Optionally, the process of using a preset modulation identification algorithm to identify the modulation mode of the abnormal wireless signal to obtain the corresponding modulation type includes:

[0012] Extract characteristic parameters of the abnormal wireless signal; compare the extracted characteristic parameters of the abnormal signal with pre-stored signal characteristics of known modulation types; if the degree of matching between the characteristic parameters of the abnormal wireless signal and the signal characteristics of the known modulation type exceeds a preset threshold, determine that the modulation type of the abnormal wireless signal is the corresponding known modulation type; if there is no characteristic matching item, use a support vector machine algorithm to identify the modulation type of the abnormal wireless signal based on the characteristic parameters of the abnormal wireless signal; if the modulation type of the abnormal signal cannot be identified using the support vector machine algorithm, use a convolutional neural network algorithm to identify the modulation type of the abnormal wireless signal.

[0013] Optionally, the process of obtaining the protocol feature of the abnormal wireless signal includes:

[0014] Based on the modulation type of the abnormal wireless signal, a corresponding protocol analysis module is obtained; based on the protocol analysis module, protocol features of the abnormal wireless signal are extracted to obtain protocol feature parameters; according to the protocol feature parameters, a deep learning algorithm is used to analyze the protocol level information of the abnormal wireless signal to obtain protocol features.

[0015] Optionally, the process of obtaining the protocol characteristics includes:

[0016] Based on the protocol feature parameters, the abnormal wireless signal is preprocessed to extract the original information at the protocol level; the convolutional neural network algorithm is used to extract the features of the original information at the protocol level to obtain the protocol feature vector; the protocol feature vector is input into the long short-term memory network for serialization modeling; the output of the long short-term memory network is weighted fused through the attention mechanism to obtain the fused feature vector; the fused feature vector is input into the fully connected layer for information analysis to obtain detailed information at the protocol level as the protocol feature.

[0017] Optionally, the process of determining whether the corresponding abnormal wireless signal is a cheating signal based on the protocol feature and a preset cheating signal feature library includes:

[0018] The protocol feature is compared with a preset cheating signal feature library. If there is a match, it is determined to be a cheating signal. If there is no match, deep feature extraction is performed on the protocol feature to obtain a unique feature vector of the abnormal wireless signal. The unique feature vector is calculated to have a similarity with a feature vector in a preset cheating signal feature library to obtain a similarity score. If the similarity score exceeds a threshold, it is a known cheating signal. If it does not exceed the threshold, it is a new cheating signal. If it is a new cheating signal, the preset cheating signal feature library is updated based on the unique feature vector of the abnormal wireless signal.

[0019] Optionally, based on the business scenario of the abnormal wireless signal, a cheating signal feature library matching the business scenario is obtained; the protocol features are compared with the features in the cheating signal feature library one by one to determine whether there is a match; if there is a match, the cheating signal feature with the highest match is determined as the feature of the abnormal wireless signal, and the abnormal wireless signal is determined to be a cheating signal.

[0020] Optionally, the process of updating the preset cheating signal feature library includes:

[0021] Determine whether the similarity score exceeds the threshold. If so, determine the abnormal signal as a known cheating signal. If the similarity score does not exceed the threshold, determine it as a suspected new cheating signal. Perform cluster analysis on the suspected new cheating signal, and determine whether it is a new cheating signal based on the clustering results. If so, add the unique feature vector of the abnormal wireless signal to the preset cheating signal feature library for update.

[0022] The present invention also provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-mentioned method for actively capturing wireless signals for preventing cheating in a driver's theoretical test.

[0023] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for actively capturing wireless signals for preventing cheating in a driver's theoretical test are implemented.

