Wideband Fast Spectrum Detection and Sensing Method, Device, Equipment and Storage Medium

Abstract

The present invention relates to a broadband fast spectrum detection and sensing method, device, equipment and storage medium; belonging to the technical field of radio frequency signal detection and spectrum sensing, the method includes: adopting a sub-channel parallel acquisition method to collect broadband signals with sparsity characteristics to obtain sample data; performing data preprocessing on the sample data including smoothing, splicing and FFT to obtain a spectrogram; extracting local maximum values from the spectrogram to obtain a first set of sensed frequency points; using an improved FCME algorithm to perform frequency domain energy detection on the first set of sensed frequency points to obtain a second set of sensed frequency points composed of signals to be sensed; clustering the second set of sensed frequency points to form multiple clusters, and each cluster corresponds to a detected signal. The present invention realizes broadband fast spectrum detection and sensing, improves the efficiency of spectrum sensing, and realizes synchronous extraction of signal bandwidth and power characteristics.

Description

Technical Field

[0001] The present invention relates to the technical field of radio frequency signal detection and spectrum sensing, and particularly to a broadband fast spectrum detection and sensing method, device, equipment and storage medium. Background Art

[0002] In traditional wireless communication systems, the use of the spectrum is often static and restricted. With the increasing demand for frequency usage, the key to improving spectrum utilization lies in optimizing the spectrum reuse mechanism. In cognitive radio technology, which is crucial for spectrum reuse, the performance of spectrum detection and sensing has an important impact on the performance of the entire system. The wider the bandwidth of spectrum sensing, the more spectrum holes can be found. The faster the spectrum sensing speed, the smaller the delay for each data transmission, and the higher the spectrum utilization rate.

[0003] As one of the currently effective single-node algorithms, the energy detection method is widely used in spectrum sensing and signal detection. It does not require prior knowledge of the signal, discriminates the threshold through the energy characteristics of the signal to detect interference, and has the characteristics of simple implementation and low complexity.

[0004] The energy detection algorithm is divided into time-domain detection and frequency-domain detection. Time-domain detection is a binary detection that can only obtain the presence or absence of a signal; while frequency-domain detection can obtain characteristic parameters such as the frequency points and energy of the signal, providing necessary information for subsequent processing. Classic frequency-domain energy detection algorithms include the consecutive mean excision (CME) algorithm and the forward CME (FCME) algorithm. Both algorithms obtain the threshold through iteration for detection. FCME is an improved algorithm of CME. Compared with the CME algorithm, FCME has better detection performance for partial band interference.

[0005] However, for broadband signals with sparsity characteristics, due to the wider bandwidth, the computational complexity of the traditional FCME algorithm is large and cannot meet the requirements of fast sensing; moreover, the traditional peak detection method cannot meet the needs in terms of detection flexibility and missed detection probability; in addition, the FCME algorithm is very sensitive to parameter settings. If the parameters are set unreasonably during the iteration process, it may lead to the situation that the threshold increase speed cannot cover the growth of normal samples, resulting in fewer and fewer samples added to the clean set, and the changes in the mean and standard deviation becoming smaller and smaller. Eventually, the iteration stops prematurely. Especially in the later stage of iteration, since the set becomes larger and larger, the influence of new samples added to the set on the statistical characteristics of the set becomes less and less obvious, resulting in an increase in the false alarm probability.

[0006] Therefore, there is a need for a method that can meet broadband fast spectrum detection and sensing, improve the efficiency of spectrum sensing, and realize the synchronous extraction of signal bandwidth and power characteristics. Summary of the Invention

[0007] In view of the above analysis, the present invention aims to disclose a broadband fast spectrum detection and sensing method, device, equipment and storage medium, which performs broadband fast spectrum detection and sensing based on improved FCME and clustering analysis, improves the efficiency of spectrum sensing, and realizes synchronous extraction of signal bandwidth and power characteristics.

