Communication method and system based on broadband interference suppression and target signal reconstruction

Abstract

The invention belongs to the technical field of communication, and discloses a communication method and system based on broadband interference suppression and target signal reconstruction, and the method comprises the steps: collecting an original broadband communication signal; carrying out compressed sampling on the preprocessed signal to obtain compressed data; performing time-frequency domain conversion on the compressed data, and performing sparse representation on the signal by using the sparse dictionary to obtain time-frequency domain conversion data; reconstructing an original broadband communication signal from the time-frequency domain conversion data by adopting an iterative reconstruction algorithm and combining an interference suppression strategy, and eliminating an interference component to obtain a target signal; and comparing the quality index of the reconstructed target signal with a preset performance index, and dynamically optimizing at least one key parameter of compressed sampling, sparse representation, iterative reconstruction or interference suppression according to a comparison result. According to the method, a complex and changeable interference environment can be adaptively dealt with, it is ensured that the system always keeps the optimal processing effect, and the robustness is far better than that of a traditional method adopting fixed parameters.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a communication method and system based on broadband interference suppression and target signal reconstruction. Background Art

[0002] With the widespread adoption of broadband communication technologies and radio frequency equipment, modern wireless signal environments are becoming increasingly complex. Traditional signal processing methods, which typically rely on fixed sampling and reconstruction techniques, struggle to achieve high-precision recovery of target signals under complex conditions such as multipath fading, low signal-to-noise ratios, and the frequent superposition of multiple interference sources.

[0003] Existing technologies for processing broadband signals generally use fixed sampling rates and standard Fourier transforms. These methods fail to fully exploit the inherent sparseness of signals in the time-frequency domain. This significantly impacts signal reconstruction and system robustness in the face of dynamically changing interference and complex background noise.

[0004] In recent years, compressed sensing theory has demonstrated tremendous potential in signal processing, enabling the reduction of sampling rates while maintaining signal integrity. However, designing an appropriate measurement matrix for realistic broadband signal environments, effectively separating interference components, and accurately reconstructing the target signal remain technical challenges. Many current compressed sensing-based methods lack flexibility in parameter selection, interference suppression strategies, and real-time feedback control. These methods are unable to adaptively address dynamic changes in signal characteristics and the external environment, resulting in signal reconstruction accuracy and overall system performance that are insufficient for practical applications.

[0005] Therefore, the industry urgently needs a new signal processing method that can effectively suppress interference in complex wireless environments and achieve high-quality reconstruction of target signals to improve the overall performance of applications such as radar detection, spectrum monitoring, and wireless communications. Summary of the Invention

[0006] In order to solve the problems existing in the above-mentioned prior art, the technical solutions provided by the present invention include: A communication method based on broadband interference suppression and target signal reconstruction includes the following steps: S1. Acquire and preprocess the original broadband communication signal containing the target signal; S2. Constructing a measurement matrix based on the principle of compressed sensing, compressing and sampling the preprocessed signal to obtain compressed data; S3. Converting the compressed data into the time-frequency domain and performing a sparse characterization of the signal using a sparse dictionary to distinguish the interference component from the target signal and obtain the time-frequency domain conversion data; S4. Using an iterative reconstruction algorithm, combined with an interference suppression strategy, reconstructing the original broadband communication signal from the time-frequency domain conversion data, while removing the interference component, to obtain the target signal; S5. Establish a real-time feedback mechanism to compare the quality index of the reconstructed target signal with the preset performance index, and dynamically optimize at least one key parameter of the compressed sampling, sparse representation, iterative reconstruction or interference suppression described in S2, S3 or S4 based on the comparison results.

[0007] Preferably, the pretreatment includes: S11. Filter the collected signal using a wavelet threshold denoising algorithm to suppress random noise; S12. performing normalization processing on the denoised signal to have zero mean and unit variance to improve the numerical stability of signal processing; S13. Divide the normalized signal into multiple data segments of fixed length to construct a subsample set with a unified format.

[0008] Preferably, step S2 includes: A measurement matrix satisfying the constrained isometry property is constructed, and each preprocessed signal data segment is linearly projected using the measurement matrix to obtain compressed data with a dimension lower than the original dimension of the data segment.

[0009] Preferably, the time-frequency conversion includes: S301. Using a preset window function, the compressed data is segmented and windowed with a set overlap rate and sliding mode; S302. Perform a fast Fourier transform on each windowed signal segment to generate a two-dimensional time-frequency matrix representing the time-frequency distribution of the signal; S303. Flatten the two-dimensional time-frequency matrix into a one-dimensional feature vector.