[0024] Compared with the prior art, the present invention has the following advantages and technical effects:

[0025] The present invention discloses a method for actively capturing wireless signals for preventing cheating in theoretical driving tests. First, by acquiring wireless signal data in the test room in real time, spectrum analysis and signal strength detection are performed to screen abnormal signals. Subsequently, modulation mode identification and protocol feature analysis are performed on the abnormal signal, and the analysis result is compared with a preset cheating signal feature library. For signals that cannot be matched, a machine learning method is used to extract their unique features, and the features are input into a dynamically updated feature library for pattern matching to determine whether they are new cheating signals. This method can not only identify known types of cheating signals, but also continuously expand the feature library through self-learning, thereby realizing dynamic identification and prevention of new cheating methods. The present invention significantly improves the accuracy and real-time performance of cheating detection of wireless signals in the test room, and provides a strong guarantee for maintaining fairness and justice in the test. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0027] Figure 1 The figure is a flow chart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0029] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0030] Embodiment 1

[0031] like Figure 1 As shown, this embodiment provides a wireless signal active capture method for anti-cheating in a driver's theoretical test, including:

[0032] Acquire the real-time data of wireless signals in the examination room and perform preprocessing, and filter out abnormal wireless signals from the preprocessed real-time data of wireless signals;

[0033] In some specific implementations, the process of screening out abnormal wireless signals includes:

[0034] A spectrum analysis algorithm is used to analyze the preprocessed real-time data of wireless signals to obtain the spectrum characteristics of the signals. According to the preset normal signal spectrum characteristic model, it is judged whether the spectrum characteristics are abnormal. If the spectrum characteristics are abnormal, it is determined as a suspected abnormal signal and marked. A signal strength detection algorithm is used to analyze the suspected abnormal signals, and abnormal wireless signals are screened out based on the signal strength analysis results and the spectrum characteristic judgment results.

[0035] Specifically, obtaining real-time data of wireless signals in the examination room is a key step in monitoring the examination environment. This can be achieved by deploying multiple highly sensitive wireless receivers that can capture WiFi signals, Bluetooth signals, and other common wireless communication signals in the 2.4GHz and 5GHz frequency bands. For example, in a 100 square meter examination room, 10 receivers can be evenly arranged, each sampling 1000 times per second to ensure full coverage and high time resolution. Transmitting the collected raw signal data to the signal analysis system is the next important step. This is usually done through a wired network or a dedicated wireless channel to avoid interference with other wireless signals in the examination room. Encryption measures such as AES-256 encryption algorithm are used during transmission to ensure data security and integrity. Preprocessing the acquired real-time signal data is key to improving the accuracy of subsequent analysis. This includes removing background noise and known interference sources. For example, adaptive filters can be used to eliminate power line noise, or wavelet transforms can be applied to remove sudden interference. Through these preprocessing steps, the signal-to-noise ratio can be increased by 5-10dB, significantly improving signal quality. Spectral analysis is the core technology for identifying abnormal signals. Commonly used methods include fast Fourier transform (FFT) and wavelet analysis. For example, for WiFi signals, the normal spectrum should show a typical sinusoidal distribution around 2.4GHz or 5GHz. If there are obvious peaks outside these frequency bands, or the spectrum shape is abnormal, it may be an indication of an abnormal signal. The preset normal signal spectrum feature model is the basis for judging abnormalities. This model can be trained based on a large amount of historical data through machine learning algorithms such as support vector machines (SVM) or deep learning networks. The model should be able to adapt to different examination room environments and equipment differences and have good generalization capabilities. For signals marked as suspected abnormalities, further signal strength detection is necessary. This can be achieved by measuring the received signal strength indication (RSSI). For example, if a signal with a strength exceeding -30dBm suddenly appears at a certain location in the examination room, and the signal strength in the surrounding area is generally around -60dBm, this is likely to be an abnormal signal source. Finally, the results of the comprehensive spectrum characteristics and signal strength analysis can more accurately identify abnormal signals. This process can be achieved using algorithms such as decision trees or fuzzy logic. For example, if a signal has both abnormal spectrum characteristics (such as a clear peak at 2.3GHz) and abnormal signal strength (such as a sudden strong signal of -20dBm), then it can be highly confident that this is an abnormal signal that needs further investigation. Through this series of steps, abnormal wireless signals in the examination room can be effectively identified and screened, providing strong technical support for maintaining examination fairness and preventing cheating. This method can not only monitor the examination environment in real time, but also provide detailed signal feature analysis, providing an important basis for subsequent evidence collection and investigation.

