Mirror image suppression method and device based on spectral entropy
By utilizing the difference in spectral entropy characteristics between the image signal and the useful signal, a low-cost and high-precision image suppression method is achieved, solving the problems of high cost and insufficient robustness in existing technologies. This method is suitable for software radio platforms.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING AEROSPACE MEASUREMENT & CONTROL TECH
- Filing Date
- 2025-12-16
- Publication Date
- 2026-05-15
AI Technical Summary
Existing image suppression technologies are costly, lack robustness, and are difficult to adapt to scenarios with dynamically changing frequencies.
By distinguishing between useful signals and mirror signals using the spectral entropy characteristics of signals, and taking advantage of the large fluctuations in spectral entropy caused by the nonlinear shift of the local oscillator frequency in mirror signals, while the spectral entropy of useful signals is stable due to the linear shift, the accurate identification and suppression of mirror signals can be achieved.
It achieves low-cost, high-precision image suppression, improves the ability to distinguish in complex scenarios, reduces hardware costs, is highly adaptable, and is suitable for software radio platforms.
Smart Images
Figure CN122052818A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of radio frequency signal processing technology, and in particular to a mirror image suppression method and apparatus based on spectral entropy. Background Technology
[0002] Image rejection technology is a key technology used in receivers to suppress image frequency interference. By eliminating the interference of the image signal, which is symmetrical to the local oscillator signal, on the useful signal, the purity of the intermediate frequency signal is ensured.
[0003] Image frequency is the interference signal generated after the radio frequency signal and the local oscillator signal are mixed. Its frequency is symmetrical with respect to the local oscillator frequency to the useful signal. The core of image suppression is to reduce the amplitude of the image signal through certain technical means (such as filters or mixers) so that it cannot be mixed into the intermediate frequency band.
[0004] In existing technologies, image suppression mainly includes premixer filtering methods and complex mixing (or orthogonal mixing, vector mixing) methods. The widely used classical superheterodyne architecture employs premixer filtering. Because superheterodyne receivers have multiple mixer stages, requiring multiple image suppression filters, this method is costly, bulky, and difficult to adapt to scenarios with dynamically changing frequencies. Using complex mixing to suppress images can achieve high receiver integration and is simpler to design than premixer filtering methods, but it has higher requirements for hardware consistency. However, practical circuit defects (such as mismatch between I and Q channels) will lead to poor image suppression performance in the receiver. Summary of the Invention
[0005] This disclosure provides a mirror image suppression method and apparatus based on spectral entropy. It addresses the problems of high cost and insufficient robustness of existing mirror image suppression methods by dynamically distinguishing useful signals from mirror signals through the spectral entropy characteristics of the signal itself, thereby achieving high-precision and low-cost mirror image suppression.
[0006] The main steps of this method include: S1, Signal Acquisition and Preprocessing: Acquire multiple sets of spectra by moving the local oscillator; S2, Spectrum Frequency Alignment: Map it to a unified reference frequency axis to achieve frequency alignment; S3, Spectral Entropy Calculation: Calculate the spectral entropy sequence of aligned frequency points; S4, Signal differentiation based on spectral entropy characteristics at the same frequency: The frequency points of mirror signals are differentiated by the difference that the spectral entropy of mirror signals fluctuates greatly at the same frequency due to the nonlinear shift of the local oscillator frequency, while the spectral entropy of useful signals is stable due to the linear shift. S5, Image Signal Suppression: Based on the differentiation results, the spectrum set is corrected to eliminate image frequency interference.
[0007] The image suppression device based on spectral entropy that applies the above method mainly includes: The signal acquisition and preprocessing module is used to acquire multiple sets of spectra by moving the local oscillator; The spectrum frequency alignment module is used to map spectrum frequencies to a unified reference frequency axis to achieve frequency alignment. The spectral entropy calculation module is used to calculate the spectral entropy sequence of aligned frequency points. The mirror signal frequency point differentiation module is used to differentiate the frequency points of mirror signals by utilizing the difference that the spectral entropy of mirror signals fluctuates greatly at the same frequency point due to the nonlinear shift of the local oscillator frequency, while the spectral entropy of useful signals is stable due to the linear shift. The image signal suppression module is used to correct the spectrum set based on the differentiation results and eliminate image frequency interference.
