A wideband spectrum signal blind detection method for a non-cooperative receiving scene

Abstract

This invention discloses a blind detection method for broadband spectrum signals in non-cooperative reception scenarios, belonging to the field of communication signal processing and spectrum sensing technology. Taking the power spectral density estimate of the broadband received signal as input, the method first preprocesses the power spectrum using small-scale closing operations; then, it constructs a multi-scale structuring element set and performs scale-by-scale morphological opening operations and residual peak-shaving iterations on the preprocessed spectrum data to obtain and correct the baseline estimation result; next, it applies one-dimensional total variational denoising to the baseline-corrected spectrum; finally, it uses a hysteresis dual-threshold mechanism to complete the candidate signal region screening and statistics. This invention effectively overcomes the problems of non-flat noise baseline fluctuations in complex electromagnetic environments and signal edge blurring under low signal-to-noise ratio conditions. It has advantages such as high baseline estimation accuracy, high detection probability, low false alarm rate, and strong robustness, and is suitable for signal detection scenarios in cognitive radio, broadband reconnaissance receivers, and spectrum monitoring equipment.

Description

Technical Field

[0001] This invention relates to the fields of communication signal processing and spectrum sensing technology, and specifically to a blind detection method for broadband spectrum signals in non-cooperative reception scenarios. Background Technology

[0002] With the rapid development of wireless communication and cognitive radio technologies, broadband spectrum sensing has become a crucial component in spectrum monitoring, electronic reconnaissance, broadband reception, and signal detection in non-cooperative communication environments. Existing broadband receiving equipment typically needs to perform blind detection of multiple unknown communication signals in complex electromagnetic environments to determine the presence of a target signal and estimate its spectral location.

[0003] However, in practical applications, thermal noise, nonlinear distortion, and external interference in broadband receiver front-end devices can cause the background noise power spectral density to exhibit non-flat, fluctuating, colored noise characteristics. This type of non-flat noise undermines the white noise assumption relied upon by traditional energy detection methods, making fixed-threshold detection prone to severe false alarms and missed detections. Simultaneously, under short-time observation conditions, the power spectral density estimation results also exhibit significant high-frequency random jitter, especially under low signal-to-noise ratio conditions, where weak signal peaks are easily submerged in baseline fluctuations and random spikes, further increasing the detection difficulty.

[0004] In existing technologies, statistical fitting methods, traditional morphological filtering methods, mean filtering methods, median filtering methods, or constant false alarm rate (CFAR) detection methods are commonly used to address the problem of spectral baseline estimation and denoising. Statistical fitting methods typically have high computational complexity, making them difficult to meet real-time processing requirements. Traditional single-scale morphological filtering struggles to balance signal peak suppression and baseline detail tracking. While multi-scale morphological iterative methods can partially improve this issue, the lack of preprocessing for negative outlier noise can easily lead to an overall downward bias in the erosion results during the opening operation, resulting in baseline estimation errors. Traditional linear smoothing methods, while reducing random noise, can blur signal edges, leading to an underestimation of bandwidth. Single-threshold detection mechanisms are prone to signal edge breaks or high false alarm rates.

[0005] Therefore, there is an urgent need for a broadband spectrum signal detection method that can achieve accurate baseline estimation, preserve signal edge characteristics, and effectively reduce false alarm rate against a complex colored noise background. Summary of the Invention

[0006] The purpose of this invention is to overcome the above-mentioned shortcomings in the prior art and provide a broadband spectrum signal blind detection method for non-cooperative reception scenarios, which solves the problems of non-flat colored noise interference in complex electromagnetic environments, signal edge blurring under low signal-to-noise ratio conditions, and insufficient robustness of traditional threshold detection.

