Radar main lobe interference cancellation system and method based on situation awareness

Abstract

The invention belongs to the technical field of radar interference protection, and particularly discloses a radar main lobe interference cancellation system and method based on situation awareness, and the system comprises a data collection module which is used for collecting an echo signal of a receiving link; the data preprocessing module is used for carrying out data preprocessing on the echo signals so as to construct a data set for interference sensing; the interference sensing module is used for performing time sequence sensing on the echo signals based on a trained online sensing model to obtain time sequence information of the echo signals; the online sensing model is obtained based on data set training; the weight decision module is used for selecting a weight state according to the time sequence information of the echo signal; and the interference cancellation module is used for cancelling main lobe interference of the main channel echo signal by using the sampling channel echo signal, performing judgment in combination with a weight state, and reserving a useful signal in the echo signal. According to the invention, main lobe interference can be suppressed, and meanwhile, the probability of deleting useful signals by mistake is remarkably reduced.

Description

Technical Field

[0001] This application belongs to the field of radar jamming protection technology, and more specifically, relates to a radar main lobe jamming cancellation system and method based on situational awareness. Background Technology

[0002] With the development of radar countermeasures technology, the widespread application of deceptive jammers has significantly reduced the detection performance of radar systems. In particular, with the development of Digital Radio Frequency Memory (DRFM) technology, jammers can rapidly detect radar signals and then copy, modulate, and forward them to create high-fidelity false targets. This type of jamming has a high similarity to the target signal, confusing or obscuring the target signal in multiple domains, posing a significant challenge to radar target detection and situational awareness in complex electromagnetic environments. Furthermore, when deceptive jamming enters the receiver from the radar main lobe, it also introduces interference energy gain and aliasing in the spatial angle, which is one of the most pressing problems to be solved in the radar field. Therefore, researching methods to suppress radar main lobe deceptive jamming has significant practical implications.

[0003] In recent years, the development and maturation of artificial intelligence technologies, especially deep learning and related theories, have greatly improved radar situational awareness, autonomous learning, autonomous reasoning, and decision-making capabilities. Deep learning technology, by establishing an interaction model between the intelligent agent and its environment, enables the agent to fully explore environmental information and improve its behavioral decision-making capabilities. Introducing deep learning into the field of radar anti-jamming allows for the optimization of radar anti-jamming behavioral strategies by collecting interaction information between the radar and jammers, thereby enhancing the radar's anti-jamming behavioral decision-making capabilities.

[0004] However, current related technologies, such as traditional adaptive cancellation techniques or deep learning applied to jammers, still suffer from the problem of useful signal loss due to interference cancellation, resulting in poor interference cancellation performance. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this application is to provide a radar main lobe interference cancellation system and method based on situational awareness, aiming to solve the problem of useful signal loss caused by interference cancellation techniques in existing technologies.

[0006] To achieve the above objectives, in a first aspect, this application provides a radar main lobe interference cancellation system based on situational awareness, comprising: a data acquisition module, a data preprocessing module, an interference sensing module, a weight decision module, and an interference cancellation module; The data acquisition module includes a main channel and a sampling channel. The input ends of both the main channel and the sampling channel are connected to the radar receiving link and are used to acquire the echo signal from the receiving link. The input end of the data preprocessing module is connected to the main channel and is used to preprocess the echo signal to construct a dataset for interference sensing. The input terminal of the interference sensing module is connected to the data preprocessing module, and is used to perform time-series sensing on the echo signal based on the trained online sensing model to obtain the time-series information of the echo signal; the online sensing model is trained based on the dataset. The input terminal of the weight decision module is connected to the interference sensing module and is used to select the weight state according to the timing information of the echo signal. The first input terminal of the interference cancellation module is connected to the data acquisition module, and the second input terminal of the interference cancellation module is connected to the weight decision module. The interference cancellation module is used to cancel the main lobe interference of the main channel echo signal using the sampling channel echo signal, and to make a judgment based on the weight state to retain the useful signal in the echo signal.

[0007] Optionally, the weight states include 0 and 1; The weight decision module is specifically used for: Based on the timing information, it is determined that the echo signal is in an interference timing sequence. The weight state is set to 1, and a weight update operation is performed. Based on the timing information, the echo signal is determined to be in a non-interference timing sequence. The weight state is set to 0, and the cancellation weight corresponding to the timing information of the previous received signal is maintained.

[0008] Optionally, the interference perception module includes a feature extraction unit and an interference decision unit; the online perception model includes an autoencoder and a deep support vector description network. The feature extraction unit is used to perform high-dimensional feature extraction on the dataset based on an autoencoder and a deep support vector description network to obtain high-dimensional features of the signal. The interference decision unit is used to set decision boundaries to sense the collected samples, compare the high-dimensional features of the signal with the decision boundaries, and output the sensing result of the echo signal. The decision boundary is determined based on the hypersphere radius and center hyperparameter of the deep support vector description network obtained through training.

