A frequency domain wideband signal detection method, an electronic device and a storage medium

Abstract

The application discloses a frequency domain wideband signal detection method, an electronic device and a storage medium. The method comprises the following steps: performing bottom noise estimation on current frame spectrum data in a wideband frequency band to obtain a bottom noise estimation value of each frequency point; dividing the current frame spectrum data into a first preset number of first data segments; for each first data segment, determining a relative bottom noise gain value corresponding to each frequency point in the first data segment according to the statistical characteristics of the first data segment; adding the relative bottom noise gain value corresponding to each frequency point to the bottom noise estimation value of the frequency point to obtain an adaptive detection threshold of each frequency point; and determining whether there is a signal according to the adaptive detection threshold of each frequency point. The scheme can improve the accuracy and stability of signal detection.

Description

Technical Field

[0001] This application generally relates to the field of wireless communication and signal detection technology. More specifically, this application relates to a frequency-domain broadband signal detection method, electronic device, and computer-readable storage medium. Background Technology

[0002] Currently, signal detection in the frequency domain typically follows a general processing flow: first, necessary preprocessing operations are performed on the received spectral data; then, the spectral noise floor level is estimated; next, a detection threshold is calculated based on the estimated noise floor; and finally, the presence or absence of a signal at each frequency point is determined according to a predetermined decision logic. Common detection methods include energy detection, matched filtering, and template matching. These methods mostly pre-set a fixed threshold and determine the presence of a signal by comparing the relative magnitude of the energy at each frequency point with that threshold. Constant false alarm rate (CFAR) detection technology, on the other hand, is based on the principle of maintaining a constant false alarm rate. It calculates the threshold based on the false alarm rate and can dynamically adjust the threshold value as the spectral background changes. This technology is divided into two processing methods based on the background distribution: non-Rayleigh distribution and Rayleigh distribution.

[0003] Existing frequency domain signal detection methods have significant shortcomings in practical engineering applications. First, for fixed-threshold detection methods, the power level of broadband signals has a large dynamic range in complex electromagnetic environments, and a single fixed threshold cannot adapt to all signal scenarios. When the threshold is set too high, low-power broadband signals are easily missed; while when the threshold is set too low, background noise and interference are easily misjudged as signals, leading to a sharp increase in the false alarm rate. Although constant false alarm rate (CFAR) detection methods can adaptively adjust the threshold to some extent according to background changes, they still rely on specific mathematical models and preset false alarm rate parameters. This leads to two limitations: firstly, the spectrum distribution in actual electromagnetic environments often cannot strictly match the theoretical model, resulting in deviations in the calculated detection threshold; secondly, the spectrum within a frequency band fluctuates greatly, and the characteristics of each frequency band differ significantly, meaning that the uniform threshold generated by CFAR detection lacks the ability to adjust for the characteristics of different frequency bands. Furthermore, during actual reception, single frames or multiple consecutive frames of data may be abnormally saturated. If such saturated frames are directly used for detection without identification, a large number of false detection results will be generated, seriously affecting the overall detection performance of the system.

[0004] In view of this, this application provides a frequency domain broadband signal detection method to improve the accuracy and stability of signal detection. Summary of the Invention

[0005] In order to at least address one or more of the technical problems mentioned above, this application proposes solutions in several aspects that can improve the accuracy and stability of signal detection.

[0006] In a first aspect, this application provides a method for detecting a frequency-domain broadband signal, comprising:

[0007] The noise floor is estimated by performing a frequency-by-frequency point estimate on the current frame spectrum data within the broadband bandwidth; The current frame spectrum data is equally divided into a first preset number of first data segments; For each first data segment, based on the statistical characteristics of the first data segment, the relative noise floor gain value corresponding to each frequency point within the first data segment is determined; the relative noise floor gain value is the amount by which the detection threshold of each frequency point is raised relative to its noise floor estimate. The relative noise floor gain value corresponding to each frequency point is added to its noise floor estimate to obtain the adaptive detection threshold for each frequency point. The presence or absence of a signal is determined point by point based on the adaptive detection threshold.

[0008] In some embodiments, after obtaining the frequency-point-by-frequency noise floor estimate and before equally dividing the current frame spectrum data into a first preset number of first data segments, the method further includes: The current frame spectrum data is equally divided into a second preset number of second data segments, and the maximum spectrum value in each second data segment is extracted to form a second maximum value sequence. The second maximum value sequence is normalized and its standard deviation is calculated; Based on the standard deviation, the preset standard deviation threshold, and the saturation frame count, determine whether the current frame spectrum data is an abnormal saturation frame; If the current frame spectrum data is an abnormally saturated frame, then discard the current frame spectrum data and perform noise floor estimation on the next frame spectrum data; If the current frame spectrum data is a normal saturated frame, then the subsequent equal division and determination of the relative noise floor gain value will continue to be performed on the current frame spectrum data.

[0009] In some embodiments, determining whether the current frame spectral data is an abnormally saturated frame based on the standard deviation, a preset standard deviation threshold, and the saturation frame count includes: Determine whether the standard deviation is less than a preset standard deviation threshold; If the standard deviation is greater than or equal to the preset standard deviation threshold, then the current frame spectrum data is determined to be a normal frame; If the standard deviation is less than the preset standard deviation threshold, then the current frame spectrum data is determined to be a suspected saturated frame, and the saturated frame count is incremented. Determine whether the saturation frame count is greater than a preset counting threshold; If the saturation frame count is less than or equal to the preset count threshold, then the current frame spectrum data is determined to be an abnormal saturation frame. If the saturation frame count is greater than the preset count threshold, then the current frame spectrum data is determined to be a normal saturation frame.

