Anti-interference method in vehicle-mounted and traffic millimeter wave radar system

Abstract

The invention discloses an anti-interference method in a vehicle-mounted and traffic millimeter wave radar system, and the method comprises the steps: obtaining a broadband echo signal of an LFMCW radar through an orthogonal down-conversion receiver, and carrying out the digital down-conversion processing, thereby forming a narrowband signal; preprocessing the narrowband signal; performing high-dimensional feature coding and mode matching on the preprocessed narrowband signals by using a super-dimensional computing technology to realize accurate identification and classification of interference signals; performing spectrum sensing on the identified interference signal, identifying the characteristics of interference frequency, amplitude, phase and the like, and abstracting the characteristics to generate simplified characteristic representation; automatically generating a series of non-interference waveform parameters by using a large language model (LLM) based on the interference characteristic data; a field programmable gate array (FPGA) is utilized to quickly generate and adjust a waveform; and the radar performance is continuously monitored in real time. According to the invention, the anti-interference capability of the system and the adaptability of the system in a complex environment are improved, so that the driving safety is effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle-mounted and traffic millimeter-wave radars, and in particular to an anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system. Background Art

[0002] In automotive and traffic millimeter-wave radar systems, accurate detection and identification of interference signals is a key factor in the system's anti-interference capability. Currently, millimeter-wave radar technology is widely used in scenarios such as autonomous driving and traffic monitoring to improve the accuracy and reliability of environmental perception. However, complex electromagnetic environments and signals from other radar systems can interfere with millimeter-wave radar to varying degrees, making it difficult for the system to provide stable and accurate detection and identification results in strong interference environments.

[0003] In traditional electronic reconnaissance applications, quadrature down-conversion mixers are used to process wideband signals. However, after conversion, the loss of signal tuning information weakens the spectral characteristics, making the resulting narrowband signal less likely to retain the interference signature of the original signal. Because narrowband signals cannot effectively represent the complete spectrum, traditional interference identification methods are inapplicable, making it difficult to distinguish and classify specific interference signals. Further exacerbating this problem, existing radar systems often rely on static or empirical waveform design, which is difficult to adapt to changing interference environments, limiting the system's anti-interference and environmental adaptability.

[0004] Radar waveform design directly impacts the system's detection performance and adaptability. Static waveforms cannot dynamically adjust their time-frequency characteristics in the face of rapidly changing interference scenarios. To address this challenge, the application of large-scale artificial intelligence models, particularly large language models (LLMs), has opened up new possibilities for intelligent radar waveform design. LLMs enable the system to rapidly abstract and analyze interference signal characteristics through natural language processing, dynamically optimizing waveform design and enhancing the radar's anti-interference capabilities and environmental adaptability. Summary of the Invention

[0005] Purpose of the invention: The present invention provides an anti-interference method for vehicle-mounted and traffic millimeter-wave radar systems, which improves the system's anti-interference capability and adaptability in complex environments, thereby effectively ensuring driving safety.

[0006] Technical solution: The anti-interference method for a vehicle-mounted and traffic millimeter-wave radar system described in the present invention specifically includes the following steps:

[0007] (1) The broadband echo signal of the LFMCW radar is obtained through an orthogonal down-conversion receiver and converted into a narrowband signal after digital down-conversion processing;

[0008] (2) Preprocess narrowband signals to recover key information and improve signal quality;

[0009] (3) Using hyperdimensional computing technology to perform high-dimensional feature encoding and pattern matching on the pre-processed narrowband signal to achieve accurate identification and classification of interference signals;

[0010] (4) Spectrum sensing of the identified interference signal is performed and abstracted to generate a simplified feature representation;

[0011] (5) Based on the interference feature data obtained in step (4), a series of interference-free waveform parameters are automatically generated using a large language model (LLM);

[0012] (6) Rapidly generate and adjust waveforms using field programmable gate arrays (FPGAs);

[0013] (7) Conduct continuous real-time monitoring of radar performance, focusing on key indicators.

