Signal detection method for broadband spectrum data

Abstract

The invention belongs to the technical field of signal detection, and particularly relates to a signal detection method for broadband spectrum data. The method comprises the following steps: equally dividing to-be-processed broadband spectrum data into N sections of spectrum data, and obtaining white noise data corresponding to each section of spectrum data according to the power spectrum entropy of each section of spectrum data; calculating the standard deviation of the white noise data corresponding to each section of spectrum data, and determining a discriminant value of comparison of adjacent frequency points according to the standard deviation of the white noise data; obtaining background noise data corresponding to each section of spectrum data through adjacent value point-by-point comparison, and taking the background noise data as a signal detection threshold; and comparing the broadband spectrum data to be processed with the signal detection threshold, and determining that the frequency points higher than the signal detection threshold are signals, thereby realizing signal detection. According to the signal detection method provided by the invention, non-interference noise data in the frequency band can be extracted according to the power spectrum entropy value of the frequency spectrum, the adjacent value discrimination value is calculated, and signal detection of the broadband frequency spectrum is realized through a segmented adjacent value comparison mode.

Description

Technical Field

[0001] The present invention belongs to the technical field of signal detection, and in particular relates to a signal detection method for broadband spectrum data. Background Art

[0002] Signal detection is an indispensable part of electromagnetic environment monitoring. Signal detection searches, detects and obtains basic parameter information of electromagnetic signals propagating in electromagnetic space. Signal detection is usually achieved by comparing the amplitude of each point in the spectrum with the threshold value. The frequency points with amplitudes higher than the threshold are considered as signals, and those lower than the threshold are considered as noise. However, there are many ways to calculate threshold values ​​in current engineering, which often require manual setting of threshold parameters and rely on engineering experience. The threshold setting is not flexible enough and it is difficult to cope with various forms of signals and changing electromagnetic background environments. In addition, the threshold of current broadband signal detection is usually corrected on the basis of the noise floor. First, the noise floor of the entire broadband signal is estimated, and then a fixed value is added to the noise floor to obtain the decision threshold. The signal spectrum power value is compared with the threshold to determine whether the signal exists. The calculation methods of the noise floor usually include the average method, the 20% noise threshold method, the frequency shift mean, the time-frequency mean, the morphological filtering, the neighbor value comparison and other methods. The above schemes often require manual selection or setting of parameters, rely on engineering experience, and lack flexibility.

[0003] The prior art proposes a method for extracting interference signals from a broadband spectrum (patent number: ZL201510542046.0). By comparing neighbor values, background noise data is extracted and signal detection is achieved. However, this patent has the following disadvantages: (1) It is necessary to manually select a sub-spectrum without abnormal interference to obtain the standard deviation of the spectrum noise, which lacks flexibility; (2) In the case of a wide frequency band, the spectrum background noise may fluctuate greatly. The use of a single neighbor value discrimination value for the entire spectrum data cannot adapt to broadband spectrum data with large background noise fluctuations; (3) The neighbor value comparison process is complicated, and the amount of filtering calculation introduced is large. Summary of the invention

[0004] In response to the above technical problems, the present invention provides a signal detection method for broadband spectrum data, which can extract interference-free noise data in the frequency band according to the power spectrum entropy value of the spectrum and calculate the adjacent value discrimination value, and realize signal detection of the broadband spectrum by segmented adjacent value comparison.

[0005] The present invention is achieved through the following technical solutions:

[0006] A signal detection method for broadband spectrum data, the method comprising:

[0007] The broadband spectrum data to be processed is evenly divided into N segments of spectrum data, and the white noise data corresponding to each segment of spectrum data is obtained according to the power spectrum entropy of each segment of spectrum data;

[0008] Calculate the standard deviation of the white noise data corresponding to each segment of spectrum data, and determine the discrimination value of the adjacent frequency point comparison according to the standard deviation of the white noise data;

[0009] Based on the discriminant value, the background noise data corresponding to each segment of the spectrum data is obtained by point-by-point comparison of adjacent values, and the background noise data of the entire broadband spectrum data to be processed is used as a signal detection threshold;

[0010] The broadband spectrum data to be processed is compared with the signal detection threshold, and the frequency points higher than the signal detection threshold are confirmed as signal existence, thereby realizing signal detection.