[0008] On the one hand, the present invention discloses a broadband fast spectrum detection and sensing method, including:

[0009] Step S1: Collect sample data of broadband signals with sparsity characteristics by using a sub-channel parallel acquisition method;

[0010] Step S2: Perform data preprocessing on the sample data, including smoothing, splicing and FFT, to obtain a spectrogram;

[0011] Step S3: Extract local maximum values from the spectrogram to obtain a first set of perceived frequency points;

[0012] Step S4: Use the improved FCME algorithm to perform frequency-domain energy detection on the first set of perceived frequency points to obtain a second set of perceived frequency points composed of signals to be sensed;

[0013] In the iterative process of the improved FCME algorithm, determine the adaptive dynamic threshold of the current iteration according to the amplitude mean and standard deviation of the noise concentration frequency points determined in the previous iteration, and find out the noise frequency points from the signal concentration determined in the previous iteration according to the adaptive dynamic threshold and move them to the noise concentration;

[0014] Step S5: Cluster the second set of perceived frequency points to form multiple clusters, and each cluster corresponds to a detected signal.

[0015] On the other hand, the present invention also discloses a device using the above-mentioned broadband fast spectrum detection and sensing method, including:

[0016] A sample data acquisition module, configured to collect sample data of broadband signals with sparsity characteristics by using a sub-channel parallel acquisition method;

[0017] A spectrogram generation module, configured to perform data preprocessing on the sample data, including smoothing, splicing and FFT, to obtain a spectrogram;

[0018] A local maximum value extraction module, configured to extract local maximum values from the spectrogram to obtain a first set of perceived frequency points;

[0019] An FCME module, configured to use the improved FCME algorithm to perform frequency-domain energy detection on the first set of perceived frequency points to obtain a second set of perceived frequency points composed of signals to be sensed;

[0020] During the iterative process of the improved FCME algorithm, the adaptive dynamic threshold for the current iteration is determined based on the amplitude mean and standard deviation of the noise concentration frequency points determined in the previous iteration. According to the adaptive dynamic threshold, the noise frequency points are found again from the signal set determined in the previous iteration and moved to the noise concentration;

[0021] A clustering module is used to cluster the second set of sensed frequency points to form multiple clusters, and each cluster corresponds to a detected signal.

[0022] On the other hand, the present invention also discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the broadband fast spectrum detection and sensing method as described above is implemented.

[0023] On the other hand, the present invention also discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the broadband fast spectrum detection and sensing method as described above is implemented.

[0024] One of the beneficial effects that can be achieved by the present invention is as follows:

[0025] The broadband fast spectrum detection and sensing method, device, equipment, and storage medium disclosed by the present invention reduce the computational amount of each iteration in spectrum detection, accelerate the convergence speed, and reduce the detection time required; it has a lower false alarm probability;

[0026] In the present invention, aiming at the sparsity characteristics of signals in the spectrum, by adding a process of finding local maxima before calculating the initial threshold in the FCME algorithm, a large number of frequency points that are not helpful for estimating the noise level are removed. Then, the frequency points retained after finding the local maxima are used to replace all the spectra, reducing the total number of frequency points participating in the calculation of the mean, standard deviation, and set partitioning of the FCME, reducing the computational amount of each iteration of the FCME algorithm, and accelerating the execution speed of the algorithm while retaining its function of filtering a large amount of noise, realizing fast sensing in the broadband spectrum range. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The drawings are only for the purpose of showing specific embodiments and are not considered as limiting the present invention. Throughout the drawings, the same reference numerals represent the same components;

[0028] Figure 1 It is a flowchart of the broadband fast spectrum detection and sensing method in an embodiment of the present invention;

[0029] Figure 2 It is a schematic connection diagram of the composition of the broadband fast spectrum detection and sensing device in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0030] The preferred embodiments of the present invention will be specifically described below in conjunction with the accompanying drawings, where the drawings form a part of this application and are used together with the embodiments of the present invention to explain the principles of the present invention.

