High-precision center frequency detection method for space electromagnetic spectrum monitoring

Abstract

The invention belongs to the field of spectrum monitoring and sensing, and particularly relates to a high-precision center frequency detection method for space electromagnetic spectrum monitoring. The method comprises the following steps: performing noise floor compensation on original AD sampling data, smoothing a curve through a Sgolay filter, performing a multi-parameter fusion peak searching algorithm to search a plurality of peak points, performing half-height searching after a peak value is searched, finding two points on the left and right sides of the relative height difference of the peak value, performing high-resolution Fourier transform in two intervals, and obtaining a high-resolution Fourier transform result; after refining the frequency, carrying out a linear interpolation fitting algorithm to obtain two end points where half-peak horizontal lines intersect, calculating to obtain a central value, namely a central frequency point, and then extrapolating 10% of peak height and width by one time, and then integrating power to obtain 99% of occupied bandwidth. The algorithm can overcome adverse effects caused by spectrum peak flattop shapes, peak top noise fluctuations and the like, accurately finds out the center frequency point and the occupied bandwidth, and has the advantages of high accuracy, strong real-time performance, good flexibility, low spectrum resource occupancy rate and the like compared with the prior art.

Description

Technical Field

[0001] The present invention belongs to the field of spectrum monitoring and sensing, and particularly relates to a high-precision center frequency detection method for spatial electromagnetic spectrum monitoring. Background Art

[0002] With the development of radio technology, the demand for spectrum resources has also increased significantly. There is a phenomenon that the communication demand for a small number of frequency bands increases year by year, while the utilization rate of a large number of other frequency bands is not high, resulting in low utilization rate of a large amount of spectrum resources and waste of spectrum resources. Therefore, monitoring the electromagnetic spectrum in space is an important link in the utilization of electromagnetic spectrum resources in space.

[0003] The center frequency detection algorithm uses a peak detection algorithm in the spectrum monitoring process. The main purpose is to effectively and quickly find peaks such as maxima and maximum values from the spectrum data as the center frequency points of the spectrum peaks. However, the traditional peak detection algorithm generally only considers the rise and fall situations around local peaks and cannot estimate the specific shape of the entire peak and the influence of baseline noise, etc. It often cannot accurately monitor the center frequency point in the case of flat-topped peaks, peak-top noise fluctuations, etc. Considering that most of the spatial electromagnetic spectrum is symmetric in shape, including some high-instantaneous-bandwidth flat-topped signals, in order to accurately find the center frequency point and occupied bandwidth, it is necessary to overcome the adverse effects brought by the flat-topped shape of the spectrum peak and peak-top noise fluctuations. At the same time, in order to process high-instantaneous-bandwidth, a sampling rate of 1 Gsps is adopted, and the frequency accuracy of 8192-point FFT only reaches 122 kHz. To accurately calculate the center frequency and bandwidth, further processing is required. Summary of the Invention

[0004] Based on the above problems, the present invention proposes a high-precision center frequency detection algorithm for spatial electromagnetic spectrum monitoring. This algorithm is a complete set of algorithm processes for finely detecting the center frequency and bandwidth of symmetric power spectrum peaks. The present invention can overcome the adverse effects brought by the flat-topped shape of the spectrum peak and peak-top noise fluctuations, and accurately find the center frequency point and occupied bandwidth.

[0005] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] A high-precision center frequency detection method for spatial electromagnetic spectrum monitoring includes the following processes:

[0007] Step 1: Perform noise floor compensation on the original AD sampling data and smooth the curve through a Sgolay filter;

[0008] Step 2: Use a multi-parameter fusion peak search algorithm to find multiple peak points for the filtered data;

[0009] Step 3: After finding the peak, take half of the height to make a semi-peak horizontal line, and find the intersection points of the semi-peak horizontal line and both sides of the peak;

[0010] Step 4: Pass through the semi-peak horizontal line, and find two closest spectral points above and below the semi-peak horizontal line at each intersection point on both sides of the peak, obtain two left and right intervals, and perform high-resolution Fourier transform on the two intervals to obtain refined frequencies;

[0011] Step 5: Perform linear interpolation fitting on the refined frequencies of the two intervals respectively, obtain two end points intersecting with the semi-peak horizontal line, and calculate to obtain the central value, which is the central frequency point.

