A Wideband Spectrum Sensing Method Based on Wavelet Transform

Abstract

This invention discloses a broadband spectrum sensing method based on wavelet transform, including background noise extraction and signal frequency domain information extraction. This method implements the background noise of the scanned spectrum, estimates the changes in the background noise, and extracts rough frequency domain information of the signal, including the center frequency and frequency range, from the scanned spectrum or the fixed-frequency detection spectrum. This prepares for subsequent more accurate parameter measurement and identification, and can detect shorter burst signals and extract frequency domain information from signals with overlapping spectra but different bandwidths and amplitudes. The multi-scale wavelet transform can better adapt to the edge shapes of different signal spectra, resulting in more accurate extraction.

Description

Technical Field

[0001] The present invention relates to the field of signal processing technology, and in particular to a broadband spectrum sensing method based on wavelet transform. Background Art

[0002] The following problems exist in existing frequency-domain spectrum sensing:

[0003] 1. Because the noise characteristics of the receiver vary in different frequency bands, and the noise in different frequency bands in space also varies, the noise floor of the scan spectrum is not flat, making it impossible to estimate the noise floor change.

[0004] 2. The duration of the burst signal is relatively short. The spectrum scan can only capture a burst within a few consecutive frames. Therefore, it is impossible to average the spectrum scan over a long period of time to reduce the impact of noise.

[0005] 3. The average of the scanned spectra of several consecutive frames shows the overlap of signals with different bandwidths and amplitudes. In this case, the frequency domain information of different signals cannot be extracted. Summary of the Invention

[0006] The purpose of the present invention is to solve the problems of existing frequency domain spectrum sensing and propose a broadband spectrum sensing method based on wavelet transform.

[0007] A broadband spectrum sensing method based on wavelet transform, including background noise extraction and signal frequency domain information extraction;

[0008] The background noise extraction steps are as follows:

[0009] S11: Scan the spectrum and divide it into several parts according to the number of points;

[0010] S12: Take the median value of each part to form a median curve;

[0011] S13: All troughs in the median curve form a trough curve;

[0012] S14: median filtering is performed on the trough curve to remove the mutation points;

[0013] S15: perform linear interpolation to finally obtain a scanning spectrum noise plane change curve;

[0014] The signal frequency domain information extraction steps are:

[0015] S21: average the scanned spectrum for several frames, and determine the average number of frames according to the situation;

[0016] S22: The spectrum is smoothed by median filtering. The number of median filter points is selected. Increasing the number of points improves the smoothing effect and also filters out narrowband signals.

[0017] S23: noise plane extraction smoothed spectrum;

[0018] S24: Subtract the smoothed spectrum from the noise plane to eliminate the influence of noise plane changes, and the noise plane is at the 0 plane;

[0019] S25: If the spectrum of the noise plane change is less than 5Db, it will be treated as zero;

[0020] S26: performing multi-scale Haar wavelet transform on the spectrum processed by the above steps;

[0021] S27: Multiply the Haar wavelet transform results of each scale to obtain the minimum wavelet transform curve;

[0022] S28: Find peaks and valleys on the wavelet transform curve, set the minimum peak distance and minimum peak height, and obtain the steepest peak;

[0023] S29: Match the peaks and troughs to obtain the frequency range of each signal;

[0024] S210: Extract the frequency range of the narrowband signal, subtract the scanned spectrum from the noise plane, and detect the spectrum range greater than a certain threshold. If the interval between two adjacent spectrum ranges is less than a certain Hz, they are merged into one spectrum range and regarded as the same signal.

[0025] Furthermore, in a broadband spectrum sensing method based on wavelet transform, the signal frequency domain information extraction step is performed after the background noise extraction S15, and the scanning spectrum noise plane change curve is used for frequency domain information extraction.

[0026] Furthermore, a broadband spectrum sensing method based on wavelet transform is provided, wherein the multi-scale Haar wavelet transform has an odd number of scales.