[0031] Embodiment 1

[0032] An embodiment of the present invention discloses a broadband fast spectrum detection and sensing method, as Figure 1 shown, including:

[0033] Step S1: Adopt a sub-channel parallel acquisition method to collect broadband signals with sparsity characteristics to obtain sample data;

[0034] Step S2: Perform data preprocessing on the sample data, including smoothing, splicing, and FFT, to obtain a spectrogram;

[0035] Step S3: Extract local maximum values from the spectrogram to obtain a first set of sensed frequency points;

[0036] Step S4: Use an improved FCME algorithm to perform frequency-domain energy detection on the first set of sensed frequency points to obtain a second set of sensed frequency points composed of signals to be sensed;

[0037] During the iterative process of the improved FCME algorithm, determine the adaptive dynamic threshold for the current iteration according to the amplitude mean and standard deviation of the noise concentration frequency points determined in the previous iteration, and find the noise frequency points from the signal concentration determined in the previous iteration according to the adaptive dynamic threshold and move them to the noise concentration;

[0038] Step S5: Cluster the second set of sensed frequency points to form multiple clusters, and each cluster corresponds to a detected signal.

[0039] Specifically, in step S1, an adjustable narrowband sensing method is adopted for sub-channel parallel acquisition. The adjustable narrowband sensing method means dividing the broadband channel into multiple sub-channels, adding a frequency band selection module in the receiver, changing the center frequency of the receiver in real time, and sequentially collecting wireless signals in these sub-channels to quickly realize the detection of multiple sub-channels in sequence; since the above acquisition consumes less time, it is considered that the acquisition of all sub-channels is carried out at the same moment.

[0040] Specifically, in step S2, use gnuradio software to perform data preprocessing on the collected samples, including smoothing, splicing, FFT, etc., to obtain a spectrogram.

[0041] Specifically, within the broadband frequency detection range of 75Mhz - 6Ghz, due to the sparsity of signals, most frequency bands should have no signals, and there is no single signal with such a large bandwidth. Therefore, not only are there no signals at some frequency points with small amplitudes, but also a considerable number of the frequency points corresponding to the local maxima of the spectrum have no signals. Thus, in step S3, local maxima are extracted from the spectrogram to obtain the first set of perceived frequency points; the first set of perceived frequency points is used to replace all frequency points and subsequent operations are continued. The purpose of this is that after finding the local maxima, the total number of frequency points will be greatly reduced, and the computational amount of each iteration will also decrease accordingly. The time complexity of finding the local maxima is , and the time complexity of finding the mean is also , and after finding the local maxima, the time complexity will become or less. Therefore, in the case of a large number of iterations, the overall delay of the algorithm after finding the local maxima will decrease.

[0042] The method for extracting local maxima in this embodiment can be carried out by using existing functions for detecting local maxima in signals provided by MATLAB or Python, or by using methods such as FFT and peak detection algorithms, and no specific limitation is made here.

[0043] Specifically, the improved FCME algorithm in step S4 includes:

[0044] Step S401: Given two sets S0 and S1, where S0 represents the noise set that does not need to be perceived, and S1 represents the signal set that needs to be perceived;

[0045] Step S402: After sorting all the frequency points in the first set of perceived frequency points according to the amplitudes of the frequency points, according to the sorting result and a set division ratio, the part with low amplitudes is classified into the noise set S0, and the part with high amplitudes is classified into the signal set S1;

[0046] Preferably, the set division ratio is 20%; 20% of the frequency points with smaller amplitudes in the first set of perceived frequency points are classified into the noise set , and the other part is classified into the signal set S1.

[0047] Step S403: Calculate the adaptive dynamic threshold according to the mean and standard deviation of the noise set ; compare the elements in the signal set S1 with the adaptive dynamic threshold, and those greater than the threshold are classified into the signal set S1, and those less than the threshold are classified into the noise set ;

[0048] Step S404: Repeatedly iterate and execute step S403. In each iteration process, use the noise set determined in the previous iteration Mean and standard deviation update the adaptive dynamic threshold; using the updated adaptive dynamic threshold, update the set and S1, move the frequency points in the signal set S1 that are less than the threshold to the noise set and set a sliding window in the noise set S0, delete the same number of frequency points with the smallest amplitude as the moved frequency points in the noise set S0 to ensure that the size of the noise set S0 remains unchanged;

[0049] Step S405: When the iteration reaches that the noise set S0 and the signal set S1 no longer update, the elements in the signal set S1 at this time are the frequency points to be sensed.