[0012] Further, the specific process of Step 2 is as follows:

[0013] Set multiple parameters:

[0014] (1) Minimum peak height: Limit the minimum peak height above the noise floor, and preselect the peaks whose power exceeds the set threshold;

[0015] (2) Minimum relative peak height difference: Set the minimum relative height difference, and eliminate adjacent peaks whose relative height difference from the main peak is lower than the minimum relative height difference;

[0016] (3) Minimum peak spacing: Set the minimum peak spacing, filter out the sidelobe signals of sharp multi-peak signals, and find the main peak;

[0017] (4) Minimum and maximum peak widths: Specify the maximum and minimum peak widths, specify the bandwidth signals in a specific range, and capture the signals of interest in the acquisition;

[0018] Find multiple peak points that simultaneously meet the above four parameters.

[0019] The present invention has the following advantages compared with the prior art:

[0020] (1) Accuracy: Through the combination of various signal processing technologies, it can monitor the signals in the target frequency band with high precision, can quickly and accurately monitor the central frequency point and in-band power of the signals, can achieve high-precision detection of the central frequency, and effectively improve the accuracy of spectrum monitoring.

[0021] (2) Strong adaptability: It is applicable to different types of electromagnetic signals and complex spatial electromagnetic environments, and has strong adaptability.

[0022] (3) Real-time performance: The algorithm has high calculation efficiency and can meet the real-time requirements of spatial electromagnetic spectrum monitoring.

[0023] (4) Reliability: Through multi-parameter fusion and optimization algorithms, the reliability of the detection results is improved, and the probability of false detection is reduced.

[0024] (5) Large instantaneous bandwidth: By controlling the down-conversion local oscillator signal of the RF front end, the system can monitor the spectrum in the frequency range of 2 - 20 GHz, with a wide monitored spectrum bandwidth range, realizing the function of large instantaneous bandwidth. Description of the Drawings

[0025] Figure 1 It is a flow chart of the high-precision center frequency detection algorithm of the present invention.

[0026] Figure 2 It is a schematic diagram of the multi-parameter fusion peak-seeking algorithm of the present invention.

[0027] Figure 3 It is an architecture diagram of the hardware system of the present invention.

[0028] Figure 4 It is a spectrum of the high-precision center frequency detection method of the present invention. Detailed Embodiments

[0029] The present invention will be further explained below with reference to the drawings.

[0030] As Figure 1 shown, a high-precision center frequency detection method for spatial electromagnetic spectrum monitoring includes the following processes:

[0031] Step 1: The original AD sampling data is used as input after noise floor compensation and smoothed by a Sgolay filter to remove random noise and maintain the shape;

[0032] The Sgolay filter is a classic smoothing filter. It smooths the signal by polynomial fitting in a sliding window rather than simply weighted averaging. The waveform of the input signal spectrum has more noise under the influence of noise, and direct peak seeking is prone to misjudgment. After inputting it into the Sgolay filter, the filter calculates the smoothed value by fitting a polynomial in a certain window width around each data point, well maintaining the shape of signal spikes and valleys while smoothing the signal, so that the signal collision characteristics are retained while filtering out the signal noise, providing a good basis for subsequent peak seeking.

[0033] Step 2: Use the multi-parameter fusion peak-seeking algorithm to find multiple peak points for the filtered data; multiple parameters for peak detection can be configured;

[0034] The present invention adopts a multi-parameter fusion peak-seeking algorithm to fuse multiple peak parameters and perform parameter matching and screening on the found peak points, and can quickly find the peak points of interest. Traditional peak detection algorithms rely solely on single changes such as left and right extreme values and slopes for peak seeking, resulting in a large number of useless peak points being found in the presence of noise and multiple peaks, making it difficult to screen. When the multi-parameter fusion peak-seeking algorithm performs peak seeking, there are multiple parameter matches for the peak as follows:

[0035] (1) Minimum peak height: Limiting the minimum peak height above the noise floor helps to eliminate stray jump peaks near the noise floor and only preselect peaks with power exceeding a certain threshold.