[0027] Furthermore, a broadband spectrum sensing method based on wavelet transform is provided. In the wavelet transform curve, as the scale becomes smaller, the edge perception becomes finer, and peaks appear when there are slight fluctuations; as the scale becomes larger, the edge perception becomes coarser, and obvious peaks appear when there are steeper changes.

[0028] The beneficial effects of the present invention are: extracting rough frequency domain information of the signal, including the center frequency and frequency range, on the scanning spectrum or fixed-frequency receiving spectrum, preparing for subsequent more accurate parameter measurement and identification, etc. The signal-to-noise ratio can be as low as 5dB, and shorter burst signals can be detected. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is the noise extraction flow chart.

[0030] Figure 2 It is the extraction of signal frequency domain information.

[0031] Figure 3 It is the median curve graph.

[0032] Figure 4 is the spectrum diagram with the noise plane variation eliminated.

[0033] Figure 5 This is a spectrum diagram of the noise-eliminated plane variation with zeroing less than 5dB.

[0034] Figure 6 is a multi-scale Haar wavelet transform image.

[0035] Figure 7 It is the multiplication graph of Haar wavelet transform results.

[0036] Figure 8 It is a wavelet transform graph with peaks and valleys.

[0037] Figure 9 It is a frequency range extraction diagram of narrowband signal.

[0038] Figure 10 It is matching the falling edge and rising edge diagram. DETAILED DESCRIPTION

[0039] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, specific embodiments of the present invention are now described with reference to the accompanying drawings.

[0040] The example analysis describes a broadband spectrum sensing method based on wavelet transform, including background noise extraction and signal frequency domain information extraction.

[0041] As attached Figure 2 As shown: the scanned spectrum is divided into several parts according to a certain number of points, and the median of each part is taken to form a curve called the median curve.

[0042] Find all the troughs in the median curve to form a trough curve;

[0043] Perform median filtering on the trough curve to remove the mutation points, and then perform linear interpolation to obtain the final scanning spectrum noise plane change curve.

[0044] The simulation experiment parameters are as follows:

[0045] The sampling rate is 204.8MHz, the number of FFT points is 8192, the frequency resolution is 25kHz, and the signal-to-noise ratio is about 5dB to 15dB.

[0046] Signal parameter modulation mode, center frequency, and bandwidth are shown in the following table:

[0047] Modulation method Center frequency (MHz) Bandwidth (MHz) 16QAM -30 3.2 16QAM -21 3.2 QPSK -20 6.4 QPSK 0 3.2 2FSK 5 3.2 BPSK 15 3.2 CW 20 N / A CW 21 N / A BPSK 21.5 0.2 QPSK 30 12.8

[0048] Table 1 Signal parameter table

[0049] The changing noise plane simulation calculation formula is as follows:

[0050] -sin(2*pi*0.01*(0:1:8191)' / 100)*5

[0051] As attached Figure 5 As shown, the smoothed spectrum is subtracted from the noise plane to obtain a spectrum that eliminates the changes in the noise plane.

[0052] As attached Figure 6 As shown, the spectrum of the eliminated noise floor changes is less than 5dB with zero placement.

[0053] As attached Figure 7 As shown, the processed spectrum is subjected to multi-scale Haar wavelet transform with scales from 20 to 28 and step 2.

[0054] As attached Figure 8 As shown, the Haar wavelet transform results of each scale are multiplied.

[0055] As attached Figure 9 As shown, the peaks and valleys are found on the wavelet transform curve, with a minimum peak distance of 40 and a minimum peak height of 2000.

[0056] As attached Figure 10 As shown, the narrowband signal frequency range is extracted, the threshold is 11dB, and the interval between two adjacent spectrum ranges is 125kHz (5 frequency points).