[0050] Specifically, in step S403, the adaptive dynamic threshold :

[0051] ;

[0052] Among them, is the mean value of the noise set in the current iteration , is the standard deviation of the noise set ; is the adaptive parameter.

[0053] The adaptive parameter is updated according to the number of frequency points newly added to the noise set in each iteration and finally reaches a stable value. Among them, if the number of new samples added to the noise set in the (n + 1)-th iteration is less than that in the n-th iteration, then increase value, otherwise, reduce value. After multiple iterations, will reach a stable value, and at the same time, the noise set can also be stably updated until most of the noise is filtered.

[0054] Specifically, the initial value of the adaptive parameter is determined according to the presented by the full noise when there is no signal in the frequency band;

[0055] The calculation formula in the iteration process is as follows:

[0056] ;

[0057] Among them, is the parameter value in the th iteration; is the number of samples added to the set in the th iteration, is a factor to avoid the denominator being zero;

[0058] Meanwhile, to ensure that after filtering out most of the noise, the signal will not be filtered out due to too few samples added to the set becoming too small, resulting in a large adaptive parameter is constrained;

[0059] The constraint conditions are:

[0060] ;

[0061] are the minimum and maximum values of the adaptive parameter determined according to the distribution of the noise.

[0062] Specifically, in step S404, by setting a sliding window for the noise set to discard the long-term impact of early data. On the one hand, discarding early data can make the threshold increase faster, accelerating the convergence speed. On the other hand, it can avoid the influence of new samples on its statistical characteristics becoming smaller due to the large set in the later stage of iteration, premature termination of iteration, reduce the false alarm probability, and improve the accuracy.

[0063] Through the specific improvements in steps S403 and S404, the traditional FCME algorithm is overcome, which is very sensitive to the parameter setting. If the parameter setting is unreasonable during the iteration process, it may lead to the threshold increase speed not being able to cover the growth of normal samples, resulting in fewer and fewer samples added to the clean set, and the changes in the mean and standard deviation becoming smaller and smaller, and finally the iteration stops prematurely. Especially in the later stage of iteration, as the set becomes larger and larger, the influence of new samples added to the set on the statistical characteristics of the set becomes less and less obvious, resulting in an increase in the false alarm probability.

[0064] Therefore, compared with the traditional FCME algorithm, the improved FCME algorithm in this embodiment reduces the computational amount of each iteration, accelerates the convergence speed, and reduces the detection time required; has a lower false alarm probability; can adaptively determine the parameter value without relying on empirical setting.

[0065] Specifically, in step S5, the second set of perceived frequency points is clustered by HDBSCAN (Hierarchical DBSCAN) to form multiple clusters, and each cluster is a signal. At the same time, its peak power and 3db bandwidth are calculated.

[0066] More specifically, in step S5, the process of HDBSCAN clustering includes:

[0067] Step S501, initialization. The two-dimensional data composed of the frequency values and signal intensities of each frequency point in the second set of perceived frequency points is used as the input of the clustering analysis, and all frequency points are marked as "unvisited" for initialization;

[0068] Since the two-dimensional data composed of the frequency points and signal strengths output by the improved FCME algorithm is inherently suitable as the input for cluster analysis, data preprocessing can be omitted, and only all data points need to be marked as "unvisited" for initialization.

[0069] Step S502, transform the metric space. According to the two-dimensional data points composed of the frequency value and signal strength of each frequency point, calculate the core distance of the data points, and construct a mutual reachability distance matrix.

[0070] Specifically, in step S502, it includes

[0071] 1) Calculate the core distance of the two-dimensional data of each frequency point in the second perceived frequency point set.

[0072] The core distance is the distance from a certain frequency point to its min_cluster_size-th nearest neighbor; the min_cluster_size is a given parameter; by setting the size of min_cluster_size, the minimum bandwidth of the detected signal is limited.

[0073] 2) Construct a mutual reachability distance matrix.