[0009] Optionally, the autoencoder includes an encoder and a decoder; The encoder includes a residual unit layer and a fully connected layer, which are used to extract features and reduce the dimensionality of the dataset corresponding to the desired signal to obtain the encoded feature vector. The residual unit layer includes three convolutional layers, a filler layer, and a fusion layer; The decoder has a structure symmetrical to the encoder and is used to recover the encoded feature vector into the desired signal.

[0010] Optionally, the interference sensing module includes a data caching unit, a data integration unit, a model training unit, and a parameter storage unit; The data buffer unit is used to temporarily store the time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal of the main channel; The data integration unit is used to call the time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal stored in the data buffer unit; The model training unit is used to input different time-domain IQ sequences and frequency-domain IQ sequences into the autoencoder and the deep support vector data description model for training, so as to obtain the optimal model parameters. The parameter storage unit is used to store the model parameters of the trained autoencoder and the model parameters of the deep support vector data description model.

[0011] Optionally, the interference sensing module further includes a parameter receiving unit and an online sensing unit; The parameter receiving unit is used to receive deep learning network model parameters from the parameter storage unit; The online sensing unit is used to sense the radar time-domain IQ signal and radar frequency-domain IQ signal received in real time based on the parameters of the deep learning network model, so as to determine the signal components and the receiving timing.

[0012] Optionally, the weight decision module includes a power detection unit and a weight calculation unit; The power detection unit is used to detect the power of the signals received by the main channel and the sampling channel, respectively. The weight calculation unit is used to calculate the cancellation weight of the sampling channel based on the power of the received signal and the timing information of the sensing signal output by the sensing module.

[0013] Optionally, the data preprocessing module includes a modulation and mixing unit, a digital filtering unit, a downsampling unit, a Fourier transform unit, and a matrix construction unit; The modulation and mixing unit is used to modulate and mix the echo signal to generate a time-domain IQ sequence; The digital filtering unit is used to filter the time-domain IQ sequence to suppress out-of-band noise; The downsampling unit is used to downsample the time-domain IQ sequence to reduce the data rate; The Fourier transform unit is used to perform Fourier transform on the time-domain IQ sequence to obtain the frequency-domain IQ sequence; The matrix construction unit is used to construct a time-frequency IQ matrix based on the time-domain IQ sequence and the frequency-domain IQ sequence.

[0014] Optionally, the data acquisition module is specifically used to perform analog-to-digital acquisition on the intermediate frequency echo signal from the receiving link according to the set sampling rate and intermediate frequency parameters, so as to obtain the main channel echo signal and the sampling channel echo signal.

[0015] Secondly, this application also provides a radar main lobe interference cancellation method based on situational awareness, including: Acquire echo signals from the receiving link; The echo signal is preprocessed to construct a dataset for interference sensing; The echo signal is subjected to time-series sensing based on a trained online sensing model to obtain the time-series information of the echo signal; the online sensing model is trained based on the dataset. The weight state is selected based on the timing information of the echo signal; The main lobe interference of the main channel echo signal is canceled by using the echo signal from the sampling channel, and the useful signal in the echo signal is retained by combining the weight state.

[0016] Optionally, the time-aware process includes: High-dimensional features of the signal are extracted from the dataset based on an autoencoder and a deep support vector description network. The decision boundary is set to sense the collected samples, the high-dimensional features of the signal are compared with the decision boundary, and the sensing result of the echo signal is output. The decision boundary is determined based on the hypersphere radius and center hyperparameter of the deep support vector description network obtained through training.

[0017] Optionally, the method further includes: The time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal of the main channel are temporarily stored in the data buffer unit. The data integration unit calls the time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal stored in the data cache unit; Different time-domain IQ sequences and frequency-domain IQ sequences are input into the autoencoder and deep support vector data description model through the model training unit to obtain the optimal model parameters; The parameter storage unit stores the model parameters of the trained autoencoder and the model parameters of the deep support vector data describing the model.

[0018] Thirdly, this application provides an electronic device, comprising: at least one memory for storing a program; and at least one processor for executing the program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to execute the method described in the first aspect or any possible implementation thereof.

[0019] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to perform the method described in the first aspect or any possible implementation thereof.

[0020] Fifthly, this application provides a computer program product that, when run on a processor, causes the processor to perform the method described in the first aspect or any possible implementation thereof.