[0010] In some embodiments, for each first data segment, the relative noise floor gain value corresponding to each frequency point within the first data segment is determined based on the statistical characteristics of the first data segment, including: The first data segment is equally divided into a third preset number of third data sub-segments, and the maximum spectral value in each third data sub-segment is recorded to obtain the first maximum value sequence of the first data segment. Select the minimum and maximum values ​​from the first maximum value sequence, and denote them as the first maximum value and the second maximum value, respectively; The noise floor estimates at the frequency points corresponding to the first maximum and the second maximum are obtained respectively and denoted as the first noise floor value and the second noise floor value. Based on the first maximum value, the second maximum value, the first noise floor value, and the second noise floor value, the relative noise floor gain value corresponding to each frequency point in the first data segment is determined.

[0011] In some embodiments, determining the relative noise floor gain value corresponding to each frequency point within the first data segment based on the first maximum value, the second maximum value, the first noise floor value, and the second noise floor value includes: Calculate the difference between the first maximum value and the first noise floor value, and record it as the first difference value; calculate the difference between the second maximum value and the second noise floor value, and record it as the second difference value; Based on the first difference, the second difference, and the preset judgment threshold, the relative noise floor gain value corresponding to each frequency point in the first data segment is determined.

[0012] In some embodiments, determining the relative noise floor gain value corresponding to each frequency point within the first data segment based on the first difference, the second difference, and a preset judgment threshold includes: Calculate the difference between the second difference and the first difference, and record it as the target difference; Determine whether the target difference is greater than the preset judgment threshold; If the target difference is greater than the preset judgment threshold, then the relative noise floor gain value corresponding to each frequency point in the first data segment is determined to be between the first difference and the second difference; If the target difference is less than or equal to the preset judgment threshold, then the relative noise floor gain value corresponding to each frequency point in the first data segment is determined to be greater than the second difference.

[0013] In some embodiments, the formula for the relative noise floor gain value to be between the first difference and the second difference is expressed as follows: G = D1 + k(D2 - D1); Where G represents the relative noise floor gain value, D2 represents the second difference, D1 represents the first difference, and k represents the preset gain coefficient, and 0 <k<1。

[0014] In some embodiments, the formula for the relative noise floor gain value being greater than the second difference is expressed as follows: G = D² + C; Where G represents the relative noise floor gain value, D2 represents the second difference value, and C is a preset constant, and 0 <C<10。

[0015] In a second aspect, this application provides an electronic device, comprising: Processor; and A memory that stores program instructions for frequency domain broadband signal detection, which, when executed by a processor, enable the aforementioned frequency domain broadband signal detection method.

[0016] In a third aspect, this application provides a computer-readable storage medium having stored thereon program instructions for frequency domain broadband signal detection, which, when executed by a processor, implement the aforementioned frequency domain broadband signal detection method.

[0017] The frequency domain broadband signal detection described above divides the spectral data within the broadband band into multiple data segments and independently calculates the relative noise floor gain value of each frequency point within each segment based on its own statistical characteristics. This allows the detection threshold to adaptively adjust to the changes in spectral fluctuation characteristics in different regions of the band, effectively avoiding the problems of missed detections and false alarms that are easily caused by using a single fixed threshold or a uniform threshold across the entire band when the signal strength is uneven. At the same time, since the detection threshold of each frequency point is obtained by superimposing the gain boost determined based on the statistical characteristics of the segment on the noise floor estimate, the threshold can always closely match the actual background environment of different frequency bands. This is beneficial for maintaining stable detection performance in complex and variable electromagnetic environments, thereby improving the accuracy of signal detection. Attached Figure Description

[0018] The above and other objects, features, and advantages of exemplary embodiments of this application will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of this application are illustrated by way of example and not limitation, and the same or corresponding reference numerals denote the same or corresponding parts, wherein: Figure 1 An exemplary flowchart of a frequency domain broadband signal detection method provided in some embodiments of this application is shown; Figure 2 An exemplary flowchart of a frequency domain broadband signal detection method according to other embodiments of this application is shown; Figure 3An exemplary flowchart illustrating the process of determining abnormally saturated frames according to an embodiment of this application is shown; Figure 4 This paper shows a schematic diagram of the statistical maximum value sequence in Simulation 1 of this application; Figure 5 The graph shows the analysis results using a pre-set relative noise floor gain value that is small; Figure 6 The graph shows the analysis results using a pre-set relative noise floor gain value, where the value is relatively large; Figure 7 The diagram shows the analysis results of calculating the relative noise floor gain value using the piecewise adaptive method of this application; Figure 8 An exemplary structural block diagram of an electronic device according to an embodiment of this application is shown. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] It should be understood that the terms "comprising" and "including" used in the specification and claims of this application indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0021] It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this specification and claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this specification and claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations.

[0022] As used in this specification and claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0023] The specific embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0024] Figure 1 An exemplary flowchart of a frequency domain broadband signal detection method 100 provided in some embodiments of this application is shown. It is understood that method 100 can be executed by any suitable device with data processing capabilities, such as, but not limited to, terminal devices and servers.

[0025] like Figure 1 As shown, in step S101, the noise floor of the current frame spectrum data in the broadband band is estimated to obtain the noise floor estimate value for each frequency point.