[0008] Compared with the prior art, the beneficial effects of this disclosure are: (1) the structure is simplified, no secondary frequency conversion or complex hardware is required, and it can be directly implemented in the primary frequency conversion architecture, thus reducing hardware costs.
[0009] (2) It is robust and distinguishes signals based on spectral entropy characteristics, without requiring prior information on amplitude and phase errors.
[0010] (3) High flexibility, pure digital domain implementation, no additional hardware required, can be integrated into software radio platform.
[0011] (4) It has strong self-adaptability and can adapt to time-varying interference environments by dynamically adjusting the spectral entropy threshold. Attached Figure Description
[0012] The above and other objects, features and advantages of this disclosure will become more apparent from the accompanying drawings, in which like reference numerals generally denote like parts.
[0013] Figure 1 Here is a flowchart of the image suppression method based on spectral entropy according to this disclosure; Figure 2 This is a schematic diagram of an exemplary mirror suppression device based on spectral entropy. Detailed Implementation
[0014] Preferred embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure will be thorough and complete, and will fully convey the scope of the present disclosure to those skilled in the art.
[0015] This disclosure provides a mirror image suppression method and apparatus based on spectral entropy. The method acquires multiple sets of spectra by moving the local oscillator, maps them to a unified reference frequency axis to achieve frequency alignment, calculates the spectral entropy sequence of the same frequency points after alignment, and utilizes the difference between the large fluctuations in spectral entropy at the same frequency points caused by the nonlinear shift of the local oscillator frequency in the mirror signal, and the stable spectral entropy of the useful signal due to the linear shift. This allows for accurate identification and suppression of the mirror signal. Therefore, it solves the problems of high cost and insufficient robustness of existing mirror image suppression methods, achieving high-precision mirror image suppression and improving the discrimination ability in complex scenarios.
[0016] As shown in the attached figure, in one exemplary embodiment, the main steps of the image suppression method based on spectral entropy are as follows: 1. Signal Acquisition and Preprocessing The RF front-end adopts a single-conversion architecture to control the moving local oscillator within a preset frequency range in steps. Move to generate N discrete local oscillator frequency points. k=1,2,...,N,N≥6 to ensure feature reliability.
[0017] For each local oscillator frequency Acquire the intermediate frequency signal after down-conversion. The intermediate frequency bandwidth is B, and the local oscillator movement step should be greater than B / F (where F is the number of FFT points in subsequent signal processing). The intermediate frequency signal is filtered out for high-frequency components by a bandpass filter and then converted into a digital signal by an ADC.
[0018] For each Adding windows (such as Hanning windows) to... Perform a Fast Fourier Transform to obtain the spectrum. Calculate the power spectrum .
[0019] Calculate the normalized power spectrum. , where F is the number of FFT points.
[0020] 2. Spectrum frequency alignment Because the local oscillator shift alters the intermediate frequency (IF) signal frequency, the power spectrum of the IF signal acquired in each mixing operation is frequency-aligned before calculating the spectral entropy. For any frequency point f... In the middle, its corresponding reference axis frequency is (Restore the intermediate frequency to the physical frequency).
[0021] After mapping, all group spectra are unified to the reference axis. The above yields the aligned spectrum set. ,in .
[0022] 3. Spectral entropy calculation On the reference axis Take discrete frequency points (m=1,2,...,M, with intervals of B / F, M<F).
[0023] For each Extract its power value in N groups of aligned spectra. , forming a power sequence .