[0021] It is understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0022] Overall, the technical solutions conceived in this application have the following beneficial effects compared with the prior art: The data preprocessing module of this application constructs a joint feature matrix through time-frequency analysis to accurately separate the time-frequency distribution of interference and signal. The interference sensing module identifies interference timing in real time based on a trained online model (autoencoder + deep support vector data description network), avoiding signal misjudgment caused by fixed thresholds in traditional methods. The weight decision module dynamically adjusts the cancellation weights based on the sensing results (weight 1 for interference periods and 0 for interference-free periods), ensuring selective cancellation only of interference components while preserving the integrity of useful signals during non-interference periods. Through this combination of timing sensing and dynamic weight control, the system significantly reduces the probability of false deletion of useful signals while suppressing main lobe interference. Attached Figure Description

[0023] Figure 1 This is one of the structural schematic diagrams of the radar main lobe interference cancellation system based on situational awareness provided in the embodiments of this application; Figure 2 This is an architecture diagram of the interference sensing module provided in an embodiment of this application; Figure 3 This is a diagram of the self-encoder architecture provided in the embodiments of this application; Figure 4 This is a diagram of the deep support vector data description network architecture provided in the embodiments of this application; Figure 5 This is a flowchart of the training process for the autoencoder and deep support vector data description model provided in the embodiments of this application; Figure 6 This is a diagram of the weighting strategy cancellation architecture provided in the embodiments of this application; Figure 7 This is the second schematic diagram of the radar main lobe interference cancellation system based on situational awareness provided in the embodiments of this application; Figure 8 This is a flowchart illustrating the radar main lobe interference cancellation method based on situational awareness provided in the embodiments of this application. Figure 9 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0025] In this article, the term "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The symbol " / " in this article indicates that the related objects are in an "or" relationship; for example, A / B means A or B.

[0026] The terms "first" and "second," etc., used in the specification and claims herein are used to distinguish different objects, not to describe a specific order of objects. For example, "first response message" and "second response message," etc., are used to distinguish different response messages, not to describe a specific order of response messages.

[0027] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0028] In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more, for example, multiple processing units means two or more processing units, multiple elements means two or more elements, etc.

[0029] The embodiments of this application are described below with reference to the accompanying drawings.

[0030] Reference Figure 1This application provides a radar main lobe interference cancellation system based on situational awareness, including: a data acquisition module 110, a data preprocessing module 120, an interference sensing module 130, a weight decision module 140, and an interference cancellation module 150. The data acquisition module 110 includes a main channel and a sampling channel. The input ends of the main channel and the sampling channel are both connected to the radar receiving link and are used to acquire the echo signal of the receiving link. The input terminal of the data preprocessing module 120 is connected to the main channel and is used to preprocess the echo signal to construct a dataset for interference sensing. The input terminal of the interference sensing module 130 is connected to the data preprocessing module 120, and is used to perform time-series sensing on the echo signal based on the trained online sensing model to obtain the time-series information of the echo signal; the online sensing model is trained based on the dataset. The input terminal of the weight decision module 140 is connected to the interference sensing module 130, and is used to select the weight state according to the timing information of the echo signal. The first input terminal of the interference cancellation module 150 is connected to the data acquisition module 110, and the second input terminal of the interference cancellation module is connected to the weight decision module 140. The interference cancellation module is used to cancel the main lobe interference of the main channel echo signal using the sampling channel echo signal, and to make a judgment based on the weight state, so as to retain the useful signal in the echo signal.

[0031] Specifically, in this embodiment, the data acquisition module 110 is the signal input terminal of the system, consisting of a main channel and a sampling channel. The main channel directly receives radar echo signals for subsequent interference cancellation and target detection; the sampling channel synchronously acquires signals to provide reference samples for interference cancellation. Both channels are connected to the radar receiving link to ensure signal synchronization and consistency, providing a high-quality data foundation for subsequent processing.

[0032] In some implementations, the data acquisition module is used to perform analog-to-digital (AD) acquisition of the intermediate frequency echo signal from the receiving link according to the set sampling rate and intermediate frequency parameters.

[0033] The data preprocessing module 120 is responsible for the preliminary processing of the main channel echo signal, including filtering, noise reduction, and normalization, to eliminate hardware-introduced noise and clutter. Through time-frequency analysis or pulse compression techniques, this module constructs a dataset that can be used for interference sensing, providing standardized input for the online sensing model.

[0034] The interference sensing module 130 performs time-series analysis on the preprocessed signal based on a trained online sensing model to identify the time-domain and frequency-domain characteristics of main lobe interference. This model, trained on historical datasets, adaptively learns the dynamic changes in interference and outputs a sensing signal containing the type and intensity of interference. In this embodiment, the online sensing model incorporates an online learning mechanism to dynamically update model parameters to adapt to environmental changes.

[0035] The weight decision module 140 dynamically selects the optimal weight state to configure the interference cancellation strategy based on the timing information output by the interference sensing module. For example, a high suppression weight is used during periods of strong interference, while the cancellation strength is reduced during periods of weak or no interference to avoid signal distortion. Weight selection can be based on preset rules. Weight states include 0 and 1; Based on the timing information, it is determined that the echo signal is in an interference timing sequence. The weight state is set to 1, and a weight update operation is performed. Based on the timing information, the echo signal is determined to be in a non-interference timing sequence. The weight state is set to 0, and the cancellation weight corresponding to the timing information of the previous received signal is maintained.