[0026] Since the noise floor of the actual broadband spectrum is usually not a flat horizontal line, but rather exhibits a slow undulating shape as the frequency changes, for example, due to factors such as the frequency response of the receiving link, the roll-off characteristics of the bandpass filter, and the difference in electromagnetic interference intensity in different frequency bands, the background noise level at different frequency points may vary significantly. Therefore, it is necessary to estimate the noise floor independently for each frequency point in order to obtain a noise floor curve that can accurately reflect the actual background noise level at each frequency point.

[0027] In practical implementation, smoothing filtering can be used to estimate the noise floor of the spectral data. Smoothing filtering effectively eliminates random fluctuations in the spectrum, ensuring that the estimated noise floor accurately reflects the background noise level within the current frequency band. Those skilled in the art should understand that smoothing filtering can employ moving average filtering, median filtering, or other suitable smoothing methods, as long as a stable noise floor estimation result is obtained; this application does not impose specific limitations on this. Noise floor estimation is the basis for subsequent calculations of relative noise floor gain and determination of the detection threshold; the accuracy of the noise floor estimation directly affects the final detection performance.

[0028] Next, in step S102, the current frame spectrum data is equally divided into a first preset number of first data segments. In other words, each first data segment contains the same number of frequency points.

[0029] The purpose of segmented processing in this application is that the spectral characteristics of different regions within a broadband bandwidth, such as signal strength, noise level, and interference distribution, often vary significantly. Using a uniform gain value and detection threshold across the entire bandwidth would inevitably fail to meet the detection needs of all frequency bands. Strong signal bands may experience rampant false alarms, while weak signal bands may suffer from severe missed detections. By equally dividing the entire broadband bandwidth, subsequent segments can perform gain calculations based on the same frequency base, ensuring consistency in the calculation benchmark between segments. Simultaneously, each segment can independently calculate the most suitable relative noise floor gain based on its own spectral statistical characteristics, thereby achieving adaptive matching between the detection threshold and the spectral characteristics of each frequency band.

[0030] It is important to note that the initial preset number of equal segments should not be too large, as this may lead to false alarms. This is because too many segments result in too few frequency points within each segment, making the statistical information within the segment insufficiently representative. This can easily lead to misinterpreting local random fluctuations as signal characteristics, thus generating false alarms. Conversely, the number of segments should not be too small either; otherwise, the segments will be too coarse and unable to reflect the spectral differences between different frequency bands, defeating the purpose of adaptive segmentation. Typically, the number of frequency points in each first data segment can be set first, and then the initial preset number can be derived in reverse, allowing the subsequently determined gain sequence to adapt to the spectral conditions.

[0031] Preferably, the number of frequency points in each first data segment ranges from 1500 to 2500. Engineering practice has verified that setting the number of frequency points in each first data segment within this range maintains sufficient sensitivity to local spectral features while avoiding unnecessary false alarms caused by overly fine segmentation.

[0032] Next, in step S103, for each first data segment, the relative noise floor gain value corresponding to each frequency point within the first data segment is determined based on the statistical characteristics of the first data segment. Here, the relative noise floor gain value is the increase in the detection threshold of each frequency point relative to its estimated noise floor value.

[0033] Unlike existing methods that directly set an absolute detection threshold, this application uses a method of noise floor plus relative gain to determine the detection threshold. First, the background noise level (i.e., noise floor) at each frequency point is estimated. Then, an adaptive gain value is superimposed on the noise floor as the detection threshold. Therefore, the detection threshold can automatically adjust according to changes in the noise floor. Even if the noise floor fluctuates across different frequency bands, the threshold can always maintain a reasonable increase relative to the noise floor, thus ensuring consistent detection sensitivity under different noise levels.

[0034] In practice, step S103 can be further refined into the following operations: First, the first data segment is equally divided into a third preset number of third data sub-segments, and the maximum spectral value within each third data sub-segment is recorded to obtain the first maximum sequence of the first data segment. The third preset number can be flexibly set according to actual needs, and its selection determines the length of the first maximum sequence, thus affecting the fineness of the characterization of the spectral envelope within the first data segment. Specifically, a larger value for the third preset number results in a finer characterization of the envelope, but also increases the computational load; a smaller value results in a coarser envelope characterization, but with higher computational efficiency. In specific implementations, the third preset number can be determined comprehensively based on the spectral resolution, the total number of frequency points, and the desired envelope smoothness; this application does not impose specific limitations on this. The first maximum sequence actually reflects the shape of the envelope on the spectrum of the first data segment. By analyzing the upper envelope, the peak characteristics of the signal within the segment can be effectively captured, while avoiding interference caused by random fluctuations at the frequency point level. By taking the maximum value of each sub-segment, fragmented noise spikes in the spectrum can be filtered out, preserving the true signal envelope characteristics, making subsequent statistical analysis based on the maximum sequence more stable and reliable.

[0035] Then, the minimum and maximum values ​​are selected from the first maximum sequence and denoted as the first maximum and the second maximum, respectively. The first maximum represents the lowest level of the signal peak within the first data segment, and the second maximum represents the highest level of the signal peak within the first data segment. Next, the estimated noise floor values ​​at the frequency points corresponding to the first and second maximums are obtained and denoted as the first noise floor value and the second noise floor value, respectively. Finally, based on the first maximum, the second maximum, the first noise floor value, and the second noise floor value, the relative noise floor gain value corresponding to each frequency point within the first data segment is determined.