[0010] Preferably, the performing sparse characterization on the signal using a sparse dictionary includes: S311. Using a sparse dictionary to sparsely represent the one-dimensional feature vector after time-frequency domain conversion; S312. Use a greedy algorithm to solve the sparse coefficient vector of the one-dimensional feature vector under the sparse dictionary.

[0011] Preferably, the sparse dictionary includes: a standard transformation basis for characterizing universal components of a signal; and an overcomplete atom set generated by pre-training with typical target signal samples and used to characterize specific structural features.

[0012] Preferably, the iterative reconstruction algorithm includes: S41. The compressed data is set as the initialization residual and an empty support set is initialized; S42. In each iteration, identifying the atom most relevant to the current residual from the sparse dictionary and adding the atom to the support set; S43. Based on the updated support set, solve the sparse coefficient and update the residual; S44. Repeat steps S42 and S43 until a preset iteration stop condition is met, thereby obtaining a reconstructed target signal.

[0013] Preferably, the interference suppression strategy includes: setting a dynamic energy threshold, and suppressing or removing sparse atoms corresponding to frequency components whose amplitudes on the spectrum exceed the threshold in subsequent iterations.

[0014] Preferably, the key parameters include compression ratio, upper limit of sparsity of iterative reconstruction and energy threshold for interference suppression.

[0015] The present invention also provides a communication system based on broadband interference suppression and target signal reconstruction, wherein the system is used to execute the above-mentioned communication method based on broadband interference suppression and target signal reconstruction.

[0016] Beneficial effects 1. This invention incorporates a real-time feedback mechanism and closed-loop control system, enabling real-time monitoring of reconstructed signal quality indicators (such as signal-to-noise ratio and reconstruction error). When performance degrades, the system automatically and dynamically adjusts key parameters such as compression rate, sparsity, and interference rejection threshold. This allows it to adaptively cope with complex and changing interference environments, ensuring optimal processing performance at all times. Its robustness far exceeds that of traditional methods that use fixed parameters.

[0017] 2. During the iterative reconstruction process, this invention incorporates a specialized frequency-domain interference suppression strategy. By setting a dynamic energy threshold, it can accurately identify and remove strong interference components with sudden changes in the spectrum. Furthermore, through time-frequency conversion and a composite sparse dictionary, it can more clearly reveal and distinguish the structures of the target signal and interference. This effectively suppresses interference while restoring the spectral structure of the target signal with high fidelity, significantly improving reconstruction accuracy.

[0018] 3. This invention utilizes a technical approach of "lightweight sampling + sparse modeling + dynamic tuning," acquiring signals at sampling rates far below the Nyquist rate, significantly reducing the burden of data processing and storage. The entire system maintains high reconstruction accuracy while balancing computational efficiency, making it suitable for deployment scenarios with limited hardware resources or data bandwidth, such as front-end SDR platforms or edge computing modules.

[0019] 4. This invention provides a complete, iteratively optimized signal processing framework. After achieving performance convergence, the system can solidify and deploy the optimized model parameters for efficient, real-time processing of newly acquired signals. This method can be widely applied to engineering tasks such as electronic countermeasures, tactical communications, radar front-end signal processing, and spectrum monitoring, providing a solid technical foundation for building intelligent, highly reliable communication systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 The figure is a flow chart of a communication method based on broadband interference suppression and target signal reconstruction provided in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is further described below with reference to the accompanying drawings. In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inner", "outer", etc., indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the devices or components referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0022] The technical field involved in the present invention is communication signal processing, especially in the increasingly complex modern wireless signal environment. How to effectively suppress broadband interference and accurately recover the target signal that is overwhelmed by the interference is a technical problem that needs to be solved urgently. Traditional signal processing methods often rely on the Nyquist sampling theorem, with a huge amount of data processing, and when faced with dynamic, non-Gaussian strong interference, the performance drops sharply, making it difficult to meet the stringent requirements of signal processing real-time and fidelity in scenarios such as wireless communications, radar detection, and spectrum monitoring. In order to overcome the above-mentioned technical defects, the present invention proposes an innovative method and system for broadband interference suppression and target signal reconstruction based on compressed sensing theory and integrating the idea of ​​adaptive closed-loop control.