[0036] This module also features state memory, enabling it to predict future weight requirements by incorporating historical interference patterns, thereby reducing decision-making delays. The output of the weight decision directly impacts the interference cancellation module, ensuring the system responds quickly in complex electromagnetic environments.

[0037] The interference cancellation module 150 receives echo signals from the main channel and the sampling channel, and uses an adaptive filtering algorithm to cancel main lobe interference. Based on the weighting state provided by the weighting decision module, this module dynamically adjusts the cancellation depth, suppressing interference while preserving the useful signal to the maximum extent possible.

[0038] Optionally, the interference perception module includes a feature extraction unit and an interference decision unit; the online perception model includes an autoencoder and a deep support vector description network. The feature extraction unit is used to perform high-dimensional feature extraction on the dataset based on an autoencoder and a deep support vector description network to obtain high-dimensional features of the signal. The interference decision unit is used to set decision boundaries to sense the collected samples, compare the high-dimensional features of the signal with the decision boundaries, and output the sensing result of the echo signal. The decision boundary is determined based on the hypersphere radius and center hyperparameter of the deep support vector description network obtained through training.

[0039] Specifically, refer to Figure 2 The interference perception module includes a feature extraction unit and an interference decision unit.

[0040] The feature extraction unit is used to extract high-dimensional features from the dataset output by the data preprocessing module. This unit includes an autoencoder and a deep support vector description network, and the model parameters of the autoencoder and deep support vector description network are determined through training. The interference decision unit is used to set decision boundaries for sensing the acquired samples. The setting of these decision boundaries involves the hypersphere radius and center hyperparameters obtained during training. The signal sensing result is obtained by comparing the high-dimensional features of the signal output from the feature extraction unit with the decision boundaries.

[0041] In this embodiment, the core function of the interference perception module is to accurately perceive interference signals through high-dimensional feature extraction and intelligent decision-making. The feature extraction unit employs an architecture combining an autoencoder and a Deep Support Vector Description Network (DeepSVDD) to perform a nonlinear transformation on the input dataset, extracting high-dimensional features that can distinguish between the desired signal and the interference signal. The autoencoder compresses and reconstructs the data through an encoder-decoder structure, while DeepSVDD constructs a hypersphere model, mapping the normal signal (desired signal) to the vicinity of the sphere's center, while the interference signal is located outside the sphere, thus providing a basis for subsequent decision-making.

[0042] The decision unit sets the decision boundary based on the hypersphere radius and center hyperparameters obtained from Deep SVDD training, and compares the extracted high-dimensional features of the signal with the boundary. If the feature lies within the hypersphere, it is determined to be the desired signal; otherwise, it is considered an interference signal. This design not only improves robustness to complex interference, but also adapts to signal changes in different scenarios by dynamically adjusting the decision boundary, ensuring the accuracy and real-time performance of the sensing results.

[0043] Optionally, the autoencoder includes an encoder and a decoder; The encoder includes a residual unit layer and a fully connected layer, which are used to extract features and reduce the dimensionality of the dataset corresponding to the desired signal to obtain the encoded feature vector. The residual unit layer includes three convolutional layers, a filler layer, and a fusion layer; The decoder has a structure symmetrical to the encoder and is used to recover the encoded feature vector into the desired signal.

[0044] Specifically, refer to Figure 3 and Figure 4 , Figure 3 This is a diagram of the self-encoder architecture provided in the embodiments of this application. Figure 4 This is a diagram of the deep support vector data description network architecture provided in the embodiments of this application; The autoencoder in this embodiment serves as a key component for feature extraction. Its encoder portion consists of two residual unit layers and a fully connected layer, responsible for deep feature extraction and dimensionality reduction of the desired signal dataset. The residual unit layer comprises three convolutional layers, one padding layer, and one fusion layer. Convolutional operations capture local features of the signal, the padding layer maintains the feature map size, and the fusion layer integrates multi-level features to enhance expressive power. The fully connected layer further maps the features into low-dimensional encoded vectors, providing a compact representation for subsequent processing.

[0045] The decoder structure is symmetrical to the encoder, aiming to restore the encoded feature vector to the original desired signal. Through layer-by-layer upsampling and convolution operations in the decoder, the encoded vector is gradually reconstructed, ultimately outputting a result that highly matches the input signal. This symmetrical design ensures that the autoencoder retains key information during dimensionality reduction, while mitigating the vanishing gradient problem through residual connections, thus improving the stability of model training and the accuracy of feature extraction.

[0046] Optionally, the interference sensing module includes a data caching unit, a data integration unit, a model training unit, and a parameter storage unit; The data buffer unit is used to temporarily store the time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal of the main channel; The data integration unit is used to call the time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal stored in the data buffer unit; The model training unit is used to input different time-domain IQ sequences and frequency-domain IQ sequences into the autoencoder and the deep support vector data description model for training, so as to obtain the optimal model parameters. The parameter storage unit is used to store the model parameters of the trained autoencoder and the model parameters of the deep support vector data description model.