[0011] Furthermore, the method specifically comprises the following steps:

[0012] (1) The broadband spectrum data P to be processed is segmented from low to high frequency and divided into N segments of spectrum data, P = [P1, P2, P3, ..., P n ,...,P N ];

[0013] (2) Calculate the power spectrum entropy of each segment of spectrum data, determine the signal occupancy according to the power spectrum entropy of each segment of spectrum data, filter out signal interference, and obtain white noise data corresponding to each segment of spectrum data;

[0014] (3) Calculate the standard deviation of the white noise data corresponding to each segment of spectrum data;

[0015] (4) determining the discrimination value of the adjacent frequency point comparison based on the standard deviation of the white noise data corresponding to each segment of the spectrum data;

[0016] (5) Obtain all frequency bands P by comparing neighboring values ​​point by point n The corresponding background noise data NP n , n∈N; then the background noise data NP of the entire broadband spectrum data to be processed is obtained, and the background noise data NP is used as the threshold value for signal detection; the obtained threshold value is compared with the original broadband spectrum data P to be processed, and the frequency points higher than the threshold value are confirmed as the existence of signals, thereby realizing signal detection.

[0017] Furthermore, step (2) is specifically as follows:

[0018] For frequency band P n Spectrum data, set the sliding window length of spectral entropy to w, then the spectral entropy value h in the i-th window i Defined as:

[0019] Among them, P i,k =|z k | 2 / E i is the probability density, is the total energy in the ith window, z k Represents the signal amplitude value of the kth point in the i-th window;

[0020] According to the set sliding window length, sliding step size and spectral entropy value h i , get the whole frequency band P n The spectral entropy sequence H = [h1,h2,h3,...,h i ,...h M ] T , where M is the number of segments obtained after sliding window processing;

[0021] Generate l groups of Gaussian white noise sequences of length M, and calculate the spectral entropy value of each group of Gaussian white noise sequences, so as to obtain the spectral entropy sequence H0 = [h 01 ,h 02 ,h 03 ,...,h 0k ,...h 0l ] T ; Among them, Gaussian white noise can be directly obtained through the formula of Gaussian white noise, which is an existing and commonly used noise model;

[0022] Definition b i Indicates whether there is signal occupancy and interference. Where h0 and δ0 are the mean and standard deviation of the spectral entropy sequence H0, respectively.

[0023] Calculate b by using the power spectrum entropy value of each segment of spectrum data and the white noise spectrum entropy sequence i , b i =0 means the signal is not occupied, extract b i =0 band to get band P n White noise data S without abnormal interference n .

[0024] Furthermore, in step (3), the frequency band P is calculated n White noise data S n The standard deviation σ n , the formula is:

[0025]

[0026] Among them, S n,k Represents white noise data S n The kth amplitude in S nis the white noise data S n The amplitude average value, X represents the white noise data S n The total number of sample points in .

[0027] Furthermore, in step (4), the discriminant value delta for comparing adjacent frequency points is determined based on the standard deviation of each segment of spectrum noise data:

[0028] delta=2σ n ;

[0029] Among them, σ n For frequency band P n White noise data S n The standard deviation of .

[0030] Furthermore, step (5) specifically includes:

[0031] Obtain spectrum data P n The average value P n mean As the first background noise point for comparison;

[0032] Let P n mean With P n The first point P n [1] For comparison, if |P n mean -P n [1]|>delta and P n [1]>P n mean , indicating that P n [1] is the signal, then let the noise data NP n [1] = P n mean When P n mean -P n [1]|>When delta does not hold, it means P n [1] is the noise, let NP n [1] = P n [1];

[0033] When k>2, all frequency points are traversed, when |P k [k]-NP k [k-1]|>delta holds, if P n [k]-NP n [k-1]≥0, then NP n [k]=NP n [k-1], otherwise NP n [k] = P n [k]; when Pk [k]-NP k [k-1]|>delta does not hold, NP n [k] = P n [k]; where P n [k] represents P n The kth data point in NP n [k] represents the background noise data NP n The kth data point in ;

[0034] Get frequency band P n Background noise data NP n ;

[0035] The background noise data NP of the entire broadband spectrum data to be processed is [NP1, NP2, NP3, ..., NP n ,...,NP N ].