[0036] Using a preset modulation identification algorithm to identify the modulation mode of the abnormal wireless signal to obtain a corresponding modulation type;

[0037] In some specific implementations, a preset modulation identification algorithm is used to identify the modulation mode of the abnormal wireless signal, and a process of obtaining a corresponding modulation type includes:

[0038] Extract characteristic parameters of the abnormal wireless signal; compare the extracted characteristic parameters of the abnormal signal with pre-stored signal characteristics of known modulation types; if the degree of matching between the characteristic parameters of the abnormal wireless signal and the signal characteristics of the known modulation type exceeds a preset threshold, determine that the modulation type of the abnormal wireless signal is the corresponding known modulation type; if there is no characteristic matching item, use a support vector machine algorithm to identify the modulation type of the abnormal wireless signal based on the characteristic parameters of the abnormal wireless signal; if the modulation type of the abnormal signal cannot be identified using the support vector machine algorithm, use a convolutional neural network algorithm to identify the modulation type of the abnormal wireless signal.

[0039] Specifically, in wireless signal analysis, identifying the modulation type of abnormal signals is a crucial step. First, a database containing signal features of multiple known modulation types needs to be established. This database may include common modulation methods such as amplitude modulation (AM), frequency modulation (FM), phase modulation (PM), etc. Each modulation type has its own unique characteristic parameters, such as bandwidth, spectrum shape, phase change, etc. For the screened abnormal signals, their characteristic parameters need to be extracted. This may involve time domain and frequency domain analysis, including information such as signal amplitude, frequency, and phase. For example, for a suspected FM signal, its frequency offset and modulation index may be of interest. Next, the extracted characteristic parameters are compared with the pre-stored known modulation types. This process can use similarity measurement methods such as Euclidean distance or cosine similarity. Assuming that the similarity between the characteristics of an abnormal signal and the pre-stored QPSK (quadrature phase shift keying) signal characteristics reaches 0.95, exceeding the preset 0.9 threshold, it can be preliminarily determined that the signal may use QPSK modulation. However, simple feature matching may not be sufficient to cope with complex actual situations. Therefore, the support vector machine (SVM) algorithm is introduced. SVM can construct hyperplanes in high-dimensional feature space to effectively separate data points of different categories. Here, multi-class SVM can be used to identify multiple modulation types. For example, an SVM model can be trained to input the feature vector of the signal and output the possible modulation types and their probabilities. If SVM cannot give a reliable recognition result, a convolutional neural network (CNN) can be used for further analysis. CNN has performed well in the field of image recognition and can also be applied to signal modulation recognition. The signal can be converted into a time-frequency graph or other two-dimensional representation and then input into CNN. CNN can automatically learn the hierarchical features of the signal, potentially capturing subtle patterns that may be overlooked by manually designed features. Finally, the modulation type recognition result of the abnormal signal is output. This result includes not only the most likely modulation type, but also other possibilities and their probabilities. For example, it may be concluded that the abnormal signal has an 80% probability of QPSK modulation, a 15% probability of 8PSK modulation, and a 5% probability of other types. This multi-level, multi-algorithm modulation type recognition method can improve the accuracy and robustness of recognition. It allows the signal to be analyzed from different angles and has alternative solutions when simple methods fail. This is especially important for processing signals in complex real-world environments, because real-world signals may be affected by noise, interference, multipath effects, and other factors, making it difficult for a single method to reliably identify the modulation type. This comprehensive approach can better understand and analyze abnormal signals, providing an important basis for subsequent signal processing and safety measures.

[0040] Performing protocol feature analysis on the abnormal wireless signal based on the modulation type to obtain the protocol feature of the abnormal wireless signal;

[0041] In some specific implementations, the process of obtaining the protocol characteristics of the abnormal wireless signal includes:

[0042] Based on the modulation type of the abnormal wireless signal, a corresponding protocol analysis module is obtained; based on the protocol analysis module, protocol features of the abnormal wireless signal are extracted to obtain protocol feature parameters; according to the protocol feature parameters, a deep learning algorithm is used to analyze the protocol level information of the abnormal wireless signal to obtain protocol features.