[0023] The core innovation of this invention lies in the following: First, rather than adopting the traditional full-scale sampling method, it exploits the sparse characteristics of the signal in a specific transform domain (such as the time-frequency domain) and obtains low-dimensional observation data of the signal through compressed sampling at a rate far lower than the Nyquist rate, thereby greatly reducing the burden of data acquisition and transmission. Second, rather than using fixed parameters and algorithms for signal reconstruction, it establishes a dynamic closed-loop feedback mechanism. This mechanism can evaluate the quality of the reconstructed signal in real time and, based on the evaluation results, adaptively adjust key parameters of core links such as compressed sampling, sparse representation, iterative reconstruction, and interference suppression, thereby enabling the entire system to intelligently adapt to changing interference environments and always maintain an optimal or near-optimal working state.

[0024] Example 1 like Figure 1 As shown, this embodiment discloses a communication method based on broadband interference suppression and target signal reconstruction, including the steps of: S1. Collect and preprocess the original broadband communication signal containing the target signal.

[0025] The purpose of this step is to obtain digital samples of the original signal and perform a series of standardization processes on it to lay a stable and reliable data foundation for subsequent compressed sensing and accurate reconstruction.

[0026] In a preferred implementation, the pre-processing process may specifically include the following sub-steps: S11: To suppress the random noise (especially high-frequency noise) that is inevitably introduced during the signal acquisition process, a wavelet threshold denoising algorithm can be used to perform preliminary filtering on the collected original signal sequence. For example, a suitable wavelet basis (such as the Daubechies-4 wavelet) can be selected to perform multi-layer wavelet decomposition on the signal. Then, by setting a suitable threshold (such as a universal threshold calculated based on the noise standard deviation), soft thresholding or hard thresholding is performed on the wavelet coefficients of each layer. Finally, a smoother signal is reconstructed through an inverse wavelet transform.

[0027] S12: To eliminate signal amplitude inconsistencies caused by factors such as different acquisition devices and channel gain variations, thereby improving the numerical stability of subsequent signal processing algorithms, the denoised signal is normalized. An effective approach is to use zero-mean unit variance (Z-Score) normalization. This involves calculating the mean and standard deviation of the signal sequence, subtracting the mean from each sampling point, and then dividing by the standard deviation. This process adjusts the signal amplitude to a uniform, controllable range (e.g., [-1, 1]), which facilitates the convergence and accuracy of subsequent algorithms.

[0028] S13: To facilitate segmentation and construct a uniformly formatted dataset, the long signal sequence processed above is divided into multiple fixed-length data segments to construct a subsample set. For example, a signal sequence of length 16384 can be divided into 64 data segments of length 256. This not only facilitates parallel processing but also allows each data segment to be treated as an independent signal sample with local time-frequency characteristics, making it suitable for subsequent compressed sensing modeling.

[0029] S2. Construct a measurement matrix based on the principle of compressed sensing, perform compressive sampling on the preprocessed signal, and obtain compressed data.

[0030] After signal preprocessing, the method enters the compressed sampling phase. This step, based on the theory of compressed sensing (CS), constructs a suitable measurement matrix and uses it to perform dimensionality reduction projection on the preprocessed signal. This allows compressed data to be obtained that retains the main information of the original signal at a sampling rate far lower than the Nyköquist sampling rate.

[0031] Specifically, the core of this step is to construct a measurement matrix Φ that satisfies the Restricted Isometry Property (RIP). The RIP property is key to accurately reconstructing sparse signals from compressed data. In practical applications, various matrix types that meet the RIP condition can be selected, such as Gaussian random matrices, Bernoulli random matrices, or partial Fourier matrices. For example, a Gaussian random matrix of dimension M×N can be constructed, where N is the original dimension of each signal data segment after preprocessing (e.g., 256) and M is the compressed dimension (e.g., 64). The compression ratio is then ρ = M / N (e.g., 0.25).

[0032] Then, using this constructed measurement matrix Φ, a linear projection is performed on each preprocessed signal data segment x∈RN, resulting in a compressed data vector y∈RM with a dimension much smaller than the original. This is mathematically represented as y=Φx. This process significantly compresses the data volume, significantly improving the efficiency of subsequent storage, transmission, and computation.

[0033] S3. Perform time-frequency domain conversion on the compressed data, and perform sparse representation on the signal using a sparse dictionary to distinguish interference components from target signals, and obtain time-frequency domain conversion data.

[0034] After obtaining the compressed data, in order to more clearly reveal the different characteristics of the target component and the interference component in the signal and use these characteristics to effectively separate them, the present invention performs time-frequency domain conversion and sparse representation on the compressed data.