[0036] Beneficial technical effects of the present invention:

[0037] The signal detection method for broadband spectrum data provided by the present invention can solve the problem of spectrum noise extraction without artificially selecting the noise frequency band, and extracts interference-free white noise frequency band data within the frequency band according to the spectral entropy value of the spectrum.

[0038] The signal detection method for broadband spectrum data provided by the present invention can solve the problem of large dynamic range of broadband spectrum noise; the broadband spectrum noise background noise fluctuation increases, and segmented noise extraction is performed to calculate the neighbor value comparison discrimination values ​​in different segments. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a flow chart of a signal detection method for broadband spectrum data in an embodiment of the present invention;

[0040] Figure 2 Flow chart of point-by-point comparison of neighboring values ​​in an embodiment of the present invention;

[0041] Figure 3 This is a signal detection effect diagram of a detection frequency band of 120MHz to 400MHz in an embodiment of the present invention;

[0042] Figure 4 This is a signal detection effect diagram of the detection frequency band of 850MHz to 1GHz in an embodiment of the present invention;

[0043] Figure 5 This is a diagram showing the effect of signal detection in a detection frequency band of 120 MHz to 1 GHz in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0045] On the contrary, the present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention as defined by the claims. Further, in order to make the public have a better understanding of the present invention, some specific details are described in detail in the detailed description of the present invention below. Those skilled in the art can fully understand the present invention without the description of these details.

[0046] The embodiment of the present invention provides a signal detection method for broadband spectrum data, such as Figure 1 As shown, the method includes:

[0047] The broadband spectrum data to be processed is evenly divided into N segments of spectrum data, and the white noise data corresponding to each segment of spectrum data is obtained according to the power spectrum entropy of each segment of spectrum data;

[0048] Calculate the standard deviation of the white noise data corresponding to each segment of spectrum data, and determine the discrimination value of the adjacent frequency point comparison according to the standard deviation of the white noise data;

[0049] Based on the adjacent value comparison difference discriminant value, the background noise data corresponding to each segment of spectrum data is obtained by point-by-point comparison of adjacent values, and the background noise data is used as a signal detection threshold;

[0050] The broadband spectrum data to be processed is compared with the signal detection threshold, and the frequency points higher than the signal detection threshold are confirmed as signal existence, thereby realizing signal detection.

[0051] In this embodiment, the method specifically includes the following steps:

[0052] (1) The broadband spectrum data P to be processed is segmented according to the frequency, and is evenly divided into N segments of spectrum data, P = [P1, P2, P3, ..., P n ,...,P N ]; wherein, dividing the broadband spectrum data to be processed into N segments can avoid the problem of excessive fluctuation of background noise within the frequency band;

[0053] (2) Calculate the power spectrum entropy of each segment of spectrum data, determine the signal occupancy according to the power spectrum entropy of each segment of spectrum data, filter out signal interference, and obtain white noise data corresponding to each segment of spectrum data;

[0054] (3) Calculate the standard deviation of the white noise data corresponding to each segment of spectrum data;

[0055] (4) determining the discrimination value of the adjacent frequency point comparison based on the standard deviation of the white noise data corresponding to each segment of the spectrum data;

[0056] (5) Obtain all frequency bands P by comparing neighboring values ​​point by point n The corresponding background noise data NP n , n∈N; then the background noise data NP of the entire broadband spectrum data to be processed is obtained, and the background noise data NP is used as the threshold value for signal detection; the obtained threshold value is compared with the original broadband spectrum data P to be processed, and the frequency points higher than the threshold value are confirmed as the existence of signals, thereby realizing signal detection.

[0057] In this embodiment, step (2) is specifically as follows:

[0058] For frequency band P n Spectrum data, set the sliding window length of spectral entropy to w, then the spectral entropy value h in the i-th window i Defined as:

[0059] Among them, P i,k =|z k | 2 / E i is the probability density, is the total energy in the ith window, z k Represents the signal amplitude value of the kth point in the window;

[0060] According to the set sliding window length, sliding step length and the above spectral entropy value h t , get the whole frequency band P n The spectral entropy sequence is expressed as H = [h1,h2,h3,...,h i ,...h M ] T ; Where M is the number of segments obtained after sliding window processing; because the sliding spectral entropy can effectively reflect the degree of disorder of the target sequence, when its spectral entropy value is higher than a threshold, it can be regarded as a white noise spectrum without abnormal interference; because the spectral entropy value of Gaussian white noise is only related to the sequence length and has nothing to do with the noise intensity, the spectral entropy value of Gaussian white noise of the same length is generated and calculated as the detection threshold for whether there is interference.