[0047] Optionally, the interference sensing module further includes a parameter receiving unit and an online sensing unit; The parameter receiving unit is used to receive deep learning network model parameters from the parameter storage unit; The online sensing unit is used to sense the radar time-domain IQ signal and radar frequency-domain IQ signal received in real time based on the parameters of the deep learning network model, so as to determine the signal components and the receiving timing.

[0048] Specifically, in this embodiment, the data caching unit temporarily stores the time-domain and frequency-domain IQ sequences of the main channel echo signal to ensure efficient data retrieval; the data integration unit coordinates the reading and integration of cached data to provide input for model training. The model training unit iteratively optimizes model parameters by inputting diverse time-domain and frequency-domain IQ sequences into the autoencoder and Deep SVDD to adapt to different signal environments.

[0049] The parameter storage unit is responsible for saving the trained model parameters, including the autoencoder weights and the Deep SVDD hypersphere parameters. These parameters can be directly accessed in subsequent online sensing, avoiding redundant training and improving real-time performance. Furthermore, the module supports dynamic parameter updates, fine-tuning the model by periodically receiving new data to ensure continuous optimization of sensing performance as the environment changes.

[0050] The interference sensing module further extends the parameter receiving unit and the online sensing unit, enabling a seamless transition from offline training to online sensing. The parameter receiving unit is responsible for loading pre-trained deep learning network model parameters from the parameter storage unit, including encoder / decoder weights of the autoencoder and hypersphere parameters of Deep SVDD, providing a foundational model for online sensing.

[0051] The online sensing unit rapidly processes real-time received radar time-domain and frequency-domain IQ signals based on loaded model parameters. By extracting signal features through an autoencoder and combining this with the decision boundaries of Deep SVDD, the unit can output the signal sensing results in real time, including signal component classification and timing information. This design ensures real-time interference sensing while reducing computational complexity through a pre-trained model, making it suitable for high-speed radar signal processing scenarios.

[0052] Reference Figure 5 This application employs a feature extraction process that combines autoencoder training with deep support vector data description training. First, preprocessed radar received signal samples are used... Train the autoencoder to initialize the encoder's network weights. And solve for the center of the hypersphere. Subsequently, the network weights of the encoder are further optimized using a loss function described by support vector data. And solve for the radius of the hypersphere. Finally, through the trained encoder and the ball center... and hypersphere radius This allows us to obtain low-dimensional features of the received signal.

[0053] Reference Figure 6This application employs the least mean square error (LMS) adaptive algorithm as the algorithm for generating radar main lobe deception interference cancellation weights.

[0054] The mathematical model of the spatial LMS algorithm is:

[0055] in, It is a weighted output signal. These are the weighting coefficients for the receiving channel. This indicates the complex conjugate transpose. The antenna receives signal vectors, which are the main channel signal, azimuth difference channel signal, and elevation difference channel signal, respectively.

[0056]

[0057] To estimate the error, The desired signal is related to the useful signal. The updated weights are:

[0058] in, To estimate the conjugate of the error, This is the updated value for the real part. Assuming the deception interference signal is a stationary random process, the LMS algorithm selects a smaller step size parameter. This ensures the system operates within the convergence region. This application considers a mean-preserving method, where all weights from the previous receive slot are averaged and used as the weights for the next hold slot. Due to the adaptive weights generated by the LMS algorithm... The weights converge to the steady-state Wiener solution in an average sense. Therefore, when using the mean-preserving method, the output signal power of non-cooperative interference cancellation is... ,in This represents the output power of the Wiener solution weights.

[0059] Unlike the time-domain LMS algorithm, which takes the same signal as its input at different times, the spatial-domain LMS algorithm takes signals from different directions at the same time. For complex signals, the weight update algorithm is as follows:

[0060]

[0061] in, Update the imaginary part value. Let be the real part of the input signal. Let be the imaginary part of the input signal. To estimate the real part of the error, This represents the imaginary part of the estimation error.

[0062] Adding decision coefficients to the warrant update algorithm This allows the interference perception results to control the updating of weights:

[0063] When the interference sensing output indicates that there is interference. Values That is, to perform the weight update operation; when the interference sensing output result is no interference, Values That is, to maintain the weight of the previous receiving time slot.

[0064] Optionally, the weight decision module includes a power detection unit and a weight calculation unit; The power detection unit is used to detect the power of the signals received by the main channel and the sampling channel, respectively. The weight calculation unit is used to calculate the cancellation weight of the sampling channel based on the power of the received signal and the timing information of the sensing signal output by the sensing module.

[0065] Specifically, the weight decision module in this embodiment consists of a power detection unit and a weight calculation unit, which effectively cancels interference through signal power sensing and weight optimization. The power detection unit measures the power of the received signals in the main channel and the sampling channel respectively, providing basic data for weight calculation. By comparing the power difference between the two channels, the module can initially determine the interference intensity, providing a basis for subsequent weight adjustment.