[0014] Furthermore, the implementation process of step (1) is as follows:

[0015] The LFMCW signal receiver receives a broadband echo signal from the target, including key information about the target's distance, speed, and surrounding environment. The signal expression is:

[0016] S RX (t) = S TX (t)·H(t)+N(t)

[0017] Among them, S RX (t) is the received signal, S TX (t) is the transmitted signal, H(t) represents the response of the target and the environment, and represents the attenuation, delay, and phase change of the signal after being reflected by the target and propagating through the environment; N(t) is the environmental noise or the internal noise of the system, which is manifested as the interference component in the signal;

[0018] Through orthogonal down-conversion processing, the broadband echo signal is converted into a low-frequency narrowband signal, thereby reducing the signal bandwidth and adapting to subsequent narrowband analysis, namely:

[0019]

[0020] Among them, S IQ (t) is the I / Q signal after processing, w c is the carrier frequency, is the complex exponential function used in the downconversion operation.

[0021] Furthermore, the implementation process of step (2) is as follows:

[0022] Filter a narrowband signal using a bandpass filter:

[0023] S filtered (t) = S narrow (t)·F(t)

[0024] Among them, S narrow (t) is a narrowband signal, F(t) is a bandpass filter, S filtered (t) is the filtered signal; the noise in the received signal is suppressed by adaptive filtering or wavelet transform; the pre-processed signal is further optimized by using signal enhancement algorithm, including amplitude adjustment, phase correction and dynamic range expansion.

[0025] Furthermore, the implementation process of step (3) is as follows:

[0026] The pre-processed signal S after orthogonal down-conversion processing and signal enhancement is preprocessed (t) Input into the hyperdimensional computing classifier model and output the result:

[0027] C=f(S preprocessed (t))

[0028] Among them, C represents the classification result, and f is the classifier function that maps the input preprocessed signal to the corresponding category label.

[0029] Furthermore, the implementation process of step (4) is as follows:

[0030] Perform spectrum analysis on the interference signal and convert it into frequency domain signal:

[0031] S(f)=|X(f)| 2

[0032] Here, X(f) represents the spectrum of the interference signal, and S(f) is the corresponding spectrum density. The spectrum density is analyzed to extract the key features of the interference signal. Feature extraction is performed on the analyzed spectrum information, focusing on parameters related to the interference signal. The extracted features include the center value of the frequency, the maximum amplitude, and the rate of change of the phase. The extracted features are abstracted using a large language model (LLM). The complex spectrum data is converted into high-level operational parameters to form a systematic signal feature description.

[0033] Furthermore, the interference-free waveform parameters in step (5) include carrier frequency, modulation mode, and pulse width.

[0034] Furthermore, the implementation process of step (5) is as follows:

[0035] LLM analyzes the current electromagnetic environment based on abstract interference characteristics and selects the most suitable waveform parameters. It then uses an optimization algorithm to further optimize the generated waveform parameters. The optimization goal is to improve the radar system's adaptability in specific environments while reducing false alarm rates and increasing detection accuracy. The waveform optimization parameters are:

[0036] minJ(θ)

[0037] Among them, minJ(θ) is the cost function, which represents the optimization goal of waveform design, and θ is the waveform parameter vector; the optimization algorithm is used to find the optimal θ value to minimize the cost function.

[0038] Furthermore, the implementation process of step (6) is as follows:

[0039] The optimized waveform parameters are input into the FPGA and processed in real time using a specific waveform design algorithm to generate a radar waveform suitable for specific application requirements.

[0040] The adjusted waveform is applied to the radar system in real time through the system's synchronous transmission module, ensuring the time consistency of the transmitted and received signals.

[0041] Furthermore, the implementation process of step (7) is as follows:

[0042] In the performance monitoring and feedback phase, the key performance indicators monitored are expressed using the following formula:

[0043] Performance Metrics = {D, V, R}

[0044] Among them, D represents ranging accuracy, V represents velocity accuracy, and R represents resolution, which is usually related to the bandwidth and frequency of the signal and can be quantified by the signal's spectrum width. These monitoring results will be collected in real time through the data acquisition module and analyzed to evaluate the performance of the radar system in a dynamic environment.