[0061] Generate l groups of Gaussian white noise sequences of length M, and calculate the spectral entropy value of each group of Gaussian white noise sequences, so as to obtain the spectral entropy sequence H0 = [h 01 ,h 02 ,h 03 ,...,h 0k ,...h 0l ] T ; Among them, Gaussian white noise can be directly obtained through the formula of Gaussian white noise, which is an existing and commonly used noise model.

[0062] Definition b i Indicates whether there is signal occupancy and interference. Where h0 and δ0 are the mean and standard deviation of the spectral entropy sequence H0, respectively.

[0063] Calculate b by using the power spectrum entropy value of each segment of spectrum data and the white noise spectrum entropy sequence i , b i =0 means the signal is not occupied, extract b i =0 band to get band P n White noise data S without abnormal interference n .

[0064] In step (3) of this embodiment, the frequency band P is calculated. n White noise data S n The standard deviation σ n , the formula is:

[0065]

[0066] Among them, S n,k Represents white noise data S n The kth amplitude in S n is the white noise data S n The amplitude average value, X represents the white noise data S n The total number of sample points in .

[0067] In step (4) of this embodiment, the discrimination value delta for comparing adjacent frequency points is determined based on the standard deviation of each segment of spectrum noise data:

[0068] delta=2σ n ;

[0069] Among them, σ n For frequency band P n White noise data S n The standard deviation of .

[0070] In this step, since the spectrum noise data samples conform to the normal distribution, the dynamic range of the spectrum noise is about 4σ n , so the discrimination value is set to half of the dynamic range delta = 2σ n .

[0071] like Figure 2 As shown, in this embodiment, step (5) specifically includes:

[0072] Obtain spectrum data P n The average value P n meanAs the first background noise point for comparison; With P n The first point P n [1] For comparison, if and Description n [1] is the signal, then let the noise data when If it is not true, it means P n [1] is the noise, let NP n [1] = P n [1];

[0073] When k>2, all frequency points are traversed, when |P k [k]-NP k [k-1]|>delta holds, if P n [k]-NP n [k-1]≥0, then NP n [k]=NP n [k-1], otherwise NP n [k] = P n [k]; when |P k [k]-NP k [k-1]|>delta does not hold, NP n [k] = P n [k]; where P n [k] represents P n The kth data point in NP n [k] represents the background noise data NP n The kth data point in ;

[0074] Finally, we get the frequency band P n Background noise data NP n ;

[0075] The background noise data NP of the entire broadband spectrum data to be processed is [NP1, NP2, NP3, ..., NP n ,...,NP N ].

[0076] Figure 3-Figure 5 The signal detection effect diagrams are respectively for the detection frequency bands of 120MHz to 400MHz, 850MHz to 1GHz, and 120MHz to 1GHz. The signal detection method for broadband spectrum data provided by the present invention can accurately extract the interference-free frequency band in the spectrum and calculate the white noise standard deviation to obtain the adjacent value comparison discrimination value, and no longer requires manual noise selection; the adjacent value comparison process is optimized, and by calculating the segmented adjacent value discrimination value, it has better adaptability to broadband spectrum data.

[0077] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A signal detection method for broadband spectrum data, characterized in that: The method comprises: The broadband spectrum data to be processed is evenly divided into N segments of spectrum data, and the white noise data corresponding to each segment of spectrum data is obtained according to the power spectrum entropy of each segment of spectrum data; Calculate the standard deviation of the white noise data corresponding to each segment of spectrum data, and determine the discrimination value of the adjacent frequency point comparison according to the standard deviation of the white noise data; Based on the discriminant value, the background noise data corresponding to each segment of the spectrum data is obtained by point-by-point comparison of adjacent values, and the background noise data of the entire broadband spectrum data to be processed is used as a signal detection threshold; The broadband spectrum data to be processed is compared with the signal detection threshold, and the frequency points higher than the signal detection threshold are confirmed as signal existence, thereby realizing signal detection.