[0066] The weight calculation unit dynamically calculates the cancellation weights of the sampling channel by combining the power detection results and the signal timing information output by the sensing module. The timing information helps the module accurately locate the start and end times of the interference signal, ensuring that the weight adjustment only takes effect when interference is present. This weight calculation method based on power sensing and timing coordination not only improves cancellation efficiency but also avoids invalid calculations when there is no interference, thus optimizing the overall system performance.

[0067] Optionally, the data preprocessing module includes a modulation and mixing unit, a digital filtering unit, a downsampling unit, a Fourier transform unit, and a matrix construction unit; The modulation and mixing unit is used to modulate and mix the echo signal to generate a time-domain IQ sequence; The digital filtering unit is used to filter the time-domain IQ sequence to suppress out-of-band noise; The downsampling unit is used to downsample the time-domain IQ sequence to reduce the data rate; The Fourier transform unit is used to perform Fourier transform on the time-domain IQ sequence to obtain the frequency-domain IQ sequence; The matrix construction unit is used to construct a time-frequency IQ matrix based on the time-domain IQ sequence and the frequency-domain IQ sequence.

[0068] Specifically, the time-frequency preprocessing unit uses a short-time Fourier transform algorithm to convert the time-domain signal into a frequency-domain signal. The specific process is described as follows: Assume the collected data is a discrete time-domain sequence:

[0069] in, The number of sampling points, the first The IQ data obtained by orthogonal synchronous sampling at discrete time points can be represented as:

[0070] Then, the corresponding frequency domain sequence can be obtained by performing a Fourier transform on the time domain sequence:

[0071] in, For the first The IQ data of a signal with discrete frequency points can be expressed as:

[0072] Finally, the complex sequences in the time and frequency domains are decomposed into real sequences of I and Q, resulting in a... A 3D matrix, where rows one through four correspond to the real and imaginary parts of the time-domain signal and the real and imaginary parts of the frequency-domain signal, respectively:

[0073] In the formula, The matrix represents the dataset of the input interference sensing network.

[0074] Reference Figure 7 , Figure 7 This is a complete structural diagram of an embodiment of this application, including the following modules: The offline training module specifically includes: Model training unit: responsible for training the interference-aware model, including autoencoders and deep support vector description networks.

[0075] Parameter storage unit: Stores the trained model parameters for use by the online perception module.

[0076] Data integration unit: Integrates and preprocesses training data to provide a high-quality dataset for model training.

[0077] Data acquisition module: It is responsible for collecting radar time-domain IQ signals and frequency-domain IQ signals to provide raw data for subsequent data processing.

[0078] Data processing module: The collected data is preprocessed, including filtering and normalization, to improve data quality.

[0079] The interference sensing module specifically includes: Online sensing unit: Utilizes model parameters obtained from offline training to perceive received radar signals in real time and identify interference signals.

[0080] Time-frequency processing unit: performs time-frequency analysis on the signal and extracts the time-frequency characteristics of the signal.

[0081] Model parameter receiving unit: Receives and updates model parameters to ensure the accuracy of online sensing.

[0082] Weighted decision module: Power detection unit: detects the power of the signals received by the main channel and the sampling channel, providing a basis for weight calculation.

[0083] Weight calculation unit: Calculates the cancellation weights of the sampling channel based on the power detection results and the signal timing information output by the sensing module.

[0084] Interference cancellation module: Based on the calculation results of the weight decision module, interference signals are canceled to improve signal quality.

[0085] Result output module: The final signal after interference cancellation is output for subsequent analysis and application.

[0086] The entire system in this application embodiment establishes an interference perception model through offline training, and uses the model to process real-time signals in the online stage to realize the perception and cancellation of interference signals, thereby improving the anti-interference capability of the radar system.

[0087] Reference Figure 8 This application also provides a radar main lobe interference cancellation method based on situational awareness, including: S101. Acquire the echo signal from the receiving link; S102. Perform data preprocessing on the echo signal to construct a dataset for interference sensing; S103. Perform time-series sensing on the echo signal based on the trained online sensing model to obtain the time-series information of the echo signal; the online sensing model is trained based on the dataset; S104. Select the weight state according to the timing information of the echo signal; S105. Use the sampling channel echo signal to cancel the main lobe interference of the main channel echo signal, and make a judgment based on the weight state to retain the useful signal in the echo signal.

[0088] Optionally, the weight states include 0 and 1; S104 specifically includes: S41. Based on the timing information, determine that the timing is in an interference sequence, set the weight state to 1, and perform a weight update operation; S42. Based on the timing information, determine that the timing is in a non-interference sequence, set the weight state to 0, and maintain the cancellation weight corresponding to the timing information of the previous received signal.

[0089] Optionally, S103 specifically includes: S31. High-dimensional feature extraction is performed on the dataset based on an autoencoder and a deep support vector description network to obtain high-dimensional signal features; S32. Set a decision boundary to sense the collected samples, compare the high-dimensional features of the signal with the decision boundary, and output the sensing result of the echo signal; The decision boundary is determined based on the hypersphere radius and center hyperparameter of the deep support vector description network obtained through training.