[0035] In a specific embodiment, the time-frequency domain conversion process can be implemented as follows: S301: Using a preset window function (such as a Hamming window), with a set overlap rate (such as 50%) and a sliding mode, perform segmented windowing processing on the compressed data vector y.

[0036] S302: Perform a Fast Fourier Transform (FFT) on each windowed signal segment, converting the one-dimensional time-domain signal into a two-dimensional time-frequency matrix that shows how the signal frequency changes over time. This matrix clearly reveals the energy distribution of the signal at different time-frequency units.

[0037] S303: To facilitate subsequent processing, the two-dimensional time-frequency matrix may be flattened in a specific order (eg, column by column or row by row) to form a one-dimensional feature vector containing complete time-frequency information.

[0038] Next is the process of using a sparse dictionary to sparsely represent the signal. The core idea is that any signal can be approximately represented by a linear combination of a few atoms (basic waveforms) in a dictionary. In some preferred embodiments, the method includes: S311: A carefully designed sparse dictionary is used to sparsely represent the one-dimensional feature vector after the time-frequency domain conversion. To effectively represent different types of signal components simultaneously, the sparse dictionary can be designed as a composite dictionary, consisting of a standard transform basis (such as the discrete cosine transform (DCT) or Fourier basis) for representing universal, stationary signal components, as well as an overcomplete atom set. This overcomplete atom set can be pre-trained using a dictionary learning algorithm (such as K-SVD) from a large number of typical, interference-free target signal samples. This allows it to strongly characterize the specific structural features of the target signal.

[0039] S312: After selecting the dictionary, an efficient sparse coding algorithm is used to solve for the sparse coefficient vector θ of the feature vector under the dictionary. Greedy algorithms, such as Orthogonal Matching Pursuit (OMP), are a preferred approach for this task. The OMP algorithm iteratively selects dictionary atoms that are most relevant to the current signal residual, gradually constructing a sparse approximation of the signal. The resulting sparse coefficient vector θ contains only a few non-zero entries. The positions and values ​​of these non-zero entries accurately depict the main components of the original signal, providing a key basis for distinguishing the target signal from interference.

[0040] S4. Using an iterative reconstruction algorithm and combining it with an interference suppression strategy, the original broadband communication signal is reconstructed from the time-frequency domain conversion data, while eliminating the interference component to obtain the target signal.

[0041] The goal of this step is to accurately reconstruct the original broadband communication signal from the compressed time-frequency domain converted data and actively and strategically remove interference components during the reconstruction process. The present invention uses an iterative reconstruction algorithm and cleverly embeds an interference suppression strategy to achieve this goal.

[0042] In some preferred embodiments, the specific process of implementing the iterative reconstruction algorithm is as follows: S41: Set the compressed data y (or its time-frequency eigenvector) as the initialization residual r0, and initialize an empty support set (for storing the index of the selected atoms).

[0043] S42: In each iteration, the inner product of the current residual and all atoms in the sparse dictionary is calculated, the atom most correlated with the residual (with the largest absolute value of the inner product) is identified, and its index is added to the support set.

[0044] S43: Based on the updated support set, the sparse coefficients are solved by the least squares method, and the residual is updated using the newly calculated signal approximation.

[0045] S44: This process is repeated until a preset terminating condition is met, such as reaching a preset upper limit on the sparsity (i.e., the number of non-zero coefficients) or the residual energy falls below a sufficiently small threshold. After the iteration is complete, the target signal is reconstructed using the sparse coefficients and the dictionary.

[0046] To achieve interference suppression during reconstruction, the present invention proposes an interference suppression strategy. This strategy is based on the observation that many strong interferences manifest as abnormal, sudden bursts of high energy in the spectrum. Therefore, a dynamic energy threshold can be set in each round or a specific round of iterative reconstruction. This threshold can be dynamically calculated based on the statistical properties of the current reconstructed signal spectrum (such as the mean and standard deviation). For example, it can be set as T = μ + γσ, where μ and σ are the mean and standard deviation of the spectrum amplitude, respectively, and γ is an adjustable coefficient. Frequency components whose spectral amplitude exceeds this threshold are determined to be highly likely interference. Accordingly, in subsequent iterative reconstructions, the sparse atoms corresponding to these high-energy interference frequencies are suppressed, for example, by temporarily removing them from the candidate dictionary or penalizing their weights when solving the sparse coefficients. In this way, the influence of interference components in the reconstruction process is effectively suppressed, resulting in a purer target signal.