[0007] To achieve the above objectives, the present invention employs the following technical solution: In a first aspect, the present invention proposes a blind detection method for broadband spectrum signals in non-cooperative reception scenarios, comprising the following steps: S1. Acquire the broadband received signal to be detected, and estimate the power spectral density of the broadband received signal to be detected to obtain the spectrum data to be processed. S2. Perform small-scale morphological closing operation on the spectrum data to be processed to suppress negative outlier noise spikes in the spectrum and obtain preprocessed spectrum data. S3. Construct a multi-scale structuring element set arranged from small to large, perform a scale-by-scale morphological opening operation on the preprocessed spectral data, calculate the residual spectrum at each scale, and perform residual peak clipping update according to a preset iteration threshold to obtain the noise baseline estimation result. S4. Use the noise baseline estimation results to perform baseline correction on the spectrum data to be processed to obtain baseline-corrected spectrum data; S5. Perform one-dimensional total variation denoising on the baseline correction spectrum data to obtain smooth spectrum data; S6. Perform hysteresis dual-threshold detection on the smoothed spectrum data. First, use the low threshold to extract candidate continuous regions, and then use the high threshold to screen the candidate continuous regions again to obtain the target signal detection mask. S7. Based on the target signal detection mask, output the area where the target signal exists, as well as the corresponding start and end frequencies, center frequency, bandwidth, and number of signals.

[0008] Furthermore, the power spectral density estimation in step S1 is achieved using fast Fourier transform, periodogram method, or Welch method.

[0009] Furthermore, in step S2, the length of the structuring element used in the morphological closing operation is less than the length of the smallest structuring element in the multi-scale structuring element set in step S3. The morphological closing operation uses small-scale flat structuring elements to perform a dilation-erosion process to fill local troughs and eliminate interference from low-level noise spikes on subsequent opening operations.

[0010] Furthermore, the multi-scale structuring element set in step S3 is determined by the minimum bandwidth, maximum bandwidth, and step bandwidth, and is converted into the corresponding structuring element length set by combining the sampling rate and the number of fast Fourier transform points; structuring elements of different scales are used to filter out signal peaks of different widths.

[0011] Furthermore, the residual peak-shaving update method in step S3 is as follows: At each scale, the residual spectrum is obtained by subtracting the current input spectrum from the morphological opening operation result; The frequency points in the residual spectrum that are greater than the preset iteration threshold are updated to the values ​​of the current scale morphological opening operation results at the corresponding frequency points; the frequency points in the residual spectrum that are not greater than the preset iteration threshold retain their original spectral values.

[0012] Furthermore, after completing all scale iterations in step S3, a short-window mean filtering step is also included to eliminate the slight staircase effect caused by morphological operations.

[0013] Furthermore, step S4 specifically includes: The baseline-corrected spectrum data is obtained by subtracting the noise baseline estimation result from the spectrum data to be processed.

[0014] Furthermore, the one-dimensional total variation denoising in step S5 is achieved by minimizing an objective function, which includes a data fidelity term and a one-dimensional total variation regularization term; wherein, the regularization parameter in the regularization term is adaptively set according to the residual noise standard deviation and signal length of the baseline corrected spectrum data.

[0015] Furthermore, in the hysteresis dual-threshold detection in step S6, the lower threshold is smaller than the higher threshold. The lower threshold is used to extract candidate continuous regions, and the higher threshold is used to eliminate pseudo-candidate regions formed by isolated noise spikes, thereby achieving the preservation of complete signal edges and the suppression of noise regions; including the following steps: S61. A low threshold is used to compare the smoothed spectrum data point by point, and all continuous frequency bands exceeding the low threshold are extracted as candidate continuous regions. S62. Traverse each candidate continuous region and determine whether there is at least one frequency point in each candidate continuous region whose amplitude exceeds the high threshold. S63. The candidate continuous region that satisfies the conditions of step S62 is determined as the target signal region, and the target signal detection mask is formed.

[0016] Furthermore, in step S7, differential operations and region statistics are performed on the target signal detection mask to obtain the total number of target signals and the start frequency, cutoff frequency, center frequency, and bandwidth of each target signal region.

[0017] Secondly, this invention proposes a broadband spectrum signal blind detection system for non-cooperative reception scenarios, which is used to implement the aforementioned broadband spectrum signal blind detection method for non-cooperative reception scenarios.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: I. Improve the accuracy of baseline estimation: By introducing small-scale closing operation preprocessing before multi-scale morphological opening operation, the interference of negative outlier noise on erosion operation is effectively eliminated, avoiding the problem of overall downward bias of baseline estimation results in traditional multi-scale morphological methods, and significantly improving the accuracy of baseline fitting under non-flat colored noise background.