[0036] (2) Minimum relative peak height difference: The relative height difference refers to the degree of prominence measured by the height of the peak itself and the relative position of adjacent peaks. Setting the minimum relative height difference can effectively filter out the small protrusions on the shoulders of the main peak without affecting the judgment of individual peaks. For example, Figure 2 the relative height difference prominence of the 3rd peak on the right is greater than that of the 1st peak on the left, so the signal of the 3rd independent peak can be successfully extracted while eliminating the fluctuation of the 1st peak.

[0037] (3) Minimum peak spacing: The peak spacing refers to the distance between peaks. Specifying the minimum peak spacing can help filter out the sidelobe signals of sharp multi-peak signals and successfully find the main peak.

[0038] (4) Minimum and maximum peak widths: Specifying the maximum and minimum peak widths can specify a signal with a specific range of bandwidths, making the capture and acquisition of the signal of interest more accurate.

[0039] Step 3: After finding the peak, take half of its height as the half-height horizontal line and find two points on each side about half of the relative peak height difference;

[0040] The half-height peak search is to solve the problem of spectral peaks with non-prominent peaks, such as flat top peaks, gentle BPSK modulation peaks, or adjacent BOC modulation split peaks. Their center frequencies are not at the peak points found. The intersection point of the center line of the symmetric graph and the peak should be used as the center frequency peak point. In fact, the center symmetric point can be obtained at any position with good symmetry. Generally, taking the half of the peak height as the symmetric point gives good results. The steps are to take half of the peak height as the horizontal line after finding the peak, and use the intersection points with both sides of the peak as the reference to obtain the symmetric center frequency.

[0041] Step 4: Through the half-height horizontal line, find the two nearest spectral points above and below the half-height horizontal line at the intersection points on both sides of the peak, obtain the left and right two intervals, and perform high-resolution Fourier transform on the two intervals to obtain the refined frequency.

[0042] High-resolution Fourier transform makes it possible to accurately determine the center frequency. To monitor the acquisition of large instantaneous bandwidth signals (satellite communication signals greater than 200 MHz) and maintain real-time performance, it is necessary to quickly perform spectrum peak monitoring based on a sampling rate of up to 1 Gsps. The algorithm uses an 8192-point FFT transform to obtain approximately 122 KHz per point while maintaining real-time monitoring. Through the semi-peak horizontal line, two spectral points above and below the semi-high line can be found on both sides of the peak. Performing a high-resolution Fourier transform with a 1 KHz step in these two intervals can accurately determine two points with the same horizontal height to the KHz level. The Fourier transform of signal sampling is as follows,

[0043]

[0044] where Ω = ΩT, T is the sampling interval, x[n] = x[nT], representing the nth sampling of the signal x[t]. The high-resolution Fourier transform of sampling can be directly calculated at the frequency points of interest. The following gives an effective algorithm for embedded devices.

[0045] Using Euler's identity, it can be divided into two parts: real and imaginary numbers:

[0046]

[0047] Through Chebyshev polynomials, starting from sinΩ and cosΩ, effective recursive operations can be performed.

[0048] cos(nΩ) = T n (cosΩ)

[0049] sin(nΩ) = sinΩU n-1 (cosΩ)

[0050] where, it is known that T 0 = 1, T 1 = cosΩ, U 0 = 1 and U 1 = 2cosΩ. The following formula can be used to calculate each T n and U n

[0051] T n+1 (cosΩ) = 2(cosΩ)T n (cosΩ) - T n-1 (cosΩ) (13)

[0052] U n+1 (cosΩ) = 2(cosΩ)U n (cosΩ) - U n-1 (cosΩ) (14)

[0053] Then, only by calculating cosΩ and sinΩ can the high-resolution Fourier transform be calculated within a small range, thus greatly improving the frequency calculation accuracy.

[0054] Step 5: Perform linear interpolation fitting on the refined frequencies of the two intervals respectively, obtain the two endpoints that intersect with the semi-peak horizontal line, calculate the central value to obtain the central frequency point, and then extrapolate the 10% peak height width by one time and integrate the power to obtain the 99% occupied bandwidth.

[0055] Perform the linear interpolation fitting algorithm to further improve the mathematical accuracy. Use the slope of a straight line to fit the upper and lower points at the intersection of the semi-peak horizontal line and the accurate frequency points of the high-precision Fourier transform to calculate the intersection points. After calculating the two intersection points, solve the intermediate frequency value between the two points to obtain the accurate central frequency. Under the premise of real-time performance, the central frequency accuracy less than 1KHz can be obtained.