[0090] Optionally, the autoencoder includes an encoder and a decoder; The encoder includes a residual unit layer and a fully connected layer, which are used to extract features and reduce the dimensionality of the dataset corresponding to the desired signal to obtain the encoded feature vector. The residual unit layer includes three convolutional layers, a filler layer, and a fusion layer; The decoder has a structure symmetrical to the encoder and is used to recover the encoded feature vector into the desired signal.

[0091] Optionally, the method further includes: S33. The time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal of the main channel are temporarily stored in the data buffer unit; S34. The data integration unit calls the time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal stored in the data buffer unit; S35. Different time-domain IQ sequences and frequency-domain IQ sequences are input into the autoencoder and deep support vector data description model through the model training unit to train them and obtain the optimal model parameters; S36. The trained autoencoder model parameters and the deep support vector data description model model parameters are stored in the parameter storage unit.

[0092] S37. Receive deep learning network model parameters from the parameter storage unit through the parameter receiving unit; S38. Based on the parameters of the deep learning network model, the online sensing unit senses the radar time-domain IQ signal and the radar frequency-domain IQ signal received in real time to determine the signal components and the receiving timing.

[0093] Optionally, S104 specifically includes; S43. Detect the power of the received signals in the main channel and the sampling channel respectively; S44. Based on the power of the received signal and the timing information of the sensing signal output by the sensing module, calculate the cancellation weight of the sampling channel.

[0094] Optionally, the data preprocessing process includes; S21. Modulate and mix the echo signal to generate a time-domain IQ sequence; S22. Filter the time-domain IQ sequence to suppress out-of-band noise; S23. Downsample the time-domain IQ sequence to reduce the data rate; S24. Perform a Fourier transform on the time-domain IQ sequence to obtain the frequency-domain IQ sequence; S25. Construct a time-frequency IQ matrix based on the time-domain IQ sequence and the frequency-domain IQ sequence.

[0095] Optionally, S101 specifically includes: Based on the set sampling rate and intermediate frequency parameters, the intermediate frequency echo signal from the receiving link is acquired by analog-to-digital conversion to obtain the main channel echo signal and the sampling channel echo signal.

[0096] Reference Figure 9 Based on the methods in the above embodiments, this application provides an electronic device that may include: a processor 910, a communications interface 920, a memory 930, and a communication bus 940. The processor 910, communications interface 920, and memory 930 communicate with each other via the communication bus 940. The processor 910 can call logical instructions in the memory 930 to execute the methods in the above embodiments.

[0097] Furthermore, the logical instructions in the aforementioned memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0098] Based on the methods in the above embodiments, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0099] Based on the methods in the above embodiments, this application provides a computer program product that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0100] It is understood that the processor in the embodiments of this application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor can be a microprocessor or any conventional processor.

[0101] The method steps in this application embodiment can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.

[0102] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0103] It is understood that the various numerical designations used in the embodiments of this application are merely for the convenience of description and are not intended to limit the scope of the embodiments of this application.

[0104] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A radar main lobe interference cancellation system based on situational awareness, characterized in that, include: The system includes a data acquisition module, a data preprocessing module, an interference sensing module, a weighting decision module, and an interference cancellation module. The data acquisition module includes a main channel and a sampling channel. The input ends of both the main channel and the sampling channel are connected to the radar receiving link and are used to acquire the echo signal from the receiving link. The input end of the data preprocessing module is connected to the main channel and is used to preprocess the echo signal to construct a dataset for interference sensing. The input end of the interference sensing module is connected to the data preprocessing module, and is used to perform time-series sensing on the echo signal based on the trained online sensing model to obtain the time-series information of the echo signal. The online perception model is trained based on the dataset; The input terminal of the weight decision module is connected to the interference sensing module and is used to select the weight state according to the timing information of the echo signal. The first input terminal of the interference cancellation module is connected to the data acquisition module, and the second input terminal of the interference cancellation module is connected to the weight decision module. The interference cancellation module is used to cancel the main lobe interference of the main channel echo signal using the sampling channel echo signal, and to make a judgment based on the weight state to retain the useful signal in the echo signal.

2. The radar main lobe interference cancellation system based on situational awareness according to claim 1, characterized in that, The weight states include 0 and 1; The weight decision module is specifically used for: Based on the timing information, it is determined that the echo signal is in an interference timing sequence. The weight state is set to 1, and a weight update operation is performed. Based on the timing information, the echo signal is determined to be in a non-interference timing sequence. The weight state is set to 0, and the cancellation weight corresponding to the timing information of the previous received signal is maintained.