[0019] 2. Enhance the sharpness of signal power spectrum edges: After baseline correction, one-dimensional total variation denoising is introduced, which can smooth random noise while maintaining the signal spectrum peak edges and bandwidth structure, avoiding the edge blurring phenomenon caused by traditional linear smoothing methods such as mean filtering.

[0020] 3. Improve detection rate and reduce false alarm rate: By using a hysteresis dual-threshold detection mechanism to classify and decide candidate continuous regions, the complete edges of the real signal are preserved, and false detections caused by isolated noise spikes are effectively eliminated, thereby improving the detection probability under low signal-to-noise ratio conditions and reducing the false alarm rate. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of a blind detection method for broadband spectrum signals in non-cooperative reception scenarios according to the present invention. Figure 2 This is a baseline detection result diagram of the actual sampled signal; Figure 3 The spectrum of the actual sampled signal after baseline correction and its detection results are shown. Detailed Implementation

[0022] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined in this application.

[0023] Please see the appendix Figure 1 This embodiment provides a blind detection method for broadband spectrum signals in non-cooperative reception scenarios, and uses actual broadband spectrum signals as an example for illustration.

[0024] In this embodiment, a signal acquisition and testing platform based on software-defined radio is constructed. The signal acquisition hardware uses a USRP B210, with a sampling rate set to 12.5 MHz and a center frequency set to 97.589 MHz. The receiver acquires broadband signals in the actual electromagnetic environment to obtain time-domain sampled data to be processed. Power spectral density estimation is performed on the time-domain sampled data to obtain the spectrum data to be detected.

[0025] The signal detection method in this embodiment includes the following steps: S1. Obtain actual broadband spectrum data.

[0026] The target frequency band is sampled using a software-defined radio platform to obtain a wideband received signal. Then, a Fast Fourier Transform and power spectral density estimation are performed on the wideband received signal to obtain the estimated power spectral density value of the sampled signal, which serves as the spectrum data to be processed. Figure 2 The measured results show that the actual sampled signal fluctuates significantly between 92 MHz and 98 MHz, exhibiting strong non-flat noise baseline characteristics, while also containing several high-amplitude spike interferences.

[0027] S2. Perform closed-loop preprocessing on the power spectral density estimate.

[0028] A morphological closing operation is performed on the power spectral density estimate using small-scale flat structuring elements to fill local troughs and eliminate negative outlier noise spikes, thereby reducing the impact of low-level noise spikes on the subsequent morphological opening erosion process and obtaining preprocessed spectral data.

[0029] The length of the structuring element used in the morphological closing operation is less than the length of the smallest structuring element in the multi-scale structuring element set described in step S3. In this embodiment, the length is set to 1 / 3000 of the power spectrum length.

[0030] S3. An improved multi-scale morphological filter is used to estimate the noise baseline.

[0031] In this embodiment, the frequency range corresponding to the structuring element scale is determined based on the typical occupied bandwidth range of the signal of interest in the application scenario to be detected, the receiver's spectral resolution, and the slowly varying scale of the background noise baseline, rather than depending on the center frequency, location, or precise bandwidth of a specific signal to be detected. Specifically, the minimum scale is used to cover the smaller occupied bandwidth that the signal of interest may have, and should be greater than the width of local random spikes in the power spectrum; the maximum scale is used to cover the larger occupied bandwidth that the signal of interest may have, and should be smaller than the slowly varying characteristic scale of the background noise baseline to avoid mistakenly removing baseline fluctuations as signal peaks. The step bandwidth is determined based on a trade-off between the spectral resolution, the dispersion of the bandwidth distribution of the signal of interest, and computational complexity. The smaller the step, the finer the scale coverage, but the computational load increases accordingly.

[0032] In this embodiment, based on the typical occupied bandwidth distribution of the signal of interest in the actual sampled spectrum, the frequency range corresponding to the structuring element scale is set to 100 kHz to 500 kHz, the step bandwidth is 100 kHz, and the iterative update threshold is 3 dB. The sampling rate is set to... The number of points in the Fast Fourier Transform is The power spectrum frequency resolution is For the first Frequency bandwidth corresponding to each scale Its structural elements can be taken as Furthermore, the length can be adjusted to an adjacent odd number to ensure that the structuring element is symmetrical about the current frequency. This constructs a multi-scale structuring element set arranged from smallest to largest, with structuring elements of different scales used to filter out signal peaks of different widths.