[0045] Beneficial effects: Compared with the existing technology, the present invention has the following beneficial effects: The present invention innovatively combines hyperdimensional computing with a large language model to build a complete anti-interference processing closed loop; first, the broadband echo signal of the LFMCW radar is obtained through an orthogonal down-conversion receiver, and a narrowband baseband signal is formed after digital down-conversion processing; the interference recognition engine based on hyperdimensional computing performs high-dimensional feature encoding and pattern matching on the signal to achieve microsecond-level interference detection and classification, and can accurately identify various interference types such as sweep frequency and noise and their time-frequency characteristics; the recognition results are passed to the LLM through a structured description interface, and based on the semantic-level understanding of the electromagnetic environment, the interference distribution characteristics are dynamically analyzed and automatically The system selects the optimal time-frequency operating region and generates adaptive waveform parameters (including frequency modulation slope, bandwidth, pulse width, etc.). Programmable hardware reconstructs the transmitted waveform in real time, forming a complete closed loop of "perception-decision-execution." This method overcomes the limitations of traditional anti-interference technology. The high-dimensional feature space provided by hyperdimensional computing ensures robust interference identification, while the cognitive reasoning capabilities of LLM enable intelligent spectrum avoidance in complex electromagnetic environments. Through the deep integration of signal processing and artificial intelligence, this method provides an efficient and intelligent anti-interference solution for vehicle-mounted and traffic millimeter-wave radar systems, significantly improving the radar's reliability and adaptability in complex electromagnetic environments. The specific steps of the two methods are as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 Flowchart of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be further described in detail below with reference to the accompanying drawings.

[0048] like Figure 1 As shown, the present invention proposes an anti-interference method for vehicle-mounted and traffic millimeter-wave radar systems, which mainly includes interference detection and identification for orthogonal down-conversion narrowband receivers and LLM-based waveform online design; specifically, the following steps are included:

[0049] Step 1: The LFMCW (Linear Frequency Modulated Continuous Wave) signal receiver receives a broadband echo signal from the target. This echo signal contains key information about the target's distance, speed, and surrounding environment. The signal expression is:

[0050] S RX (t) = S TX (t)·H(t)+N(t)

[0051] Among them, S RX (t) is the received signal, S TX(t) is the transmitted signal, H(t) represents the response of the target and the environment. This function represents the attenuation, delay, and phase change of the signal after it is reflected by the target and propagated through the environment. N(t) is the noise, which is usually environmental noise or internal noise of the system and appears as an interference component in the signal.

[0052] To extract and analyze this information, the broadband echo signal is first converted into a low-frequency narrowband signal through orthogonal down-conversion processing, thereby reducing the signal bandwidth and adapting to subsequent narrowband analysis. That is:

[0053]

[0054] Among them, S IQ (t) represents the processed I / Q signal, w c is the carrier frequency, is the complex exponential function used in the downconversion operation.

[0055] This process uses a mixer to downconvert the signal to a quadrature channel to obtain I / Q components with phase and amplitude information, laying the foundation for subsequent interference detection and signal classification. This step effectively reduces the signal frequency while preserving the target information, making subsequent preprocessing and classification analysis more efficient.

[0056] Step 2: Perform a series of preprocessing operations on the narrowband signal obtained through the orthogonal down-conversion process to restore key information in the signal and improve the signal quality.

[0057] First, a bandpass filter is used to filter the narrowband signal to remove unnecessary high-frequency and low-frequency noise, ensuring that only the frequency components relevant to the target are retained. This step helps improve the accuracy of subsequent signal analysis and makes the key features of the signal more prominent. The filtering formula is as follows:

[0058] S filtered (t) = S narrow (t)·F(t)

[0059] Next, through advanced noise suppression techniques such as adaptive filtering or wavelet transforms, the system suppresses noise in the received signal, reducing the impact of environmental noise and internal system noise on the signal, thereby improving the signal-to-noise ratio (SNR). This noise suppression not only improves signal clarity but also enhances the reliability of subsequent interference detection.