3. The radar main lobe interference cancellation system based on situational awareness according to claim 1, characterized in that, The interference perception module includes a feature extraction unit and an interference decision unit; the online perception model includes an autoencoder and a deep support vector description network. The feature extraction unit is used to perform high-dimensional feature extraction on the dataset based on an autoencoder and a deep support vector description network to obtain high-dimensional features of the signal. The interference decision unit is used to set decision boundaries to sense the collected samples, compare the high-dimensional features of the signal with the decision boundaries, and output the sensing result of the echo signal. The decision boundary is determined based on the hypersphere radius and center hyperparameter of the deep support vector description network obtained through training.

4. The radar main lobe interference cancellation system based on situational awareness according to claim 3, characterized in that, The autoencoder includes an encoder and a decoder; The encoder includes a residual unit layer and a fully connected layer, which are used to extract features and reduce the dimensionality of the dataset corresponding to the desired signal to obtain the encoded feature vector. The decoder has a structure symmetrical to the encoder and is used to recover the encoded feature vector into the desired signal.

5. The radar main lobe interference cancellation system based on situational awareness according to claim 4, characterized in that, The interference sensing module includes a data caching unit, a data integration unit, a model training unit, and a parameter storage unit. The data buffer unit is used to temporarily store the time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal of the main channel; The data integration unit is used to call the time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal stored in the data buffer unit; The model training unit is used to input different time-domain IQ sequences and frequency-domain IQ sequences into the autoencoder and the deep support vector data description model for training, so as to obtain the optimal model parameters. The parameter storage unit is used to store the model parameters of the trained autoencoder and the model parameters of the deep support vector data description model.

6. The radar main lobe interference cancellation system based on situational awareness according to claim 5, characterized in that, The interference sensing module further includes a parameter receiving unit and an online sensing unit; The parameter receiving unit is used to receive deep learning network model parameters from the parameter storage unit; The online sensing unit is used to sense the radar time-domain IQ signal and radar frequency-domain IQ signal received in real time based on the parameters of the deep learning network model, so as to determine the signal components and the receiving timing.

7. The radar main lobe interference cancellation system based on situational awareness according to claim 1, characterized in that, The weight decision module includes a power detection unit and a weight calculation unit; The power detection unit is used to detect the power of the signals received by the main channel and the sampling channel, respectively. The weight calculation unit is used to calculate the cancellation weight of the sampling channel based on the power of the received signal and the timing information of the sensing signal output by the sensing module.

8. The radar main lobe interference cancellation system based on situational awareness according to claim 1, characterized in that, The data preprocessing module includes a modulation and mixing unit, a digital filtering unit, a downsampling unit, a Fourier transform unit, and a matrix construction unit; The modulation and mixing unit is used to modulate and mix the echo signal to generate a time-domain IQ sequence; The digital filtering unit is used to filter the time-domain IQ sequence to suppress out-of-band noise; The downsampling unit is used to downsample the time-domain IQ sequence to reduce the data rate; The Fourier transform unit is used to perform Fourier transform on the time-domain IQ sequence to obtain the frequency-domain IQ sequence; The matrix construction unit is used to construct a time-frequency IQ matrix based on the time-domain IQ sequence and the frequency-domain IQ sequence.

9. The radar main lobe interference cancellation system based on situational awareness according to claim 1, characterized in that, The data acquisition module is specifically used to perform analog-to-digital acquisition on the intermediate frequency echo signal from the receiving link according to the set sampling rate and intermediate frequency parameters, so as to obtain the main channel echo signal and the sampling channel echo signal.

10. A radar main lobe interference cancellation method based on situational awareness, characterized in that, The system is applied to the situational awareness-based radar main lobe interference cancellation system as described in any one of claims 1-9, comprising: Acquire echo signals from the receiving link; The echo signal is preprocessed to construct a dataset for interference sensing; The echo signal is subjected to time-series sensing based on a trained online sensing model to obtain the time-series information of the echo signal; the online sensing model is trained based on the dataset. The weight state is selected based on the timing information of the echo signal; The main lobe interference of the main channel echo signal is canceled by using the echo signal from the sampling channel, and the useful signal in the echo signal is retained by combining the weight state.

11. The radar main lobe interference cancellation method based on situational awareness according to claim 10, characterized in that, The time-sensing process includes: High-dimensional features of the signal are extracted from the dataset based on an autoencoder and a deep support vector description network. The decision boundary is set to sense the collected samples, the high-dimensional features of the signal are compared with the decision boundary, and the sensing result of the echo signal is output. The decision boundary is determined based on the hypersphere radius and center hyperparameter of the deep support vector description network obtained through training.

12. The radar main lobe interference cancellation method based on situational awareness according to claim 11, characterized in that, The method further includes: The time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal of the main channel are temporarily stored in the data buffer unit. The data integration unit calls the time-domain IQ sequence and frequency-domain IQ sequence corresponding to the echo signal stored in the data cache unit; Different time-domain IQ sequences and frequency-domain IQ sequences are input into the autoencoder and deep support vector data description model through the model training unit to obtain the optimal model parameters; The parameter storage unit stores the model parameters of the trained autoencoder and the model parameters of the deep support vector data describing the model.

Images