[0033] Morphological opening operations are performed on the preprocessed spectral data scale by scale. At each scale, the difference between the current input spectrum and the opening operation result is calculated to obtain the residual spectrum. If the residual at a certain frequency point is greater than the iterative update threshold, the original value of that frequency point is replaced with the opening operation result at the current scale; if the residual at a certain frequency point is not greater than the iterative update threshold, the original value of that frequency point is retained. Through the above scale-by-scale iterative peak-shaving process, the peak components of the signal are gradually filtered out to obtain a noise baseline estimation result that closely matches the actual noise floor profile.

[0034] In this embodiment, after all scale iterations are completed, a short-window mean filtering step is also included to eliminate the slight staircase effect introduced by morphological operations and obtain a smooth noise baseline.

[0035] like Figure 2 The experimental results show that the baseline estimation can accurately track the variation trend of non-flat colored noise in the actual sampling environment, providing a reliable noise reference for subsequent signal detection.

[0036] S4. Perform baseline correction.

[0037] Subtracting the noise baseline estimation result obtained in step S3 from the original power spectral density estimate yields the baseline-corrected spectral data S. c After baseline correction, the slow variation trend caused by noise floor fluctuations in the original spectrum is effectively removed, and the target signal component and random noise component are further separated.

[0038] S5. Perform one-dimensional total variation denoising on the baseline-corrected spectrum.

[0039] To further suppress high-frequency random jitter while preserving the peak edge characteristics of the signal spectrum, one-dimensional total variation denoising was performed on the baseline-corrected spectrum to obtain smoothed spectrum data S. t .

[0040] One-dimensional total variational denoising is achieved by minimizing an objective function, which includes a data fidelity term and a one-dimensional total variational regularization term: In this embodiment, the regularization parameter is set as follows: in, This is an estimate of the standard deviation of residual noise in the baseline-corrected spectral data. The length of the spectrum data to be processed.

[0041] This step effectively suppresses high-frequency noise spikes without significantly weakening signal edge and bandwidth information, thereby improving the stability of subsequent threshold detection.

[0042] S6. Perform delayed double threshold detection.

[0043] Since baseline correction of the power spectrum has been completed in step S4, the overall spectral noise floor after baseline correction converges to around 0 dB. Therefore, in this embodiment, the smoothed spectral data is first sorted by amplitude from smallest to largest, and the 30% of frequency points with the lowest amplitude are selected as noise floor samples. The mean or median of these samples is then calculated as the noise floor estimate. In the noise floor estimate Based on this, level biases are applied separately to obtain the low threshold and the high threshold: in, Less than In this embodiment, Take 5 dB, Take 8 dB.

[0044] First, a low threshold is adopted. The smoothed spectral data is compared point by point to extract all continuous candidate regions that meet the low threshold condition; then, each candidate region is traversed to determine whether there is at least one frequency point in each candidate region that exceeds the high threshold. If a candidate region exists, it is identified as the target signal region; if it does not exist, it is identified as a noise region and is eliminated.

[0045] The dual-threshold mechanism described above can both preserve the complete edges of the real signal and suppress false detections caused by isolated noise spikes.

[0046] S7. Output the detection results.

[0047] Based on the target signal region obtained in step S6, the detection mask is boundary-localized, and the start frequency, end frequency, center frequency, bandwidth, and number of target signals for each target signal are output. In this embodiment, as shown... Figure 3 The test results show that the method proposed in this invention can accurately filter out high-amplitude noise spikes in the actual test environment, avoid misjudging noise spikes as real signals, and effectively detect actual frequency band signals. This indicates that the invention has strong background noise adaptability and detection robustness in complex actual electromagnetic environments.