[0047] S5. Establish a real-time feedback mechanism to compare the quality index of the reconstructed target signal with the preset performance index, and dynamically optimize at least one key parameter of the compressed sampling, sparse representation, iterative reconstruction or interference suppression described in S2, S3 or S4 based on the comparison results.

[0048] A significant feature of the present invention that distinguishes it from the prior art is that it establishes a real-time feedback and closed-loop control logic, which gives the entire method a strong adaptive capability.

[0049] The core of this feedback mechanism is that after signal reconstruction is completed, the reconstructed target signal is immediately evaluated for quality, calculating key quality indicators such as signal-to-noise ratio (SNR) and mean square error (MSE). These real-time quality indicators are then compared with a set of preset indicators that represent ideal performance.

[0050] If the comparison results indicate that current performance has degraded or fallen short of expectations (for example, if the SNR is below 18dB or the MSE is above 0.005), the feedback mechanism will be triggered. Based on the degree and trend of the deviation, it will intelligently and dynamically optimize and adjust at least one key parameter involved in steps S2, S3, or S4. In some preferred embodiments, these dynamically optimized key parameters may include: Compression ratio: When the reconstruction quality is consistently poor, the compression ratio can be appropriately increased (for example, from 0.25 to 0.3125). This means collecting more measurements and providing richer information for reconstruction.

[0051] Upper limit of sparsity for iterative reconstruction: If the signal complexity is high, the upper limit of sparsity can be increased (for example, from 15 to 20), allowing more dictionary atoms to be used to fit the signal, thereby improving reconstruction accuracy.

[0052] Energy threshold for interference suppression: When faced with interference of varying strengths and forms, the coefficient γ used in calculating the energy threshold can be adjusted (e.g., from 3.0 to 2.5) to more sensitively or robustly identify and suppress interference.

[0053] This closed-loop, performance feedback-based parameter optimization mechanism ensures that the method and system of the present invention can automatically adapt to changes in signal statistical characteristics and external interference environment, and can continuously seek and maintain the optimal operating point without human intervention. This is extremely valuable in the complex and changing scenarios of actual applications.

[0054] Example 2 To further illustrate the present invention, the following describes an embodiment of a communications system for executing the method described in Example 1. The system may include a signal acquisition and preprocessing module, a compressed sampling module, a time-frequency conversion and sparse representation module, an iterative reconstruction and interference suppression module, and a core real-time feedback and control module. These modules may be hardware circuits (such as FPGAs or ASICs), software programs (running on DSPs or general-purpose processors), or a combination of hardware and software.

[0055] In a specific application scenario, the system is used to process an IEEE802.11n OFDM signal with a center frequency of 2.437 GHz.

[0056] Acquisition and preprocessing module: A high-speed oscilloscope is used to acquire the signal at a sampling rate of 1 GHz, obtaining a digital sequence of length 16,384. This module performs the aforementioned Daubechies-4 wavelet denoising, zero-mean unit variance normalization, and divides the sequence into 64 subsamples of length 256.

[0057] Compressed Sampling Module: This module maintains a 64x256 Gaussian random measurement matrix and compresses each 256-point subsample, outputting 64-point compressed data. The initial compression ratio is 0.25.

[0058] Time-frequency and sparse representation module: This module performs STFT (Hamming window, window length 32, overlap rate 50%) on the 64-point compressed data, and then uses a composite dictionary consisting of a DCT basis and pre-trained target signal atoms to calculate the sparse coefficients through the OMP algorithm. The sparsity upper limit is initially set to 15.

[0059] Reconstruction and Interference Suppression Module: This module performs the OMP iterative reconstruction process. Simultaneously, it calculates the energy threshold of the current spectrum (initial γ = 3.0) and suppresses the sparse atoms corresponding to frequency components with energy exceeding the threshold, ultimately outputting the reconstructed 256-point target signal.

[0060] Feedback and Control Module: This module serves as the "brain" of the system. It continuously monitors the SNR and MSE of the output signal. If the SNR continues to drop below 18dB, it instructs the compressed sampling module to increase the measurement matrix dimensions to 80x256 (raising the compression ratio to 0.3125). If the MSE remains above 0.005, it instructs the reconstruction module to increase the upper limit of sparsity to 20. This closed-loop control ensures that the system can quickly adjust itself and restore optimal performance even in the face of sudden changes in the interference pattern.