[0074] For any two points a and b, their mutual reachability distance is defined as:

[0075] max{core distance(a), core distance(b), Euclidean distance(a, b)}.

[0076] Under the detection conditions of a large monitoring frequency band range and a small spectral resolution, setting the parameter min_cluster_size to 1 ensures that signals with a very small bandwidth in the order of dozens of kHz can also be detected instead of being treated as outliers.

[0077] Step S503, construct a hierarchical clustering tree. Based on the mutual reachability distance matrix, generate a minimum spanning tree MST, which represents the hierarchical connection relationship between data points; convert the minimum spanning tree MST into a hierarchical clustering tree by gradually merging the points or clusters with the closest distances to form a hierarchical structure.

[0078] Step S504, generate a condensed tree. Prune the hierarchical clustering tree to remove cluster branches, and only retain the cluster structure that meets the density and scale conditions to generate a condensed tree.

[0079] Specifically, according to the min_cluster_size parameter, prune the hierarchical clustering tree, remove the cluster branches with a size less than min_cluster_size, and generate a condensed tree. The condensed tree only retains the cluster structure that meets the density and scale conditions, simplifying subsequent analysis.

[0080] Step S505: Extract stable clusters, traverse the condensed tree to calculate the stability of each potential cluster; select the optimal cluster partition by maximizing stability, and merge overlapping or similar sub-clusters;

[0081] Step S506: Assign final labels, partition the data points into the corresponding stable clusters, and mark the points that are not assigned to any cluster as separate clusters and output the clustering results.

[0082] In summary, the broadband fast spectrum detection and sensing method disclosed in this embodiment reduces the computational complexity of each iteration in spectrum detection, speeds up the convergence rate, and reduces the detection time required; it has a lower false alarm probability;

[0083] In this embodiment, aiming at the sparsity characteristics of signals in the spectrum, by adding a process of finding local maxima before calculating the initial threshold in the traditional FCME algorithm, a large number of frequency points that are not helpful for estimating the noise level are removed, and then the frequency points retained after finding the local maxima are used to replace the entire spectrum, reducing the total number of frequency points participating in the calculation of the mean, standard deviation, and set partition of FCME, reducing the computational complexity of each iteration of the FCME algorithm, and speeding up the execution speed of the algorithm on the premise of retaining its function of filtering a large amount of noise, realizing fast sensing in the broadband spectrum range.

[0084] Embodiment 2

[0085] An embodiment of the present invention discloses a device adopting the broadband fast spectrum detection and sensing method as described in Embodiment 1, as Figure 2 shown, including:

[0086] A sample data acquisition module, configured to collect sample data of broadband signals with sparsity characteristics by adopting a sub-channel parallel acquisition method;

[0087] A spectrogram generation module, configured to perform data preprocessing including smoothing, splicing, and FFT on the sample data to obtain a spectrogram;

[0088] A local maximum extraction module, configured to extract local maxima from the spectrogram to obtain a first set of perceived frequency points;

[0089] An FCME module, configured to perform frequency-domain energy detection on the first set of perceived frequency points by adopting an improved FCME algorithm to obtain a second set of perceived frequency points composed of signals to be sensed;

[0090] During the iterative process of the improved FCME algorithm, an adaptive dynamic threshold for the current iteration is determined according to the amplitude mean and standard deviation of the frequency points in the noise concentration determined in the previous iteration, and noise frequency points are found from the signal concentration determined in the previous iteration according to the adaptive dynamic threshold and moved to the noise concentration;

[0091] A clustering module, configured to cluster the second set of sensed frequency points to form a plurality of clusters, where each cluster corresponds to a detected signal.

[0092] The specific technical details and corresponding beneficial effects in this embodiment are the same as those disclosed in Embodiment 1. Please refer to them specifically and will not be elaborated one by one here.

[0093] Embodiment 3

[0094] An embodiment of the present invention discloses an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor.

[0095] In one example, the above-mentioned processor may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0096] The memory may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical, or other physical / tangible memory storage device. Therefore, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to execute the broadband fast spectrum detection and sensing method according to Embodiment 1.