[0048] Furthermore, this embodiment demonstrates that by combining small-scale closing operation preprocessing, improved multi-scale morphological baseline estimation, one-dimensional total variational denoising, and hysteresis dual-threshold detection, problems such as non-flat noise baseline fluctuations, high-frequency random glitches, and weak signal edge blurring in the actual broadband spectrum can be effectively solved, thereby improving the accuracy and reliability of actual broadband signal detection.

[0049] It should be noted that the sampling rate, center frequency, structuring element scale range, step bandwidth, iterative update threshold, total variation denoising parameters, and dual-threshold parameters given in this embodiment are only specific settings in one preferred embodiment. In different application scenarios, the above parameters can be adjusted accordingly based on receiver bandwidth, spectral resolution, noise environment, and target signal characteristics without affecting the basic principles and implementation of this invention.

[0050] This embodiment also provides a broadband spectrum signal blind detection system for non-cooperative reception scenarios to implement the above method, including: The data acquisition module is used to acquire the broadband received signal to be detected and to estimate the power spectral density of the broadband received signal to be detected to obtain the spectrum data to be processed. The preprocessing module is used to perform small-scale morphological closing operations on the spectrum data to be processed, suppress negative outlier noise spikes in the spectrum, and obtain preprocessed spectrum data. The noise baseline estimation module is used to construct a multi-scale structuring element set arranged from small to large, perform a scale-by-scale morphological opening operation on the preprocessed spectral data, calculate the residual spectrum at each scale, and perform residual peak clipping update according to a preset iteration threshold to obtain the noise baseline estimation result. A baseline correction module is used to perform baseline correction on the spectrum data to be processed using the noise baseline estimation result, so as to obtain baseline-corrected spectrum data. A one-dimensional total variation denoising module is used to perform one-dimensional total variation denoising on the baseline corrected spectrum data to obtain smooth spectrum data. The hysteresis dual-threshold detection module is used to perform hysteresis dual-threshold detection on the smoothed spectrum data. First, it uses the low threshold to extract candidate continuous regions, and then uses the high threshold to screen the candidate continuous regions again to obtain the target signal detection mask. The target signal output module is used to output the presence area of ​​the target signal, as well as the corresponding start and end frequencies, center frequency, bandwidth, and number of signals, based on the target signal detection mask.

[0051] For the system embodiments, since they basically correspond to the method embodiments, relevant details can be found in the descriptions of the method embodiments; the implementation methods of the remaining modules will not be repeated here. The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the present invention according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0052] The system embodiments of the present invention can be applied to any device with data processing capabilities, such as a computer or other similar device. The system embodiments can be implemented in software, hardware, or a combination of both. Taking software implementation as an example, as a logical device, it is formed by the processor of any data processing device loading the corresponding computer program instructions from non-volatile memory into memory for execution.

[0053] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. Those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A blind detection method for broadband spectrum signals in non-cooperative reception scenarios, characterized in that, Includes the following steps: S1. Acquire the broadband received signal to be detected, and estimate the power spectral density of the broadband received signal to be detected to obtain the spectrum data to be processed. S2. Perform small-scale morphological closing operation on the spectrum data to be processed to suppress negative outlier noise spikes in the spectrum and obtain preprocessed spectrum data. S3. Construct a multi-scale structuring element set arranged from small to large, perform a scale-by-scale morphological opening operation on the preprocessed spectral data, calculate the residual spectrum at each scale, and perform residual peak clipping update according to a preset iteration threshold to obtain the noise baseline estimation result. S4. Use the noise baseline estimation results to perform baseline correction on the spectrum data to be processed to obtain baseline-corrected spectrum data; S5. Perform one-dimensional total variation denoising on the baseline correction spectrum data to obtain smooth spectrum data; S6. Perform hysteresis dual-threshold detection on the smoothed spectrum data. First, use the low threshold to extract candidate continuous regions, and then use the high threshold to screen the candidate continuous regions again to obtain the target signal detection mask. S7. Based on the target signal detection mask, output the area where the target signal exists, as well as the corresponding start and end frequencies, center frequency, bandwidth, and number of signals.

2. The broadband spectrum signal blind detection method for non-cooperative reception scenarios according to claim 1, characterized in that, In step S2, the length of the structuring element used in the morphological closing operation is less than the length of the minimum structuring element in the multi-scale structuring element set in step S3.