[0057] The modulation mode, center frequency, and bandwidth used in the calculation are shown in the following table:

[0058]

[0059]

[0060] Table 2 Signal parameters in calculation

[0061] The third signal was not detected because the signal-to-noise ratio was less than 5dB. The large errors in center frequency and bandwidth detection were because edge detection was performed at the position with the largest change, rather than at the position 3dB below the maximum value of the signal power spectrum. The signal-to-noise ratio increased to approximately 7dB to 20dB.

[0062] The results of modulation mode, center frequency and bandwidth are shown in the following table:

[0063] Modulation method Center frequency (MHz) Bandwidth (MHz) 16QAM -30.1125 3.375 16QAM -21.35 3.9 QPSK -20.15 6.3 QPSK 0.0125 3.875 2FSK 4.9875 3.175 BPSK 15.025 3.05 CW 20.0125 0.075 CW 21.0125 0.075 BPSK 21.5125 0.175 QPSK 30.2 15.15

[0064] Table 3 Signal parameter results

[0065] This solution uses a broadband spectrum sensing method based on wavelet transform to achieve the noise floor of the scanned spectrum, estimate the changes in the noise floor, and extract the rough frequency domain information of the signal, including the center frequency and frequency range, from the scanned spectrum or fixed-frequency received spectrum. This prepares for subsequent more accurate parameter measurement and identification. The signal-to-noise ratio can be as low as 5dB, and it can detect shorter burst signals and extract the frequency domain information of signals with overlapping spectra but different bandwidths and amplitudes. The multi-scale wavelet transform can better adapt to the edge shapes of different signal spectra, making the extraction more accurate.

[0066] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A broadband spectrum sensing method based on wavelet transform, characterized in that: Including background noise extraction and signal frequency domain information extraction; The background noise extraction steps are as follows: S11: Scan the spectrum and divide it into several parts according to the number of points; S12: Take the median value of each part to form a median curve; S13: All troughs in the median curve form a trough curve; S14: median filtering is performed on the trough curve to remove the mutation points; S15: perform linear interpolation to finally obtain a scanning spectrum noise plane change curve; The signal frequency domain information extraction steps are as follows: S21: average the scanned spectrum for several frames, and determine the average number of frames according to the situation; S22: The spectrum is smoothed by median filtering. The number of median filter points is selected. Increasing the number of points improves the smoothing effect and also filters out narrowband signals. S23: noise plane extraction smoothed spectrum; S24: Subtract the smoothed spectrum from the noise plane to eliminate the influence of noise plane changes, and the noise plane is at the 0 plane; S25: If the spectrum of the noise plane change is less than 5Db, it will be treated as zero; S26: performing multi-scale Haar wavelet transform on the spectrum processed by the above steps; S27: Multiply the Haar wavelet transform results of each scale to obtain the minimum wavelet transform curve; S28: Find peaks and valleys on the wavelet transform curve, set the minimum peak distance and minimum peak height, and obtain the steepest peak; S29: Match the peaks and troughs to obtain the frequency range of each signal; S210: Extract the frequency range of the narrowband signal, subtract the scanned spectrum from the noise plane, and detect the spectrum range greater than a certain threshold. If the interval between two adjacent spectrum ranges is less than a certain Hz, they are merged into one spectrum range and regarded as the same signal.

2. The method for broadband spectrum sensing based on wavelet transform according to claim 1, characterized in that: The signal frequency domain information extraction step is performed after the background noise extraction S15, and the scanned spectrum noise plane change curve is used for frequency domain information extraction.

3. The broadband spectrum sensing method based on wavelet transform according to claim 1, characterized in that: The multi-scale Haar wavelet transform has an odd number of scales.

4. The method for broadband spectrum sensing based on wavelet transform according to claim 1, wherein: The wavelet transform curve has a finer edge perception as the scale becomes smaller, and peaks appear when there are slight fluctuations. As the scale increases, the edge perception becomes coarser, and obvious peaks appear when the changes are steeper.

Images