[0043] Specifically, determining the corresponding protocol parsing module according to the modulation type of the abnormal signal is a key step in signal analysis. For example, for frequency shift keying (FSK) modulated signals, a dedicated FSK demodulator and decoder may be required. For quadrature amplitude modulation (QAM) signals, a QAM demodulation algorithm may be required. This targeted parsing can greatly improve the accuracy of protocol feature extraction. In the process of protocol feature extraction, key parameters such as frame synchronization sequence, header structure, and data load can be focused on. Taking the 802.11 wireless LAN protocol as an example, information such as the type, source address, and destination address of the MAC frame can be extracted. These feature parameters lay the foundation for subsequent deep learning analysis. When using deep learning algorithms to parse protocol-level information, models such as recurrent neural networks (RNNs) or long short-term memory networks (LSTMs) can be used. These models can effectively capture the timing dependencies in protocol data streams. For example, when analyzing the TCP protocol, the characteristics of the handshake process, data transmission, and connection termination stages can be learned to identify abnormal behaviors. Multi-dimensional comparison can be used to determine whether the abnormal signal conforms to the preset feature model. For example, for DNS tunnel attacks, you can check whether the frequency, length, and entropy of DNS queries are abnormal. If a large number of extremely long DNS query requests are found, and the entropy of the query content is significantly higher than that of normal DNS traffic, an alarm mechanism may be triggered. When using clustering algorithms to classify abnormal signals, you can choose methods such as K-means or DBSCAN. Taking network intrusion detection as an example, abnormal traffic can be clustered according to features such as source IP address, destination port, and protocol type. In this way, you may find that one type of scanning behavior comes from a specific IP segment and targets a specific service, while another type may be a feature of a distributed denial of service (DDoS) attack. Association rule mining helps to discover potential connections between abnormal signals. For example, using the Apriori algorithm to analyze Web application attack logs, you may find that SQL injection attempts are often associated with a specific User-Agent field. This association feature not only helps to improve detection accuracy, but also provides clues for tracing the source of the attack. Through this series of analysis steps, valuable security intelligence can be extracted from the original abnormal signals. This multi-level and multi-angle analysis method can not only improve the accuracy of anomaly detection, but also provide network security managers with a more comprehensive and in-depth understanding of threats, so as to formulate more targeted defense strategies.

[0044] In some specific implementations, according to the protocol characteristic parameters, a deep learning algorithm is used to parse the protocol layer information of the abnormal wireless signal, and the process of obtaining the protocol characteristics includes:

[0045] Based on the protocol feature parameters, the abnormal wireless signal is preprocessed to extract the original information at the protocol level; the convolutional neural network algorithm is used to extract the features of the original information at the protocol level to obtain the protocol feature vector; the protocol feature vector is input into the long short-term memory network for serialization modeling; the output of the long short-term memory network is weighted fused through the attention mechanism to obtain the fused feature vector; the fused feature vector is input into the fully connected layer for information analysis to obtain detailed information at the protocol level as the protocol feature.

[0046] Specifically, protocol feature parameter extraction is a key step in abnormal signal analysis. Taking wireless communication protocols as an example, parameters such as frame synchronization, modulation mode, and encoding mode can be extracted. For the 802.11 protocol, information such as MAC frame header and frame body can be extracted. Noise reduction and interpolation can be performed in the preprocessing stage to improve the accuracy of subsequent analysis. Convolutional neural networks play an important role in protocol feature extraction. Taking TCP / IP protocol stack analysis as an example, a multi-layer convolution structure can be designed, with the first layer extracting packet header features, the second layer extracting load features, and the third layer fusing the features of the first two layers. This hierarchical feature extraction can effectively capture the structured information of the protocol. Long short-term memory networks are suitable for analyzing protocol behaviors with temporal characteristics. For example, when analyzing the MQTT protocol, operation sequences such as connection establishment, message publishing, and subscription can be input into LSTM, and the model can learn normal protocol interaction patterns. When an abnormal sequence occurs, such as frequent disconnection and reconnection, the model can quickly identify it. The attention mechanism can highlight key information in protocol analysis. Taking HTTP protocol analysis as an example, different weights can be given to different parts such as request methods, URLs, and header fields. For POST requests that may contain malicious payloads, the model will pay more attention to the content of the request body to improve the accuracy of anomaly detection. The fully connected layer performs the final information analysis and maps high-dimensional features to specific protocol semantics. For example, when analyzing the DNS protocol, the features extracted previously can be mapped to specific meanings such as query type and response code.