[0036] Furthermore, based on the first maximum, second maximum, first noise floor value, and second noise floor value, the relative noise floor gain value corresponding to each frequency point within the first data segment is determined. This can be achieved by performing the following operations: First, calculate the difference between the first maximum and the first noise floor value, denoted as the first difference; then calculate the difference between the second maximum and the second noise floor value, denoted as the second difference. The first difference reflects the degree of increase in the lowest peak value relative to the noise floor at its corresponding frequency point within the segment, while the second difference reflects the degree of increase in the highest peak value relative to the noise floor at its corresponding frequency point within the segment. By comparing the relationship between these two differences, it can be determined whether there are real signal components within the segment, rather than simply noise floor fluctuations. Then, based on the first difference, the second difference, and a preset judgment threshold, the relative noise floor gain value corresponding to each frequency point within the first data segment is determined.

[0037] In the embodiments of this application, the relative noise floor gain value corresponding to each frequency point in the first data segment is determined based on the first difference, the second difference, and the preset judgment threshold. Specifically, the relative noise floor gain value corresponding to each frequency point in the first data segment can be determined based on whether the difference between the second difference and the first difference is greater than the preset judgment threshold.

[0038] Specifically, the following operations can be performed: First, calculate the difference between the second difference and the first difference, denoted as the target difference. This target difference is essentially a measure of the dynamic range of the signal peaks within the first data segment, i.e., the degree of difference between the strongest and weakest peaks within the first data segment. Next, determine whether the target difference is greater than a preset judgment threshold. The preset judgment threshold is used to distinguish whether a real signal exists within the segment. If the target difference is greater than the preset judgment threshold, it indicates that the peak levels at different frequencies within the first data segment differ significantly, exceeding the range that normal random fluctuations in the noise floor can explain. Therefore, it can be considered that a real signal component exists within the segment. Conversely, if the target difference is less than or equal to the preset judgment threshold, it indicates that the peak levels at each frequency within the first data segment are relatively close, and the difference is insufficient to be considered a signal. Therefore, it can be considered that no signal exists within the segment. Based on this, if the target difference is greater than the preset judgment threshold, the relative noise floor gain value corresponding to each frequency point within the first data segment is determined to be between the first difference and the second difference. Conversely, if the target difference is less than or equal to the preset judgment threshold, then the relative noise floor gain value corresponding to each frequency point in the first data segment is determined to be greater than the second difference.

[0039] In practice, the formula (1) for the relative noise floor gain value to be between the first difference and the second difference can be expressed as follows: G=D1+k(D2-D1) formula (1); Where G represents the relative noise floor gain value, D2 represents the second difference, D1 represents the first difference, and k represents the preset gain coefficient, and 0 <k<1。

[0040] The specific value of the preset gain coefficient k can be flexibly set according to actual engineering needs. When a relatively aggressive signal capture is required, a smaller k value can be selected to make the threshold lower and the detection sensitivity higher; when a relatively conservative approach is needed to avoid false alarms, a larger k value can be selected to make the threshold higher. The typical range of k value selection is between 0.3 and 0.7, but this application is not limited to this.

[0041] Formula (2) for the relative noise floor gain value being greater than the second difference value can be expressed as follows: Formula (2) for G=D2+C; Where G represents the relative noise floor gain value, D2 represents the second difference value, and C is a preset constant, and 0 <C<10。

[0042] Formula (2) applies to frequency bands determined to be without signal. In this case, the relative noise floor gain value within the band is set to be greater than the second difference, i.e., higher than the highest peak rise level within the band, so that the spectral data within the band are all below the detection threshold, effectively avoiding false alarms. The preset constant C can be arbitrarily selected between 0 and 10. The larger the value of C, the easier it is to raise the detection threshold of the band, and the stronger the anti-false alarm capability. However, the value of C should not be too large, so as not to fail to detect the signal in the band in a timely manner when it appears in subsequent frames. Preferably, C can be a value between 1 and 5.

[0043] Next, in step S104, the relative noise floor gain value corresponding to each frequency point is added to its noise floor estimate value to obtain the adaptive detection threshold for each frequency point.

[0044] Specifically, firstly, a gain sequence of the same length as the spectral data is constructed: for each first data segment, all frequency points within that segment share the same relative noise floor gain value, i.e., the G value calculated for that segment. Therefore, the gain sequence exhibits a stepped shape, with the same gain value for each frequency point within the same segment, and possible jumps in gain values ​​between segments. The gain sequence can be represented as {G1, G1, ..., G2, G2, ..., GS2, GS2, ...}, where G1, G2, ..., GS2 are the relative noise floor gain values ​​for each first data segment, and the number of repetitions of each gain value is equal to the number of frequency points contained in the corresponding first data segment. Then, for each frequency point, its detection threshold is equal to the noise floor estimate of that frequency point plus the relative noise floor gain value of the first data segment in which that frequency point is located. Since the relative noise floor gain values ​​of different first data segments can be different, the detection threshold throughout the entire broadband band exhibits a segmented variation characteristic, which can match the spectral fluctuation characteristics of different regions within the band. The gain value of the frequency band with signal is moderate, which can ensure that the signal can be reliably detected by passing the threshold, while suppressing noise false alarms as much as possible; the gain value of the frequency band without signal is higher, raising the threshold to a level that will not produce false alarms.

[0045] Finally, in step S105, the presence or absence of a signal is determined point by point according to the adaptive detection threshold.

[0046] Specifically, for each frequency point, if the spectral amplitude of that frequency point is greater than the adaptive detection threshold at that frequency point, it is determined that a signal has been detected at that frequency point; conversely, if the spectral amplitude of that frequency point is less than or equal to the adaptive detection threshold at that frequency point, it is determined that no signal has been detected at that frequency point. By comparing frequency points one by one, a binary detection result sequence of the same length as the spectrum can be obtained, where frequency points that exceed the threshold are marked as having a signal, and frequency points that do not exceed the threshold are marked as having no signal.