[0060] Finally, signal enhancement algorithms are used to further optimize the preprocessed signal, including amplitude adjustment, phase correction, and dynamic range expansion. These processes ensure the identifiability and resolution of key information within the signal, providing high-quality input for subsequent hyperdimensional computing classifier analysis. These integrated processes lay a solid foundation for accurate identification and classification of interference signals.

[0061] Step 3: The system uses hyperdimensional computing technology to conduct an in-depth analysis of the pre-processed orthogonal signals. The goal of this step is to accurately identify and classify the interference signals through the trained classifier model so that appropriate measures can be taken in subsequent processing steps.

[0062] First, the pre-processed signal S after orthogonal down-conversion processing and signal enhancement is preprocessed (t) is input into the classifier model. The classifier model has been optimized through a large amount of training data and has the ability to identify various interference signals.

[0063] The output of the classifier can be expressed as follows:

[0064] C=f(S preprocessed (t))

[0065] Here, C represents the classification result, and f is the classifier function, which maps the input preprocessed signal to the corresponding category label. Through this process, the system can accurately identify the interference present in the signal and classify it into different types, providing important basis for subsequent signal processing and decision-making.

[0066] To further improve recognition accuracy and efficiency, a hyperdimensional computing classifier leverages the advantages of high-dimensional feature space to process complex signal patterns. Through efficient feature extraction and classification, the system accurately identifies interference signals, enhancing the radar system's overall performance and anti-interference capabilities.

[0067] Step 4: In the spectrum sensing and abstraction stage, the interference signal is immediately analyzed in depth to extract key signal features.

[0068] First, the collected interference signal is analyzed using techniques such as Fast Fourier Transform (FFT) to convert it into a frequency domain signal. The formula is as follows:

[0069] S(f)=|X(f)| 2

[0070] Here, X(f) represents the spectrum of the interference signal, and S(f) is the corresponding spectral density. This formula describes the energy distribution of the signal in the frequency domain and calculates the strength of each frequency component, thereby helping to identify the characteristics of the interference signal. By analyzing the spectral density, the system can effectively extract key characteristics of the interference signal, such as frequency and amplitude, providing a foundation for subsequent signal processing and classification.

[0071] Next, the system extracts features from the analyzed spectral information, focusing on parameters related to the interfering signal. These features may include the center frequency, maximum amplitude, and rate of change of phase, all of which are crucial for understanding the nature of the interfering signal. Based on this, the system abstracts the extracted features using a large language model (LLM). Using natural language processing techniques, the LLM transforms complex spectral data into high-level, actionable parameters, thereby forming a systematic description of the signal characteristics.

[0072] The core of this phase lies in combining technical analysis with intelligent processing. LLM's powerful semantic understanding capabilities enable it to identify the patterns and characteristics of interference signals and automatically generate specific parameters for subsequent waveform design and adjustment. This transformation process not only improves signal processing efficiency but also provides a solid data foundation for dynamic optimization of the radar system, enabling it to maintain efficiency and flexibility in a constantly changing electromagnetic environment.

[0073] Step 5: During the waveform parameter design and optimization phase, the system leverages the interference signature data obtained through spectrum sensing and abstraction to automatically generate a series of interference-free waveform parameters using a large language model (LLM). These parameters include, but are not limited to, key characteristics such as carrier frequency, modulation method, and pulse width, designed to maximize the radar signal's anti-interference capabilities and performance.

[0074] First, LLM analyzes the current electromagnetic environment based on abstracted interference signatures and automatically selects the most appropriate waveform parameters. This process is both efficient and flexible, as LLM can factor in various external interference factors in real time to generate a targeted waveform design.