[0047] Based on the protocol feature and a preset cheating signal feature library, it is determined whether the corresponding abnormal wireless signal is a cheating signal.

[0048] In some specific implementations, the process of determining whether the corresponding abnormal wireless signal is a cheating signal based on the protocol feature and a preset cheating signal feature library includes:

[0049] The protocol feature is compared with a preset cheating signal feature library. If there is a match, it is determined to be a cheating signal. If there is no match, deep feature extraction is performed on the protocol feature to obtain a unique feature vector of the abnormal wireless signal. The unique feature vector is calculated to have a similarity with a feature vector in a preset cheating signal feature library to obtain a similarity score. If the similarity score exceeds a threshold, it is a known cheating signal. If it does not exceed the threshold, it is a new cheating signal. If it is a new cheating signal, the preset cheating signal feature library is updated based on the unique feature vector of the abnormal wireless signal.

[0050] Furthermore, according to the business scenario of the abnormal wireless signal, a cheating signal feature library matching the business scenario is obtained; the protocol features are compared with the features in the cheating signal feature library one by one to determine whether there is a match; if there is a match, the cheating signal feature with the highest matching degree is determined as the feature of the abnormal wireless signal, and the abnormal wireless signal is determined to be a cheating signal.

[0051] Specifically, according to the business scenario of the abnormal signal, a pre-established cheating signal feature library matching the business scenario is obtained; the feature set of the abnormal signal is compared with the features in the cheating signal feature library one by one to determine whether there is a match; if there is a match, the cheating signal feature with the highest matching degree is determined as the feature of the abnormal signal, and the abnormal signal is determined to be a cheating signal; if there is no match, the feature set of the abnormal signal is input into the machine learning model for classification and prediction, and whether the abnormal signal is a cheating signal is determined according to the classification result; according to the determination result of the abnormal signal, combined with the business scenario and risk control rules, the business behavior corresponding to the abnormal signal is processed accordingly; the parsed features, determination results and corresponding processing measures of the abnormal signal are recorded in the log for subsequent data analysis and model optimization.

[0052] Furthermore, the process of updating the preset cheating signal feature library includes:

[0053] Determine whether the similarity score exceeds the threshold. If so, determine the abnormal signal as a known cheating signal. If the similarity score does not exceed the threshold, determine it as a suspected new cheating signal. Perform cluster analysis on the suspected new cheating signal, and determine whether it is a new cheating signal based on the clustering results. If so, add the unique feature vector of the abnormal wireless signal to the preset cheating signal feature library for update.

[0054] Furthermore, abnormal signal data is obtained, and features of the abnormal signal are extracted to obtain a unique feature vector of the signal; similarity is calculated between the extracted unique feature vector and the feature vector in a pre-established cheating signal feature library to obtain a similarity score; it is determined whether the similarity score exceeds a preset threshold, and if so, the abnormal signal is determined to be a known cheating signal; if the similarity score does not exceed the threshold, the abnormal signal is determined to be a suspected new cheating signal, triggering a new cheating signal confirmation process; cluster analysis is performed on the suspected new cheating signal, and it is determined whether it is a new cheating signal based on the clustering results; if it is confirmed to be a new cheating signal, the unique feature vector of the signal is extracted, added to the cheating signal feature library, and the feature library is updated in real time; based on the updated cheating signal feature library, machine learning algorithms such as support vector machines are used to perform real-time detection and judgment on subsequent abnormal signals, so as to continuously improve the detection rate of cheating signals.