[0047] After obtaining the frequency-by-frequency decision results, signal clustering (or connected component extraction) can be performed: the binarized detection result sequence is scanned from left to right, and consecutive signal frequencies are grouped into independent signal candidate segments. If two signal candidate segments are separated by a non-signal frequency, they are considered as two different signals. Through this step, complete signals can be extracted from the scattered single-frequency decision results, preparing for subsequent parameter estimation.

[0048] After determining the presence or absence of a signal and extracting the independent signal, the process may further include steps to estimate the center frequency and bandwidth of the detected signal.

[0049] Center frequency estimation can be achieved using various methods: for example, the midpoint method, which takes the arithmetic mean of the starting and ending frequencies of the signal segment as the center frequency; another example is the peak method, which takes the frequency corresponding to the frequency point with the highest energy in the signal segment as the center frequency; yet another example is the energy-weighted centroid method, which calculates the weighted average of the frequencies using the energy of each frequency point in the signal segment as the weight, and in addition, methods such as parabolic interpolation can be used to achieve sub-frequency accuracy center frequency estimation.

[0050] Bandwidth estimation can also employ various methods: for example, the threshold bandwidth method, which directly uses the difference between the start and end frequencies of the signal segment as the bandwidth, a method that best suits the detection logic of this application; another example is the 3dB bandwidth method, which finds the frequency points where the energy drops to half the peak value (i.e., -3dB) on both sides of the signal peak, and the difference between these two points is the 3dB bandwidth, which is the most standard bandwidth definition in the field of signal processing; in addition, there are statistical bandwidth estimation methods such as energy equivalent bandwidth. Those skilled in the art can choose according to actual needs, and this application does not impose specific limitations on this.

[0051] The above combination Figure 1 The frequency domain broadband signal detection method described herein divides the spectral data within the broadband band into multiple data segments and independently calculates the relative noise floor gain value of each frequency point within each data segment based on its own statistical characteristics. This allows the detection threshold to adaptively adjust to the changes in spectral fluctuation characteristics in different regions of the frequency band, effectively avoiding the problems of missed detections and false alarms that are easily caused by using a single fixed threshold or a uniform threshold across the entire frequency band when the signal strength is uneven. At the same time, since the detection threshold of each frequency point is obtained by superimposing the gain boost determined based on the statistical characteristics of the segment on the noise floor estimate, the threshold can always closely match the actual background environment of different frequency bands, which is conducive to maintaining stable detection performance in complex and variable electromagnetic environments, thereby improving the accuracy of signal detection.

[0052] To eliminate false alarms caused by abnormal saturation frames and further improve the accuracy and stability of signal detection, some embodiments of this application may also include the operation of detecting and removing abnormal saturation frames. In practical engineering applications, due to the complexity of the electromagnetic environment, a single frame or several consecutive frames may have their spectrum abnormally inflated to near full scale, known as abnormal saturation frames. The causes of abnormal saturation frames are varied, including sudden strong pulse interference, instantaneous saturation of the receiving link, and AD sampling overflow. If abnormal saturation frames are not identified and removed, and are directly sent to subsequent detection processes, almost the entire frame spectrum will exceed the detection threshold, generating a large number of false alarms and severely contaminating the detection results. Therefore, judging and removing abnormal saturation frames from each frame of data before formally performing signal detection is a crucial preliminary step to ensure detection reliability.

[0053] Figure 2 An exemplary flowchart of a frequency domain broadband signal detection method 200 according to other embodiments of this application is shown. Figure 2 The method 200 shown is in Figure 1 This is a further improvement based on method 100 shown. It should be noted that, as... Figure 2 The timing of each step shown can be combined with the above. Figure 1 After obtaining the frequency-point-by-frequency noise floor estimate in step S101, and before dividing the current frame spectrum data equally into a first preset number of first data segments in step S102.

[0054] like Figure 2 As shown, in step S201, the current frame spectrum data is equally divided into a second preset number of second data segments, and the maximum spectral value within each second data segment is extracted to form a second maximum value sequence. This second maximum value sequence is actually a sampling sequence of the upper envelope of the current frame spectrum data, and its shape can reflect the overall contour features of the frame spectrum.

[0055] In practice, the second preset number is related to the frequency resolution. A higher frequency resolution, meaning a narrower bandwidth represented by each frequency point, results in a larger total number of frequency points within the entire broadband band, allowing for a correspondingly larger value for the second preset number. Conversely, a lower frequency resolution allows for a correspondingly smaller value for the second preset number. Preferably, the second preset number can be between 256 and 2048.

[0056] In step S202, the second maximum value sequence is normalized and its standard deviation is calculated. The purpose of normalization is to eliminate the influence of overall amplitude differences between different frames on the standard deviation calculation, so that the standard deviation can purely reflect the fluctuation of the envelope on the spectrum of that frame. Specifically, normalization can be performed using other commonly used normalization methods such as maximum-minimum normalization, division by mean or median normalization, as long as it can eliminate the absolute differences in overall amplitude and preserve the relative fluctuation characteristics of the data. This application does not impose any specific limitations on this. After the normalization is completed, the standard deviation of the normalized second maximum value sequence is calculated again for subsequent abnormal saturation frame judgment.