[0061] By repeatedly executing the complete cycle of compressed sampling, sparse reconstruction, feedback optimization, and interference suppression, system performance gradually converges to the desired target. Once the system reaches a stable convergence state, the optimized set of parameters (including the final measurement matrix, sparse dictionary, sparsity, threshold coefficients, etc.) can be solidified, forming an efficient signal processing model optimized for a specific environment. This model is used to quickly and efficiently process newly acquired signals, and its output can be directly used for subsequent modulation identification or channel decoding.

[0062] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A communication method based on broadband interference suppression and target signal reconstruction, characterized in that: Including steps: S1. Acquire and preprocess the original broadband communication signal containing the target signal; S2. Constructing a measurement matrix based on the principle of compressed sensing, compressing and sampling the preprocessed signal to obtain compressed data; S3. Converting the compressed data into the time-frequency domain and performing a sparse characterization of the signal using a sparse dictionary to distinguish the interference component from the target signal and obtain the time-frequency domain conversion data; S4. Using an iterative reconstruction algorithm, combined with an interference suppression strategy, reconstructing the original broadband communication signal from the time-frequency domain conversion data, while removing the interference component, to obtain the target signal; S5. Establish a real-time feedback mechanism to compare the quality index of the reconstructed target signal with the preset performance index, and dynamically optimize at least one key parameter of the compressed sampling, sparse representation, iterative reconstruction or interference suppression described in S2, S3 or S4 based on the comparison results.

2. The communication method based on broadband interference suppression and target signal reconstruction according to claim 1, wherein: The pretreatment includes: S11. Filter the collected signal using a wavelet threshold denoising algorithm to suppress random noise; S12. performing normalization processing on the denoised signal to have zero mean and unit variance to improve the numerical stability of signal processing; S13. Divide the normalized signal into multiple data segments of fixed length to construct a subsample set with a unified format.

3. The communication method based on broadband interference suppression and target signal reconstruction according to claim 1, wherein: Step S2 includes: A measurement matrix satisfying the constrained isometry property is constructed, and each preprocessed signal data segment is linearly projected using the measurement matrix to obtain compressed data with a dimension lower than the original dimension of the data segment.

4. The communication method based on broadband interference suppression and target signal reconstruction according to claim 1, wherein: The time-frequency conversion includes: S301. Using a preset window function, the compressed data is segmented and windowed with a set overlap rate and sliding mode; S302. Perform a fast Fourier transform on each windowed signal segment to generate a two-dimensional time-frequency matrix representing the time-frequency distribution of the signal; S303. Flatten the two-dimensional time-frequency matrix into a one-dimensional feature vector.

5. The communication method based on broadband interference suppression and target signal reconstruction according to claim 4, characterized in that: The sparse representation of the signal by using a sparse dictionary includes: S311. Using a sparse dictionary to sparsely represent the one-dimensional feature vector after time-frequency domain conversion; S312. Use a greedy algorithm to solve the sparse coefficient vector of the one-dimensional feature vector under the sparse dictionary.

6. The communication method based on broadband interference suppression and target signal reconstruction according to claim 5, characterized in that: The sparse dictionary includes: a standard transformation basis for characterizing universal components of a signal; and an overcomplete atom set generated by pre-training with typical target signal samples and used to characterize specific structural features.

7. The communication method based on broadband interference suppression and target signal reconstruction according to claim 1, wherein: The iterative reconstruction algorithm includes: S41. The compressed data is set as the initialization residual and an empty support set is initialized; S42. In each iteration, identifying the atom most relevant to the current residual from the sparse dictionary and adding the atom to the support set; S43. Based on the updated support set, solve the sparse coefficient and update the residual; S44. Repeat steps S42 and S43 until a preset iteration stop condition is met, thereby obtaining a reconstructed target signal.

8. The communication method based on broadband interference suppression and target signal reconstruction according to claim 1, wherein: The interference suppression strategy includes: setting a dynamic energy threshold, and suppressing or eliminating sparse atoms corresponding to frequency components whose amplitudes on the spectrum exceed the threshold in subsequent iterations.

9. The communication method based on broadband interference suppression and target signal reconstruction according to claim 1, wherein: The key parameters include compression rate, upper limit of sparsity of iterative reconstruction and energy threshold for interference suppression.

10. A communication system based on broadband interference suppression and target signal reconstruction, characterized in that: The system is used to execute the communication method based on broadband interference suppression and target signal reconstruction as described in any one of claims 1 to 9.

Images