[0055] Specifically, abnormal signal data acquisition is the first step in detecting cheating behavior. In practical applications, abnormal traffic patterns in communication networks can be captured by monitoring systems. For example, in an online examination system, it is found that a candidate answers questions abnormally fast and has a very high correct answer rate, which may be a potential cheating signal. Feature extraction is the process of converting the original signal into a numerical vector that can be quantified and compared. For the above abnormal answering behavior, features such as answering speed, correct rate, and answering mode can be extracted. Assume that the extracted feature vector is [0.95, 0.98, 0.87], which represents the abnormal degree of answering speed, correct rate, and answering mode, respectively. Similarity calculation is the process of comparing the extracted feature vector with a known cheating pattern. Commonly used methods include Euclidean distance, cosine similarity, etc. Assume that there is a vector [0.92, 0.97, 0.85] in the pre-established cheating signal feature library, and the calculated similarity score is 0.98. Threshold judgment is a key step in determining whether to classify a signal as a known cheating type. If the threshold is set to 0.95, the above signal will be judged as a known cheating behavior. This method can quickly identify cheating behaviors similar to known patterns and improve detection efficiency. For signals that do not exceed the threshold, the system will mark them as suspected new cheating signals. This processing method reflects the flexibility and adaptability of the system and can cope with the evolving cheating methods. Cluster analysis is an important means to identify new cheating signals. By grouping multiple suspected signals, potential new cheating patterns can be discovered. For example, a group of suspected signals are clustered using the K-means algorithm, and it is found that they form a new tight cluster, which may represent a new cheating method. After confirming it as a new cheating signal, the system will extract its feature vector and add it to the feature library. This step ensures that the system can continuously learn and adapt to new cheating methods. For example, a cheating method that uses specific software for remote assistance may have a feature vector of [0.88, 0.93, 0.91]. After adding it to the feature library, the system can identify similar cheating behaviors. The application of machine learning algorithms such as support vector machines enables the system to judge cheating behaviors more intelligently. Through training with a continuously updated feature library, the algorithm can learn more complex decision boundaries and improve the accuracy and robustness of detection. The advantage of this method is that it can handle high-dimensional feature spaces and adapt to the diversity and complexity of cheating methods. Real-time update and detection mechanisms ensure that the system can respond quickly to emerging cheating methods. This dynamic adaptability is crucial to maintaining the fairness of exams and can effectively curb the spread of cheating. By continuously improving the detection rate, the system can provide more reliable support for educational institutions and exam organizers to maintain the fairness and effectiveness of exams.

[0056] This embodiment also provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-mentioned method for actively capturing wireless signals for preventing cheating in a driver's theoretical test.

[0057] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for actively capturing wireless signals for preventing cheating in a driver's theoretical test are implemented.

[0058] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A wireless signal active capture method for anti-cheating in driver's theory test, characterized in that: The following steps are involved: Acquire the real-time data of wireless signals in the examination room and perform preprocessing, and filter out abnormal wireless signals from the preprocessed real-time data of wireless signals; Using a preset modulation identification algorithm to identify the modulation mode of the abnormal wireless signal to obtain a corresponding modulation type; Performing protocol feature analysis on the abnormal wireless signal based on the modulation type to obtain the protocol feature of the abnormal wireless signal; Based on the protocol feature and a preset cheating signal feature library, it is determined whether the corresponding abnormal wireless signal is a cheating signal.

2. The method for actively capturing wireless signals for preventing cheating in a driver's theoretical test according to claim 1 is characterized in that: The process of screening out abnormal wireless signals includes: A spectrum analysis algorithm is used to analyze the preprocessed real-time data of wireless signals to obtain the spectrum characteristics of the signals. According to the preset normal signal spectrum characteristic model, it is judged whether the spectrum characteristics are abnormal. If the spectrum characteristics are abnormal, it is determined as a suspected abnormal signal and marked. A signal strength detection algorithm is used to analyze the suspected abnormal signals, and abnormal wireless signals are screened out based on the signal strength analysis results and the spectrum characteristic judgment results.