[0057] In step S203, based on the standard deviation, a preset standard deviation threshold, and the number of saturated frames, it is determined whether the current frame's spectral data is an abnormally saturated frame. The principle behind this determination is as follows: the envelope of a normal frame's spectrum should have a certain degree of fluctuation, resulting in a larger normalized standard deviation; while in an abnormally saturated frame, because the spectral values ​​across the entire frequency band are raised to near saturation levels, the envelope tends to be flat, resulting in a smaller normalized standard deviation. Therefore, by comparing the relationship between the normalized standard deviation and the preset standard deviation threshold, it is possible to preliminarily determine whether the current frame is a saturated frame. Preferably, the preset standard deviation threshold can be set between 8 and 12.

[0058] Next, if the current frame's spectral data is an abnormally saturated frame, then the current frame's spectral data is discarded, and the noise floor estimation operation is performed on the next frame's spectral data, i.e., the process returns to step S101 to continue execution. Discarding abnormally saturated frames can effectively avoid large-scale false alarms introduced by saturation and ensure the reliability of the detection results.

[0059] Conversely, if the current frame spectrum data does not belong to an abnormally saturated frame, the subsequent equal division and determination of the relative noise floor gain value are performed on the current frame spectrum data, that is, steps S102 and S103 are continued.

[0060] It's important to note that when multiple consecutive frames are identified as saturated frames, it may indicate the presence of a strong signal covering the entire frequency band in the current channel, rather than abnormal saturation of the receiving device. Continuing to discard frame data in this situation would lead to missed detections of genuine signals. Therefore, this application uses a saturated frame counting mechanism to record the number of consecutive saturated frames. When the number of consecutive saturated frames exceeds a preset counting threshold, this saturation phenomenon is considered a continuous, normal signal scenario rather than an occasional anomaly. These frames are then treated as normal frames, allowing them to enter the subsequent signal detection process, thus achieving a balance between avoiding false alarms and preventing missed detections.

[0061] For ease of description, the following will refer to such frames, which have saturation characteristics but are treated as normal frames, as normal saturated frames, to distinguish them from normal frames in the usual sense (i.e., non-saturated frames with sufficient fluctuations in the upper envelope). However, it should be understood that their essence is a normal frame under a continuous strong signal scenario, rather than an abnormal saturation state.

[0062] Figure 3 An exemplary flowchart of a process 300 for determining abnormal saturated frames based on standard deviation, a preset standard deviation threshold, and saturated frame count, according to an embodiment of this application, is shown. It will be understood that the following description, in conjunction with... Figure 3 The description is a specific implementation of the aforementioned step S203. Therefore, in conjunction with the preceding text... Figure 2 The described characteristics can be similarly applied here.

[0063] like Figure 3 As shown, in step S301, it is determined whether the standard deviation is less than a preset standard deviation threshold.

[0064] If the standard deviation is greater than or equal to the preset standard deviation threshold, then step S302 is executed to determine that the current frame spectrum data is a normal frame. This indicates that the upper envelope of the current frame spectrum has sufficient fluctuation and conforms to the characteristics of a normal spectrum. There is no need to perform abnormal handling operations such as saturation frame count accumulation and frame data discarding. The subsequent signal detection process, such as equal division and gain calculation, can be directly entered.

[0065] Conversely, if the standard deviation is less than the preset standard deviation threshold, step S303 is executed to determine that the current frame spectrum data is a suspected saturated frame, and the saturated frame count is incremented. A standard deviation less than the preset standard deviation threshold indicates that the upper envelope of the current frame spectrum is relatively flat, and it is initially judged to be a suspected saturated frame. However, it is not ultimately determined to be an abnormal saturated frame at this time, but further judgment is made through the saturated frame count.

[0066] Next, in step S304, it is determined whether the saturation frame count is greater than a preset counting threshold. In actual operation, the preset counting threshold can be set according to actual needs. Preferably, the preset counting threshold can be set between 10 and 30. The setting of this counting threshold reflects the system's tolerance for consecutive saturation frames. The smaller the counting threshold is set, the more sensitive the system is to abnormal saturation frames and the faster the response; the larger the counting threshold is set, the more the system tends to regard consecutive saturation as a normal phenomenon, reducing the possibility of missed detection.

[0067] If the saturation frame count is less than or equal to a preset counting threshold, then step S305 is executed to determine that the current frame spectrum data is an abnormal saturation frame. This indicates that although the current frame is suspected of being saturated, the number of consecutively occurring saturation frames has not yet reached the counting threshold, so it is still considered an abnormal situation and discarded.

[0068] Conversely, if the saturation frame count exceeds a preset counting threshold, step S306 is executed to determine that the current frame spectrum data is a normal saturation frame. This indicates that the number of consecutive saturation frames has exceeded the counting threshold. At this point, it is believed that there is indeed a strong signal covering the entire frequency band in the channel, and this frame should be entered into the subsequent signal detection process as a normal frame, rather than simply being discarded.

[0069] The technical effects of this application will be further illustrated below through specific simulation experiments.