[0075] Next, the system will use optimization algorithms to further optimize the generated waveform parameters. These optimization algorithms may include genetic algorithms, particle swarm optimization, etc., which can evaluate and compare the performance of different waveform configurations to ensure that the selected parameters have the best signal quality and resolution in actual applications. The goal of optimization is to improve the adaptability of the radar system in a specific environment, while reducing the false alarm rate and improving the accuracy of detection. The waveform optimization parameters are

[0076] minJ(θ)

[0077] Here, min J(θ) is the cost function, representing the optimization objective for waveform design, and θ is the waveform parameter vector. The goal of this formula is to use an optimization algorithm to find the optimal value of θ to minimize the cost function. This typically involves evaluating system performance, such as signal interference immunity and accuracy, to ensure that the generated waveform parameters have optimal performance and adaptability in practical applications.

[0078] Through this comprehensive process, the system not only achieves dynamic waveform adjustment and optimization, but also maintains efficient radar signal processing capabilities in complex and changing electromagnetic environments. This automated and intelligent waveform design approach significantly improves the overall performance of the radar system and lays the foundation for future intelligent transportation and automotive applications.

[0079] Step 6: During the online waveform generation and adjustment phase, the system uses an FPGA (field programmable gate array) to rapidly generate and adjust the waveform. This process occurs immediately after waveform parameter design and optimization to ensure rapid response to environmental changes.

[0080] First, the optimized waveform parameters are input into the FPGA and processed in real time using a specific waveform design algorithm to generate a radar waveform suitable for specific application requirements.

[0081] The adjusted waveform is then applied to the radar system in real time via the system's synchronous transmission module, ensuring the time consistency of transmitted and received signals. This mechanism enables the radar system to dynamically adapt to complex electromagnetic environments, improving detection accuracy and anti-interference capabilities, thereby enhancing the performance and adaptability of automotive radar.

[0082] Step 7: During the performance monitoring and feedback phase, the system continuously monitors radar performance in real time, focusing on key indicators such as ranging accuracy, speed accuracy, and resolution. During the performance monitoring and feedback phase, the key performance indicators monitored can be expressed using the following formula:

[0083] Performance Metrics = {D, V, R}

[0084] Where D represents distance accuracy, V represents velocity accuracy, and R represents resolution, which is typically related to the signal's bandwidth and frequency and can be quantified by the signal's spectral width. These monitoring results are collected in real time by the data acquisition module and analyzed to evaluate the radar system's performance in dynamic environments.

[0085] If performance degradation or non-compliance with pre-set standards is detected, the system automatically feeds this feedback into a large language model (LLM). The LLM analyzes this real-time data to further optimize waveform parameters. This closed-loop feedback mechanism allows the radar system to adapt to environmental changes, ensuring continuous optimization of performance, anti-jamming capabilities, and reliability.

[0086] The present invention has been described in detail above with reference to specific embodiments. However, these descriptions should not be construed as limiting the present invention. Those skilled in the art will appreciate that various equivalent substitutions, modifications, or improvements may be made to the technical solutions and implementations of the present invention without departing from the spirit and scope of the present invention, all of which fall within the scope of the present invention. The scope of protection of the present invention shall be determined by the appended claims.

Claims

1. An anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system, characterized in that: The following steps are involved: (1) The broadband echo signal of the LFMCW radar is obtained through an orthogonal down-conversion receiver and converted into a narrowband signal after digital down-conversion processing; (2) Preprocess narrowband signals to recover key information and improve signal quality; (3) Using hyperdimensional computing technology to perform high-dimensional feature encoding and pattern matching on the pre-processed narrowband signal to achieve accurate identification and classification of interference signals; (4) Spectrum sensing of the identified interference signal is performed and abstracted to generate a simplified feature representation; (5) Based on the interference feature data obtained in step (4), a series of interference-free waveform parameters are automatically generated using a large language model (LLM); (6) Rapidly generate and adjust waveforms using field programmable gate arrays (FPGAs); (7) Conduct continuous real-time monitoring of radar performance, focusing on key indicators.