2. The signal detection method of broadband spectrum data according to claim 1, characterized in that: The method specifically comprises the following steps: (1) The broadband spectrum data P to be processed is segmented from low to high frequency and divided into N segments of spectrum data, P = [P1, P2, P3, ..., P n ,...,P N ]; (2) Calculate the power spectrum entropy of each segment of spectrum data, determine the signal occupancy according to the power spectrum entropy of each segment of spectrum data, filter out signal interference, and obtain white noise data corresponding to each segment of spectrum data; (3) Calculate the standard deviation of the white noise data corresponding to each segment of spectrum data; (4) determining the discrimination value of the adjacent frequency point comparison based on the standard deviation of the white noise data corresponding to each segment of the spectrum data; (5) Obtain all frequency bands P by comparing neighboring values ​​point by point n The corresponding background noise data NP n , n∈N; then the background noise data NP of the entire broadband spectrum data to be processed is obtained, and the background noise data NP is used as the threshold value for signal detection; the obtained threshold value is compared with the original broadband spectrum data P to be processed, and the frequency points higher than the threshold value are confirmed as the existence of signals, thereby realizing signal detection.

3. The signal detection method of broadband spectrum data according to claim 2, characterized in that: Step (2) is specifically as follows: For frequency band P n Spectrum data, set the sliding window length of spectral entropy to w, then the spectral entropy value h in the i-th window i Defined as: Among them, P i,k =|z k | 2 / E i is the probability density, is the total energy in the ith window, z k Represents the signal amplitude value of the kth point in the i-th window; According to the set sliding window length, sliding step size and spectral entropy value h i , get the whole frequency band P n The spectral entropy sequence H = [h1,h2,h3,...,h i ,...h M ] T , where M is the number of segments obtained after sliding window processing; Generate l groups of Gaussian white noise sequences of length M, and calculate the spectral entropy value of each group of Gaussian white noise sequences, so as to obtain the spectral entropy sequence H0 = [h 01 ,h 02 ,h 03 ,...,h 0k ,...h 0l ] T ; Definition b i Indicates whether there is signal occupancy and interference. Where h0 and δ0 are the mean and standard deviation of the spectral entropy sequence H0, respectively. Calculate b by using the power spectrum entropy value of each segment of spectrum data and the white noise spectrum entropy sequence i , b i =0 means the signal is not occupied, extract b i =0 band to obtain band P n White noise data S without abnormal interference n .

4. The signal detection method of broadband spectrum data according to claim 3, characterized in that: In step (3), calculate the frequency band P n White noise data S n The standard deviation σ n , the formula is: Among them, S n,k Represents white noise data S n The k-th amplitude in is the white noise data S n The amplitude average value, X represents the white noise data S n The total number of sample points in .

5. A signal detection method for broadband spectrum data according to claim 4, characterized in that: In step (4), the discriminant value delta for comparing adjacent frequency points is determined based on the standard deviation of the noise data of each segment of the spectrum: delta=2σ n ; Among them, σ n For frequency band P n White noise data S n The standard deviation of .

6. A signal detection method for broadband spectrum data according to claim 5, characterized in that: Step (5) specifically includes: Obtain spectrum data P n The average As the first background noise point for comparison; make With P n The first point P n [1] For comparison, if and Description n [1] is the signal, then let the noise data when If it is not true, it means P n [1] is the noise, let NP n [1] = P n [1]; When k>2, all frequency points are traversed, when |P k [k]-NP k [k-1]|>delta holds, if P n [k]-NP n [k-1]≥0, then NP n [k]=NP n [k-1], otherwise NP n [k] = P n [k]; when P k [k]-NP k [k-1]|>delta does not hold, NP n [k] = P n [k]; where P n [k] represents P n The kth data point in NP n [k] represents the background noise data NP n The kth data point in ; Get frequency band P n Background noise data NP n ; The background noise data NP of the entire broadband spectrum data to be processed is [NP1, NP2, NP3, ..., NP n ,...,NP N ].

Images