[0070] Simulation 1: Verification of the effect of abnormal saturation frame detection To assess the accuracy of the abnormal saturation frame detection in step S203, the following simulation experiment was conducted. Figure 4 A schematic diagram of the statistical maximum value sequence in Simulation 1 of this application is shown. Figure 4 In the graph, the blue line represents the raw spectral data within the broadband band, and the red dots represent the second maximum sequence, which can also be understood as the upper envelope of the spectral data. Figure 4 As shown, in the case of abnormal saturation frames, the overall spectral data is raised to near saturation levels, with a flat upper envelope and very small fluctuations. After normalizing the sequence, its standard deviation is calculated. When the standard deviation is less than the preset standard deviation threshold DVal, the frame is considered a possible abnormal saturation frame. Simultaneously, the saturation frame count F1 is accumulated. When the saturation frame count F1 is greater than the preset count threshold Num, it indicates that the saturation state has persisted for some time, and this saturation phenomenon is considered normal, allowing for normal subsequent detection procedures. When the saturation frame count F1 is less than the preset count threshold Num, it indicates that the frame is an occasional abnormal saturation frame, and no further detection procedures are performed; it is directly discarded. Through this mechanism, this application can effectively distinguish between occasional abnormal saturation and persistent normal saturation, avoiding false alarms caused by abnormal frames and preventing missed detections in real full-band strong signal scenarios.

[0071] Simulation 2: Verification of the piecewise adaptive gain effect The following comparative simulation experiments were conducted to assess the detection performance of the adaptive detection threshold generated in this application. Figures 5 to 7 The figures show the analysis results of Simulation 2. In each figure, the blue line represents the original spectral data within the broadband band, the red line represents the noise floor estimated using smoothing operations, and the green line represents the adaptive detection threshold determined based on the relative noise floor gain.

[0072] Figure 5 This illustrates the case where a preset relative noise floor gain value is used, and that value is relatively small. From Figure 5As can be clearly seen, in the portion of the frequency band where there is no signal, due to the excessively low detection threshold, the amplitude of a large number of pure noise frequencies exceeds the detection threshold (i.e., exceeds the threshold). These noise points that exceed the threshold will be misjudged as signals, resulting in a large number of false alarms. This indicates that setting the gain value too low will lead to a low detection threshold, which cannot effectively suppress false alarms caused by background noise.

[0073] Figure 6 This illustrates the case where a pre-set relative noise floor gain value is used, and that value is relatively large. From Figure 6 As can be observed, because the detection threshold was set too high, the amplitudes of the stronger signals that should have been detected failed to exceed the detection threshold, resulting in all these real signals being missed and causing a serious false alarm situation. This indicates that setting the gain value too high will lead to a high detection threshold, reducing the detection sensitivity and making it impossible to effectively capture real signals.

[0074] Figure 7 This illustrates the calculation of the relative noise floor gain value using the piecewise adaptive method of this application. From Figure 7 As can be observed, the detection threshold, represented by the green line, is appropriately lowered where there are signal peaks, allowing the real signal to reliably exceed the threshold and be detected; while it is appropriately raised where there is no signal, effectively suppressing noise spikes below the threshold. It is worth noting that the detection threshold exhibits abrupt changes at the boundaries of different frequency bands, which reflects the segmented gain calculation. Each segment independently determines a different relative noise floor gain value based on its own statistical characteristics, thus displaying a segmented variation in the threshold. From an overall perspective… Figure 7 All strong signals and some weak signals can reliably exceed the detection threshold, while almost no false alarms occur in the pure noise region, which significantly improves the accuracy of detection.

[0075] The above simulation experiments verified the following engineering practice conclusions: In actual engineering, the electromagnetic environment is complex and changeable, and the fluctuation pattern of the spectrum is difficult to predict. The pre-set fixed gain value (i.e., the engineering experience value) cannot encompass every possible spectrum form. Therefore, it is necessary to adaptively calculate the gain according to the spectrum characteristics of each segment within the frequency band, which is of great significance for improving the accuracy of signal detection. In addition, the rationality of the data segment division in this method directly affects the detection performance. Engineering experience shows that when the number of frequency points of each first data segment is set in the range of 1500 to 2500, better detection results can be obtained. At the same time, the practice of judging whether there is a signal in the segment based on the difference between the second difference and the first difference, and raising the threshold to D2+C using the above formula (2) in the no-signal segment, also plays an important role in reducing false alarms.

[0076] Next, combine Figure 8An exemplary description of an electronic device 800 provided in an embodiment of this application is given below. For example... Figure 8 As shown, the electronic device 800 in this application embodiment may include a processor 801, a memory 802, and a communication bus 803.

[0077] In specific embodiments, the processor 801 described above can be at least one of the following: Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), CPU, controller, microcontroller, and microprocessor. It is understood that for different devices, the electronic device used to implement the above processor function can also be other types, and this embodiment does not specifically limit it.

[0078] In this embodiment, the communication bus 803 is used to establish communication between the processor 801 and the memory 802; the memory 802 stores program instructions for frequency domain broadband signal detection; when the processor 801 executes the program instructions stored in the memory 802, it implements the combination of this application. Figures 1 to 3 The described method for detecting broadband signals in the frequency domain.

[0079] The above combination Figure 8 This document describes an electronic device for frequency-domain broadband signal detection that can be used to execute the present application. It should be understood that the device structure or architecture described herein is merely exemplary, and the implementation methods and entities of this application are not limited thereto, but can be modified without departing from the spirit of this application. It is understood that the description of the various embodiments in this disclosure emphasizes the differences between the various embodiments, while their similarities or corresponding parts can be referred to mutually. For the sake of brevity, this disclosure will not elaborate on each one.

[0080] Based on the foregoing description in conjunction with the accompanying drawings, those skilled in the art will understand that the embodiments of this application can also be implemented by software programs. Therefore, this application also provides a computer-readable storage medium. This computer-readable storage medium stores program instructions for frequency-domain broadband signal detection, which can be used to implement the embodiments of this application. Figures 1 to 3 The method for detecting broadband signals in the frequency domain is described.