2. The anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system according to claim 1, characterized in that: The implementation process of step (1) is as follows: The LFMCW signal receiver receives a broadband echo signal from the target, including key information about the target's distance, speed, and surrounding environment. The signal expression is: S RX (t)=S TX (t)·H(t)+N(t) Among them, S RX (t) is the received signal, S TX (t) is the transmitted signal, H(t) represents the response of the target and the environment, and represents the attenuation, delay, and phase change of the signal after being reflected by the target and propagating through the environment; N(t) is the environmental noise or the internal noise of the system, which is manifested as the interference component in the signal; Through orthogonal down-conversion processing, the broadband echo signal is converted into a low-frequency narrowband signal, thereby reducing the signal bandwidth and adapting to subsequent narrowband analysis, namely: Among them, S IQ (t) is the I / Q signal after processing, w c is the carrier frequency, is the complex exponential function used in the downconversion operation.

3. The anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system according to claim 1, characterized in that: The implementation process of step (2) is as follows: Filter a narrowband signal using a bandpass filter: S filtered (t)=S narrow (t)·F(t) Among them, S narrow (t) is a narrowband signal, F(t) is a bandpass filter, S filtered (t) is the filtered signal; the noise in the received signal is suppressed by adaptive filtering or wavelet transform; the pre-processed signal is further optimized by using signal enhancement algorithm, including amplitude adjustment, phase correction and dynamic range expansion.

4. The anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system according to claim 1, characterized in that: The implementation process of step (3) is as follows: The pre-processed signal S after orthogonal down-conversion processing and signal enhancement is preprocessed (t) Input into the hyperdimensional computing classifier model and output the result: C=f(S preprocessed (t)) Among them, C represents the classification result, and f is the classifier function that maps the input preprocessed signal to the corresponding category label.

5. The anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system according to claim 1, characterized in that: The implementation process of step (4) is as follows: Perform spectrum analysis on the interference signal and convert it into frequency domain signal: S(f)=|X(f)| 2 Here, X(f) represents the spectrum of the interference signal, and S(f) is the corresponding spectrum density. The spectrum density is analyzed to extract the key features of the interference signal. Feature extraction is performed on the analyzed spectrum information, focusing on parameters related to the interference signal. The extracted features include the center value of the frequency, the maximum amplitude, and the rate of change of the phase. The extracted features are abstracted using a large language model (LLM). The complex spectrum data is converted into high-level operational parameters to form a systematic signal feature description.

6. The anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system according to claim 1, characterized in that: The interference-free waveform parameters in step (5) include carrier frequency, modulation mode, and pulse width.

7. The anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system according to claim 1, characterized in that: The implementation process of step (5) is as follows: LLM analyzes the current electromagnetic environment based on abstract interference characteristics and selects the most suitable waveform parameters. It then uses an optimization algorithm to further optimize the generated waveform parameters. The optimization goal is to improve the radar system's adaptability in specific environments while reducing false alarm rates and increasing detection accuracy. The waveform optimization parameters are: min J(θ) Where min J(θ) is the cost function, which represents the optimization goal of the waveform design, and θ is the waveform parameter vector. The optimization algorithm is used to find the optimal θ value to minimize the cost function.

8. The anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system according to claim 1, characterized in that: The implementation process of step (6) is as follows: The optimized waveform parameters are input into the FPGA and processed in real time using a specific waveform design algorithm to generate a radar waveform suitable for specific application requirements. The adjusted waveform is applied to the radar system in real time through the system's synchronous transmission module, ensuring the time consistency of the transmitted and received signals.

9. The anti-interference method in a vehicle-mounted and traffic millimeter-wave radar system according to claim 1, characterized in that: The implementation process of step (7) is as follows: In the performance monitoring and feedback phase, the key performance indicators monitored are expressed using the following formula: Performance Metrics = {D, V, R} Among them, D represents ranging accuracy, V represents velocity accuracy, and R represents resolution, which is usually related to the bandwidth and frequency of the signal and can be quantified by the signal's spectrum width. These monitoring results will be collected in real time through the data acquisition module and analyzed to evaluate the performance of the radar system in a dynamic environment.

Images