[0024] Calculate the spectral entropy sequence at this frequency point: T represents the local time window. The spectral entropy is:
[0025] 4. Signal differentiation based on spectral entropy characteristics at the same frequency points Extract each frequency point Spectral entropy sequence characteristics: Sequence volatility: (Standard deviation reflects the degree of fluctuation in spectral entropy); Sequence mean ratio: ( (The mean of the spectral entropy sequence of the known useful signal). Distinguishing criteria: like and It was determined to be a mirror signal frequency point (the spectral entropy sequence fluctuates greatly and has a high mean). Threshold determination: Statistical analysis of mirror signal frequency points using training data. and Distribution, taking the upper limit of the 95% confidence interval as and The threshold can be configured and dynamically adjusted in real time via software.
[0026] 5. Mirror signal suppression Based on the differentiation results, mark the set of mirror frequency points on the reference axis. . Spectrum set Corrections are made to eliminate image frequency interference. The minimum spectral entropy is found in the corresponding spectral entropy sequence within the image frequency set. The corresponding power sequence Take the average value The final spectrum after suppressing the mirror frequency points is obtained. For the remaining non-mirror frequency points, the average value of the aligned spectrum set is directly taken to obtain the final spectrum.
[0027] In this embodiment, multiple sets of spectra are mapped to a unified reference axis by frequency alignment. By utilizing the characteristics of large fluctuations in mirror signals and stable useful signals in spectral entropy sequences at the same frequency points, high-precision mirror suppression is achieved, improving the discrimination capability in complex scenarios.
[0028] Application Examples An example of image suppression based on spectral entropy according to this disclosure is attached. Figure 2 As shown, it specifically includes: a local oscillator, a mixer, a bandpass filter, an ADC, an FPGA, and a host computer.
[0029] Assuming the RF input frequency is 5.83GHz, the local oscillator is controlled by the FPGA and moves in steps, with the local oscillator frequency range being 5.500GHz~5.520GHz, in steps of 1MHz; After each local oscillator shift, the ADC acquires one set of data from the intermediate frequency (IF) signal, with N=20 sets of samples. The bandpass filter bandwidth B is 100MHz, and the IF signal center frequency is 320MHz, therefore the IF signal range is 270MHz~370MHz.
[0030] The first local oscillator frequency, 5.500 GHz, corresponds to an input RF signal range of 5.77 GHz to 5.87 GHz, with a mirror frequency range of 5.13 GHz to 5.23 GHz. The last local oscillator frequency, 5.520 GHz, corresponds to an input RF signal range of 5.79 GHz to 5.89 GHz, with a mirror frequency range of 5.15 GHz to 5.25 GHz. The reference frequency axis is the common portion of the 20 RF signal ranges, 5.79 GHz to 5.87 GHz, thus providing 20 sets of power spectrum data for each frequency.
[0031] A spectral entropy image suppression method based on multiple spectrum sampling: After acquiring 20 sets of intermediate frequency signals and sending them to the FPGA, the power spectrum is calculated through windowing and FFT processing (FFT points are 1024). and normalized power spectrum Each set of spectra is mapped to a reference axis to obtain the aligned set of spectra. For each frequency point on the reference axis Twenty power values were extracted. A time window of T = 6 was used to calculate the spectral entropy sequence of the frequency points. .Pick , Then in the spectral entropy sequence and The frequency points are considered mirror frequencies. For example, if a mirror frequency exists at 5.81 GHz, the minimum spectral entropy is found in the spectral entropy sequence corresponding to that frequency point, and the average of the spectra of the six frequency points corresponding to the minimum spectral entropy is taken as the final spectrum of that frequency point. For the remaining frequency points, the final spectrum is obtained by averaging the 20 sets of spectral data.
[0032] The above technical solutions are merely exemplary embodiments of the present invention. For those skilled in the art, based on the application methods and principles disclosed in the present invention, it is easy to make various types of improvements or modifications, and not limited to the methods described in the specific embodiments of the present invention. Therefore, the methods described above are merely preferred and not restrictive.