[0081] It should be noted that although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart can be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0082] While numerous embodiments of this application have been shown and described herein, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and intent of this application. It should be understood that various alternatives to the embodiments of this application described herein may be employed in the practice of this application. The appended claims are intended to define the scope of protection of this application and therefore cover equivalents or alternatives within the scope of these claims.

Claims

1. A method for detecting broadband signals in the frequency domain, characterized in that, include: The noise floor is estimated by performing a frequency-by-frequency point estimate on the current frame spectrum data within the broadband bandwidth; The current frame spectrum data is equally divided into a first preset number of first data segments; For each first data segment, based on the statistical characteristics of the first data segment, the relative noise floor gain value corresponding to each frequency point within the first data segment is determined; the relative noise floor gain value is the amount by which the detection threshold of each frequency point is raised relative to its noise floor estimate. The relative noise floor gain value corresponding to each frequency point is added to its noise floor estimate to obtain the adaptive detection threshold for each frequency point. The presence or absence of a signal is determined point by point based on the adaptive detection threshold.

2. The method according to claim 1, characterized in that, After obtaining the frequency-point-by-frequency noise floor estimate and before equally dividing the current frame spectrum data into a first preset number of first data segments, the process further includes: The current frame spectrum data is equally divided into a second preset number of second data segments, and the maximum spectrum value in each second data segment is extracted to form a second maximum value sequence. The second maximum value sequence is normalized and its standard deviation is calculated; Based on the standard deviation, the preset standard deviation threshold, and the saturation frame count, determine whether the current frame spectrum data is an abnormal saturation frame; If the current frame spectrum data is an abnormally saturated frame, then discard the current frame spectrum data and perform noise floor estimation on the next frame spectrum data; If the current frame spectrum data is a normal saturated frame, then the subsequent equal division and determination of the relative noise floor gain value will continue to be performed on the current frame spectrum data.

3. The method according to claim 2, characterized in that, Based on the standard deviation, the preset standard deviation threshold, and the saturation frame count, determining whether the current frame spectrum data is an abnormally saturated frame includes: Determine whether the standard deviation is less than a preset standard deviation threshold; If the standard deviation is greater than or equal to the preset standard deviation threshold, then the current frame spectrum data is determined to be a normal frame; If the standard deviation is less than the preset standard deviation threshold, then the current frame spectrum data is determined to be a suspected saturated frame, and the saturated frame count is incremented. Determine whether the saturation frame count is greater than a preset counting threshold; If the saturation frame count is less than or equal to the preset count threshold, then the current frame spectrum data is determined to be an abnormal saturation frame. If the saturation frame count is greater than the preset count threshold, then the current frame spectrum data is determined to be a normal saturation frame.

4. The method according to claim 1, characterized in that, For each first data segment, based on the statistical characteristics of that first data segment, determine the relative noise floor gain value corresponding to each frequency point within that first data segment, including: The first data segment is equally divided into a third preset number of third data sub-segments, and the maximum spectral value in each third data sub-segment is recorded to obtain the first maximum value sequence of the first data segment. Select the minimum and maximum values ​​from the first maximum value sequence, and denote them as the first maximum value and the second maximum value, respectively; The noise floor estimates at the frequency points corresponding to the first maximum and the second maximum are obtained respectively and denoted as the first noise floor value and the second noise floor value. Based on the first maximum value, the second maximum value, the first noise floor value, and the second noise floor value, the relative noise floor gain value corresponding to each frequency point in the first data segment is determined.

5. The method according to claim 4, characterized in that, Based on the first maximum value, the second maximum value, the first noise floor value, and the second noise floor value, determine the relative noise floor gain value corresponding to each frequency point within the first data segment, including: Calculate the difference between the first maximum value and the first noise floor value, and record it as the first difference value; calculate the difference between the second maximum value and the second noise floor value, and record it as the second difference value; Based on the first difference, the second difference, and the preset judgment threshold, the relative noise floor gain value corresponding to each frequency point in the first data segment is determined.

6. The method according to claim 5, characterized in that, Based on the first difference, the second difference, and a preset judgment threshold, the relative noise floor gain value corresponding to each frequency point within the first data segment is determined, including: Calculate the difference between the second difference and the first difference, and record it as the target difference; Determine whether the target difference is greater than the preset judgment threshold; If the target difference is greater than the preset judgment threshold, then the relative noise floor gain value corresponding to each frequency point in the first data segment is determined to be between the first difference and the second difference; If the target difference is less than or equal to the preset judgment threshold, then the relative noise floor gain value corresponding to each frequency point in the first data segment is determined to be greater than the second difference.

7. The method according to claim 6, characterized in that, The formula for the relative noise floor gain value to be between the first difference and the second difference is expressed as follows: G = D1 + k(D2 - D1); Where G represents the relative noise floor gain value, D2 represents the second difference, D1 represents the first difference, and k represents the preset gain coefficient, and 0 <k<1。 8. The method according to claim 6, characterized in that, The formula for the relative noise floor gain value being greater than the second difference is expressed as follows: G = D² + C; Where G represents the relative noise floor gain value, D2 represents the second difference value, and C is a preset constant, and 0 <C<10。 9. An electronic device, comprising: processor; as well as A memory storing program instructions for frequency domain broadband signal detection, which, when executed by a processor, cause the frequency domain broadband signal detection method according to any one of claims 1-8 to be implemented.

10. A computer-readable storage medium having stored thereon program instructions for detecting a broadband signal in the frequency domain, which, when executed by a processor, implement the broadband signal detection method in the frequency domain according to any one of claims 1-8.

Images