[0097] The processor runs a computer program corresponding to the executable program code by reading the executable program code stored in the memory, so as to implement the broadband fast spectrum detection and sensing method in Embodiment 1.

[0098] In one example, a communication interface and a bus may also be included.

[0099] Wherein, the memory, the processor, and the communication interface are connected through the bus and complete communication with each other.

[0100] The communication interface is mainly used to implement communication between various modules, devices, units, and / or devices in the embodiments of the present application. The input device and / or the output device may also be accessed through the communication interface.

[0101] A bus includes hardware, software, or both, and components of an electronic device are coupled to each other. By way of example and not limitation, a bus can include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-E) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses or a combination of two or more of these. Where appropriate, a bus can include one or more buses. Although embodiments of the present application describe and illustrate specific buses, the present application contemplates any suitable bus or interconnect.

[0102] Embodiment 4

[0103] An embodiment of the present invention discloses a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the broadband fast spectrum detection and sensing method in Embodiment 1 can be implemented and the same technical effects can be achieved. To avoid repetition, it will not be elaborated here. Among them, the above computer-readable storage medium can include a non-transitory computer-readable storage medium, such as a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disc, etc., which is not limited herein.

[0104] As mentioned above, the above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention.

[0105] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A broadband fast spectrum detection and sensing method, characterized in that Including: Step S1: Use the sub-channel parallel acquisition method to collect broadband signals with sparsity characteristics to obtain sample data; Step S2: Perform data preprocessing on the sample data, including smoothing, splicing, and FFT, to obtain a spectrogram; Step S3: Extract local maximum values from the spectrogram to obtain a first set of perceived frequency points; Step S4: Use the improved FCME algorithm to perform frequency-domain energy detection on the first set of perceived frequency points to obtain a second set of perceived frequency points composed of signals to be perceived; During the iterative process of the improved FCME algorithm, determine the adaptive dynamic threshold for the current iteration according to the amplitude mean and standard deviation of the frequency points in the noise set determined not to be perceived in the previous iteration. According to the adaptive dynamic threshold, find the noise frequency points from the signal set determined to be perceived in the previous iteration and move them to the noise set; Repeat the iteration until the noise set and the signal set no longer update. At this time, the elements in the signal set are the frequency points to be perceived; Step S5: Cluster the second set of perceived frequency points to form multiple clusters, and each cluster corresponds to a detected signal.

2. The broadband fast spectrum detection and sensing method according to claim 1, characterized in that The improved FCME algorithm in step S4 includes: Step S401: Given two sets and , where represents the noise set that does not need to be sensed, and S1 represents the signal set that needs to be sensed; Step S402: After sorting all the frequency points in the first set of perceived frequency points according to the amplitude of the frequency points, according to the sorting result and the set division ratio, the part with a low amplitude is classified into the noise set , and the part with a high amplitude is classified into the signal set ; Step S403. Calculate an adaptive dynamic threshold according to the mean value and standard deviation of the noise set ; Compare the elements in the signal set with the adaptive dynamic threshold, and those greater than the threshold are classified into the signal set , and those less than the threshold are classified into the noise set ; Step S404: Repeatedly and iteratively execute step S403. In each iteration, update the adaptive dynamic threshold using the mean and standard deviation of the noise set determined in the previous iteration; use the updated adaptive dynamic threshold to update the sets and ; move the frequency points in the signal set that are less than the threshold to the noise set , and set a sliding window in the noise set ; delete the same number of frequency points with the smallest amplitudes as the moved frequency points in the noise set to ensure that the size of the noise set remains unchanged; ​ Step S405: When iterating to the set of noise sets and the signal set no longer update, at this time, the elements in the signal set S1 are the frequency points to be sensed.

3. The broadband fast spectrum detection and sensing method according to claim 2, characterized in that Adaptive dynamic threshold Calculated by the following formula: ； Among them, is the noise set for the current iteration mean value of , is the standard deviation of the noise set ; is an adaptive parameter, which is updated according to the number of frequency points newly added to the noise set in each iteration and finally reaches a stable value.