Claims
1. A mirror image suppression method based on spectral entropy, characterized in that, Includes the following steps: S1, acquires multiple sets of spectra by moving the local oscillator; S2 maps the spectrum frequencies to a unified reference frequency axis to achieve frequency alignment; S3, calculate the spectral entropy sequence of the same frequency points after alignment; S4. By utilizing the difference between the large fluctuations in spectral entropy at the same frequency point caused by the nonlinear shift of the local oscillator frequency in the mirror signal and the stable spectral entropy of the useful signal caused by the linear shift, the frequency points of the mirror signal can be distinguished. S5, based on the differentiation results, corrects the spectrum set to eliminate mirror frequency interference.
2. The method according to claim 1, characterized in that, Step S1 includes: The RF front-end adopts a single-conversion architecture to control the moving local oscillator within a preset frequency range in steps. Move to generate N discrete local oscillator frequency points. k=1,2,...,N, N≥6 to ensure feature reliability; For each local oscillator frequency Acquire the intermediate frequency signal after down-conversion. The intermediate frequency bandwidth is B, and the local oscillator movement step should be greater than B / F, where F is the number of FFT points for subsequent signal processing. The intermediate frequency signal is filtered out by a bandpass filter to remove high-frequency components and then converted into a digital signal by an ADC. For each Add windows to the window. Perform a Fast Fourier Transform to obtain the spectrum. Calculate the power spectrum ; Calculate the normalized power spectrum. , where F is the number of FFT points.
3. The method according to claim 2, characterized in that, Step S2 includes: Before calculating the spectral entropy, the power spectrum of the intermediate frequency signal acquired in each mixing operation is frequency-aligned: for any frequency point f at... In the process of restoring the intermediate frequency to the physical frequency, the corresponding reference axis frequency is: ； After mapping, all group spectra are unified to the reference axis. The above yields the aligned spectrum set. ,in .
4. The method according to claim 3, characterized in that, Step S3 includes: On the reference axis Take discrete frequency points m = 1, 2, ..., M, with intervals of B / F, where M < F; For each Extract its power value in N groups of aligned spectra. , forming a power sequence ; Calculate the spectral entropy sequence at this frequency point: T is the local time window; No. The spectral entropy is: 。 5. The method according to claim 4, characterized in that, Step S4 includes: S41, Extract each frequency point The spectral entropy sequence features include: Sequence volatility: , representing the standard deviation, reflects the degree of fluctuation in spectral entropy; Sequence mean ratio: ,in, The mean of the spectral entropy sequence of the known useful signal; S42, Distinguishing Criteria: like and It was determined to be a mirror signal frequency point; S43, Threshold determination: Statistical analysis of mirror signal frequency points using training data. and Distribution, taking the upper limit of the 95% confidence interval as and .
6. The method according to any one of claims 3-5, characterized in that, Step S5 includes: Based on the differentiation results, mark the set of mirror frequency points on the reference axis. ; Spectrum set To correct and eliminate image frequency interference, the minimum spectral entropy is found in the spectral entropy sequence corresponding to the image frequency set. The corresponding power sequence Take the average value The final spectrum after suppressing the mirror frequency points is used; for the other non-mirror frequency points, the average value of the aligned spectrum set is directly taken to obtain the final spectrum.
7. A mirror image suppression device based on spectral entropy applying the method of any one of claims 1-6, characterized in that, include: The signal acquisition and preprocessing module is used to acquire multiple sets of spectra by moving the local oscillator; The spectrum frequency alignment module is used to map spectrum frequencies to a unified reference frequency axis to achieve frequency alignment. The spectral entropy calculation module is used to calculate the spectral entropy sequence of aligned frequency points. The mirror signal frequency point differentiation module is used to differentiate the frequency points of mirror signals by utilizing the difference that the spectral entropy of mirror signals fluctuates greatly at the same frequency point due to the nonlinear shift of the local oscillator frequency, while the spectral entropy of useful signals is stable due to the linear shift. The image signal suppression module is used to correct the spectrum set based on the differentiation results and eliminate image frequency interference.