[0056] As Figure 3 shown, the hardware system of the present invention mainly consists of a radio frequency front-end unit, a control unit, a signal processing unit, and a storage unit. The four units work together. The control unit controls the radio frequency front-end unit to collect signals in different frequency bands through instructions. The collected data is transmitted to the signal processing unit to calculate the signal center frequency point and in-band power. The original collected data and the data processed by the signal processing unit can be stored in the storage unit.

[0057] Radio frequency front-end unit: The radio frequency front-end unit is responsible for collecting signals in different frequency bands and mainly consists of a receiving antenna, a down-converter, and an analog-to-digital converter. The radio frequency analog signal is received through the receiving antenna, the high-frequency signal is down-converted to a lower-frequency signal through the down-converter, and the analog signal is converted into a digital signal through the analog-to-digital converter.

[0058] Control unit: The control unit is mainly responsible for the task management and instruction issuance of electromagnetic spectrum monitoring.

[0059] Signal processing unit: The signal processing unit is mainly responsible for processing and analyzing the collected data and calculating the signal center frequency point and in-band power.

[0060] Storage unit: The storage unit is responsible for storing AD data, spectrum data, and the processing and analysis results of the signal processing unit.

[0061] The following describes an embodiment of the present invention with reference to the accompanying drawings.

[0062] The collected target signal is a bandwidth signal with a central frequency point of 11.7GHz and a bandwidth of 240MHz. Figure 4 For the high-precision central frequency point detection algorithm to process and calculate the central frequency point result. Through Figure 4It can be seen that through Sgolay filtering and multi-parameter fusion peak searching, the peak points within the target bandwidth range can be found. After calculating the half-height peak value, it is found that the half-height peak line is in the middle of two spectral points, and the frequency difference between the spectral points is 122 KHZ, with a large error. Through high-resolution Fourier transform and slope fitting difference algorithm, there can be an intersection between the half-height peak line and the fitting curve, and the frequency value of the intersection point is the half-height peak frequency value.

[0063] After calculating the left half-height peak frequency point as 11.570061 GHz and the right half-height peak frequency point as 11.829935 GHz through the center frequency point detection algorithm, the center frequency can be calculated as 11.699998 GHz, with an error of 2 KHz.

[0064] The content not described in detail in this invention book belongs to the prior art well-known to those skilled in the art.

Claims

1. A high-precision center frequency detection method for space electromagnetic spectrum monitoring, characterized in that: The process includes: Step 1: Noise floor compensation is performed on the original AD sampling data, and the curve is smoothed by Sgolay filter; Step 2: Use a multi-parameter fusion peak-finding algorithm to find multiple peak points on the filtered data; Step 3: After finding the peak value, take half of the height to make a half-peak horizontal line, and find the intersection point between the half-peak horizontal line and both sides of the peak; Step 4: Find two spectrum points that are closest to the half-peak horizontal line at the intersection points on both sides of the peak through the half-peak horizontal line, obtain the left and right intervals, and perform high-resolution Fourier transform in the two intervals to obtain the refined frequency; Step 5: Perform linear interpolation fitting on the refined frequencies of the two intervals respectively, obtain the two endpoints that intersect the half-peak horizontal line, and calculate the center value as the center frequency point.

2. A high-precision center frequency detection method for space electromagnetic spectrum monitoring according to claim 1, characterized in that: The specific process of step 2 is as follows: Set multiple parameters: (1) Minimum peak height: limits the minimum peak height above the noise floor and pre-selects peaks whose power exceeds the set threshold; (2) Minimum peak relative height difference: Set the minimum relative height difference and remove the adjacent peaks whose relative height difference with the main peak is lower than the minimum relative height difference; (3) Minimum peak spacing: Set the minimum peak spacing to filter out the sidelobe signals of sharp multi-peak signals and find the main peak; (4) Minimum and maximum peak width: Specify the maximum and minimum peak widths, specify bandwidth signals within a specific range, and capture and collect signals of interest; Find multiple peak points that simultaneously satisfy the above four parameters.

Images