3. The method for actively capturing wireless signals for preventing cheating in driver's theoretical test according to claim 1 is characterized in that: The process of using a preset modulation identification algorithm to identify the modulation mode of the abnormal wireless signal and obtaining the corresponding modulation type includes: Extract characteristic parameters of the abnormal wireless signal; compare the extracted characteristic parameters of the abnormal signal with pre-stored signal characteristics of known modulation types; if the degree of matching between the characteristic parameters of the abnormal wireless signal and the signal characteristics of the known modulation type exceeds a preset threshold, determine that the modulation type of the abnormal wireless signal is the corresponding known modulation type; if there is no characteristic matching item, use a support vector machine algorithm to identify the modulation type of the abnormal wireless signal based on the characteristic parameters of the abnormal wireless signal; if the modulation type of the abnormal signal cannot be identified using the support vector machine algorithm, use a convolutional neural network algorithm to identify the modulation type of the abnormal wireless signal.

4. The method for actively capturing wireless signals for preventing cheating in driver's theory test according to claim 1 is characterized in that: The process of obtaining the protocol characteristics of abnormal wireless signals includes: Based on the modulation type of the abnormal wireless signal, a corresponding protocol analysis module is obtained; based on the protocol analysis module, protocol features of the abnormal wireless signal are extracted to obtain protocol feature parameters; according to the protocol feature parameters, a deep learning algorithm is used to analyze the protocol level information of the abnormal wireless signal to obtain protocol features.

5. The method for actively capturing wireless signals for preventing cheating in a driver's theoretical test according to claim 4 is characterized in that: The process of obtaining protocol characteristics includes: Based on the protocol feature parameters, the abnormal wireless signal is preprocessed to extract the original information at the protocol level; the convolutional neural network algorithm is used to extract the features of the original information at the protocol level to obtain the protocol feature vector; the protocol feature vector is input into the long short-term memory network for serialization modeling; the output of the long short-term memory network is weighted fused through the attention mechanism to obtain the fused feature vector; the fused feature vector is input into the fully connected layer for information analysis to obtain detailed information at the protocol level as the protocol feature.

6. The method for actively capturing wireless signals for preventing cheating in driver's theory test according to claim 1 is characterized in that: The process of determining whether the corresponding abnormal wireless signal is a cheating signal based on the protocol feature and a preset cheating signal feature library includes: The protocol feature is compared with a preset cheating signal feature library. If there is a match, it is determined to be a cheating signal. If there is no match, deep feature extraction is performed on the protocol feature to obtain a unique feature vector of the abnormal wireless signal. The unique feature vector is calculated to have a similarity with a feature vector in a preset cheating signal feature library to obtain a similarity score. If the similarity score exceeds a threshold, it is a known cheating signal. If it does not exceed the threshold, it is a new cheating signal. If it is a new cheating signal, the preset cheating signal feature library is updated based on the unique feature vector of the abnormal wireless signal.

7. The method for actively capturing wireless signals for preventing cheating in a driver's theoretical test according to claim 6 is characterized in that: According to the business scenario of the abnormal wireless signal, a cheating signal feature library matching the business scenario is obtained; the protocol features are compared with the features in the cheating signal feature library one by one to determine whether there is a match; If there is a matching item, the cheating signal feature with the highest matching degree is determined as the feature of the abnormal wireless signal, and the abnormal wireless signal is determined to be a cheating signal.

8. The method for actively capturing wireless signals for preventing cheating in driver's theoretical test according to claim 6 is characterized in that: The process of updating the preset cheating signal signature library includes: Determine whether the similarity score exceeds the threshold. If so, determine the abnormal signal as a known cheating signal. If the similarity score does not exceed the threshold, determine it as a suspected new cheating signal. Perform cluster analysis on the suspected new cheating signal, and determine whether it is a new cheating signal based on the clustering results. If so, add the unique feature vector of the abnormal wireless signal to the preset cheating signal feature library for update.

9. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for actively capturing wireless signals for preventing cheating in a driver's theoretical test as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for actively capturing wireless signals for preventing cheating in a driver's theoretical test as described in any one of claims 1 to 8 are implemented.

Images