4. The broadband fast spectrum detection and sensing method according to claim 3, characterized in that Adaptive parameter The initial value is determined according to the Gaussian distribution presented by the full noise when there is no signal in the frequency band; The calculation formula during the iterative process is as follows: ； Among them, is the value of the parameter for the th iteration; is the number of samples added to the th iteration set , and is a factor to avoid a zero denominator; Meanwhile, to ensure that after filtering out most of the noise, the signal is not filtered out due to the too small number of samples added to the set resulting in becoming very large, the adaptive parameter is constrained; The constraint condition is: ； The adaptive parameter determined according to the distribution of the noise The minimum value and the maximum value 5. The broadband fast spectrum detection and sensing method according to claim 1, characterized in that In step S5, use HDBSCAN clustering for the second set of perceived frequency points to form multiple clusters, and each cluster is a signal. At the same time, calculate its peak power and 3dB bandwidth.

6. The broadband fast spectrum detection and sensing method according to claim 5, characterized in that In step S5, the process of HDBSCAN clustering includes: Step S501: Initialize, use the two-dimensional data composed of the frequency values and signal intensities of each frequency point in the second set of perceived frequency points as the input for clustering analysis, and mark all frequency points as "unvisited" for initialization; Step S502: Convert the metric space, calculate the core distance of the data points according to the two-dimensional data points composed of the frequency values and signal intensities of each frequency point, and construct a mutually reachable distance matrix; Step S503: Construct a hierarchical clustering tree. Based on the mutually reachable distance matrix, generate a minimum spanning tree MST to represent the hierarchical connection relationship between data points; convert the minimum spanning tree MST into a hierarchical clustering tree, and form a hierarchical structure by gradually merging the points or clusters with the closest distances; Step S504: Generate a condensed tree, prune the hierarchical clustering tree to remove cluster branches, and only retain the cluster structure that meets the density and scale conditions to generate a condensed tree; Step S505: Extract stable clusters, traverse the condensed tree to calculate the stability of each potential cluster; select the optimal cluster partition by maximizing stability, and merge overlapping or similar sub-clusters; Step S506: Assign final labels, divide the data points into the corresponding stable clusters, and mark the points not assigned to any cluster as separate clusters and output the clustering results.

7. The broadband fast spectrum detection and sensing method according to claim 6, characterized in that In step S502, it includes 1) Calculate the core distance of the two-dimensional data of each frequency point in the second sensing frequency point set; The core distance is the distance from a certain frequency point to its min_cluster_size-th nearest neighbor; the min_cluster_size is a given parameter; the minimum bandwidth of the signal to be detected is defined by setting the size of min_cluster_size; 2) Construct a mutual reachability distance matrix; For any two points a and b, their mutual reachability distance is defined as: max{core distance(a), core distance(b), Euclidean distance(a, b)}.

8. An apparatus using the broadband fast spectrum detection and sensing method according to any one of claims 1-7, characterized in that, It includes: A sample data acquisition module, which is used to collect sample data of broadband signals with sparsity characteristics by adopting a sub-channel parallel acquisition method; A spectrogram generation module, which is used to perform data preprocessing including smoothing, splicing, and FFT on the sample data to obtain a spectrogram; A local maximum extraction module, which is used to extract local maximum values from the spectrogram to obtain a first sensing frequency point set; An FCME module, which is used to perform frequency domain energy detection on the first sensing frequency point set by adopting an improved FCME algorithm to obtain a second sensing frequency point set composed of signals to be sensed; During the iteration process of the improved FCME algorithm, determine the adaptive dynamic threshold of the current iteration according to the amplitude mean and standard deviation of the frequency points in the noise set determined not to be sensed in the previous iteration, and move the noise frequency points found from the signal set determined to be sensed in the previous iteration to the noise set according to the adaptive dynamic threshold; Repeat the iteration until the noise set and the signal set are no longer updated. At this time, the elements in the signal set are the frequency points to be sensed; A clustering module, which is used to cluster the second sensing frequency point set to form multiple clusters, and each cluster corresponds to a detected signal.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the broadband fast spectrum detection and sensing method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the broadband fast spectrum detection and sensing method according to any one of claims 1 to 7.

Images