3. The broadband spectrum signal blind detection method for non-cooperative reception scenarios according to claim 1, characterized in that, The multi-scale structuring element set in step S3 is determined by the minimum bandwidth, maximum bandwidth, and step bandwidth, and is converted into the corresponding structuring element length set by combining the sampling rate and the number of fast Fourier transform points; structuring elements of different scales are used to filter out signal peaks of different widths.

4. The broadband spectrum signal blind detection method for non-cooperative reception scenarios according to claim 1, characterized in that, The residual peak smoothing update method in step S3 is as follows: At each scale, the residual spectrum is obtained by subtracting the current input spectrum from the morphological opening operation result; The frequency points in the residual spectrum that are greater than the preset iteration threshold are updated to the values ​​of the current scale morphological opening operation results at the corresponding frequency points; The frequency points in the residual spectrum that are not greater than a preset iteration threshold retain their original spectral values.

5. The broadband spectrum signal blind detection method for non-cooperative reception scenarios according to claim 1, characterized in that, After completing all scale iterations, step S3 also includes a short-window mean filtering step on the iteration output results.

6. The broadband spectrum signal blind detection method for non-cooperative reception scenarios according to claim 1, characterized in that, Step S4 is as follows: The baseline-corrected spectrum data is obtained by subtracting the noise baseline estimation result from the spectrum data to be processed.

7. The broadband spectrum signal blind detection method for non-cooperative reception scenarios according to claim 1, characterized in that, The one-dimensional total variation denoising in step S5 is achieved by minimizing an objective function, which includes a data fidelity term and a one-dimensional total variation regularization term; wherein, the regularization parameter in the regularization term is adaptively set according to the residual noise standard deviation and signal length of the baseline corrected spectrum data.

8. The broadband spectrum signal blind detection method for non-cooperative reception scenarios according to claim 1, characterized in that, The hysteresis dual-threshold detection in step S6 includes the following steps: S61. A low threshold is used to compare the smoothed spectrum data point by point, and all continuous frequency bands exceeding the low threshold are extracted as candidate continuous regions. S62. Traverse each candidate continuous region and determine whether there is at least one frequency point in each candidate continuous region whose amplitude exceeds the high threshold. S63. The candidate continuous region that satisfies the conditions of step S62 is determined as the target signal region, and the target signal detection mask is formed.

9. A blind detection method for broadband spectrum signals in non-cooperative reception scenarios according to claim 1, characterized in that, In step S7, differential operations and region statistics are performed on the target signal detection mask to obtain the total number of target signals and the start frequency, cutoff frequency, center frequency, and bandwidth of each target signal region.

10. A broadband spectrum signal blind detection system for non-cooperative reception scenarios, used to implement the broadband spectrum signal blind detection method for non-cooperative reception scenarios in claim 1, characterized in that, include: The data acquisition module is used to acquire the broadband received signal to be detected and to estimate the power spectral density of the broadband received signal to be detected to obtain the spectrum data to be processed. The preprocessing module is used to perform small-scale morphological closing operations on the spectrum data to be processed, suppress negative outlier noise spikes in the spectrum, and obtain preprocessed spectrum data. The noise baseline estimation module is used to construct a multi-scale structuring element set arranged from small to large, perform a scale-by-scale morphological opening operation on the preprocessed spectral data, calculate the residual spectrum at each scale, and perform residual peak clipping update according to a preset iteration threshold to obtain the noise baseline estimation result. A baseline correction module is used to perform baseline correction on the spectrum data to be processed using the noise baseline estimation result, so as to obtain baseline-corrected spectrum data. A one-dimensional total variation denoising module is used to perform one-dimensional total variation denoising on the baseline corrected spectrum data to obtain smooth spectrum data. The hysteresis dual-threshold detection module is used to perform hysteresis dual-threshold detection on the smoothed spectrum data. First, it uses the low threshold to extract candidate continuous regions, and then uses the high threshold to screen the candidate continuous regions again to obtain the target signal detection mask. The target signal output module is used to output the presence area of ​​the target signal, as well as the corresponding start and end frequencies, center frequency, bandwidth, and number of signals, based on the target signal detection mask.

Images