A method for noise reduction in RF links based on spectrum cancellation

Abstract

This invention introduces a method for noise reduction in radio frequency links based on spectrum cancellation. The specific steps are as follows: S1: Signal modeling and acquisition; S2: Spectrum analysis and frequency band division; S3: Noise floor gain estimation and quantization calculation of raised noise power; S4: Spectrum cancellation and splicing processing; S5: Effect evaluation. This invention captures the statistical characteristics of noise floor in real time by referencing the frequency band and dynamically estimates the noise floor gain of the target frequency band. It can flexibly adapt to the time-varying characteristics of noise floor under different interference scenarios and achieve accurate tracking and elimination of noise floor without manual intervention. For the transition frequency band, a linear weighted cancellation strategy is adopted, which effectively avoids the spectrum jump and signal distortion problems caused by traditional hard threshold processing or fixed filtering, ensuring a smooth transition of the signal spectrum after processing and fully preserving the detailed features of the target signal. It has good suppression effects on various typical and complex interferences such as aiming interference, jamming interference, sweeping interference, and comb spectrum interference.

Description

Technical Field

[0001] This invention relates to the field of radio frequency signal processing technology, and in particular to a method for eliminating background noise in a radio frequency link based on spectrum cancellation. Background Technology

[0002] In the field of radar signal processing, target detection and identification presuppose the effective extraction of weak target signals from complex electromagnetic environments. However, background noise and various types of interference remain key bottlenecks in radar signal processing. When a radar is in operation, it must deal with multiple interference factors, including natural noise such as random noise caused by natural factors such as ambient temperature; man-made interference such as intentional electromagnetic interference generated by enemy electronic countermeasures; and multipath effects. These noise and interference sources highly overlap with the target signal in the time and frequency domains, causing the target echo signal to be masked. This situation directly affects the effect of subsequent signal processing, such as the accuracy of signal detection, the precision of parameter estimation, and the stability of target tracking, all of which will be significantly reduced. Especially for long-range, small-cross-section targets, because their echo signal power is very weak, the effect of background noise suppression plays a decisive role to a large extent, directly determining whether the radar system can successfully achieve effective target detection.

[0003] The sources of noise floor in radar signals are significantly complex. From a physical perspective, thermal noise is generated by the thermal motion of electrons, exhibits a Gaussian distribution, and is widely present in the RF front-end and signal processing links of radar receivers. External interference, such as industrial electromagnetic radiation and communication signals, is random and diverse, including frequency targeting interference (achieving wideband coverage through random phase modulation), blocking interference (using strong power narrowband signals to suppress specific frequency bands), frequency sweeping interference (periodically sweeping within the target frequency band), and comb spectrum interference (discretely distributed across multiple frequency points). When these interferences are superimposed on the noise floor, they cause a significant decrease in the signal-to-noise ratio (SNR) of the target signal. The presence of interference signals further compresses the dynamic range of the target signal, causing traditional threshold detection methods to frequently result in missed or false detections.

[0004] While existing noise reduction technologies have developed various solutions, they still have significant limitations in complex scenarios. From an algorithmic perspective, mainstream methods can be categorized into time-domain filtering (e.g., mean denoising, sliding window smoothing), frequency-domain processing (e.g., bandpass filtering, frequency band template denoising), statistical learning (e.g., CFAR constant false alarm rate detection), and time-frequency analysis (e.g., wavelet denoising, EMD empirical mode decomposition). Time-domain methods suppress high-frequency noise through local smoothing, but easily obscure the pulse characteristics of the target signal, such as the pulse width and repetition period of radar echoes. Frequency-domain methods rely on the assumption of spectral separation between the target signal and noise; however, real-world interference, such as comb-spectrum interference, often overlaps with the target frequency band, and over-filtering can lead to the loss of target information. The CFAR algorithm adaptively estimates background noise and sets a threshold, but its performance is significantly affected by the training window selection and lacks robustness in non-uniform noise environments. While wavelet denoising and EMD methods can balance time-frequency resolution, their ability to suppress strong interference, such as blocking interference, is limited, and parameter tuning, such as wavelet basis selection and the number of decomposition levels, lacks a unified standard.

[0005] More importantly, existing research often focuses on single interference scenarios or the verification of a few algorithms, lacking a systematic consideration of complex radar environments. On the one hand, signal models often simplify actual interference characteristics, such as ignoring real-world problems like the superposition of multiple interference sources and the time-varying nature of noise statistics. This results in algorithms performing well in laboratory environments but experiencing a sharp drop in adaptability when deployed to real radar systems. On the other hand, performance comparisons of different denoising algorithms lack quantitative standards, and subjective visual evaluations such as spectrum comparisons are insufficient to support algorithm selection in engineering applications. For example, in scenarios involving a mixture of frequency-sweeping interference and Gaussian noise, wavelet denoising may be better at preserving target signal details, while band template denoising is more efficient at suppressing narrowband interference. However, existing research has not yet established a matching relationship between different interference types and optimal algorithms.

[0006] Therefore, in radar signal processing, constructing a disturbance signal model that closely resembles real-world scenarios, incorporating typical interference and noise floor, and systematically evaluating the performance of various denoising algorithms under different interference combinations, such as through quantitative indicators like RMSE, becomes crucial for overcoming the bottleneck in noise floor elimination technology. This research not only provides a theoretical basis for algorithm selection in radar systems under complex electromagnetic environments but also improves the accuracy of target signal extraction by optimizing denoising strategies, ultimately enhancing the radar's adaptability to weak targets and complex interference scenarios, thus possessing significant engineering application value. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for eliminating background noise in radio frequency links based on spectrum cancellation, which is suitable for radio frequency link signal purification in complex interference environments.

[0008] The technical solution adopted in this invention is: A method for noise reduction in an RF link based on spectrum cancellation, comprising the following steps: S1: Signal modeling and acquisition; acquisition of mixed signals in the RF link. It includes the target signal, the interference signal, and the noise floor; the mathematical model is: ;in, For target signal, This is an interference signal. This is the background noise.

[0009] Specifically, step S1 includes: Step 1.1: Parameter initialization; This includes system basic parameters, time parameters, target signal parameters, and interference and noise parameters; Step 1.2: Generate the target signal; Target signal Generate a linear frequency modulated pulse signal using the radio frequency system; the expression is: In the formula, For the target signal amplitude, For pulse window functions, For the center frequency, For signal bandwidth, The frequency modulation slope is T, and the pulse width is T. Initial phase; pulse window function For Hanning Window: ; Step 1.3: Generate interference signal Interference signal This is a typical set of interferences in complex electromagnetic environments, including frequency targeting interference. Blocking interference Frequency sweeping interference and comb-like interference ; The expression for the aiming interference is: ;in, For the amplitude of interference, The frequency modulation factor for the targeting interference; It is a modulated signal; The initial phase of the interference; Blocking interference The expression is: ;in, For blocking interference frequency modulation coefficients, The initial phase of the interference; Frequency sweeping interference The expression is: ;in, For the frequency sweep rate, The frequency modulation coefficient for the sweep frequency interference. The initial phase of the interference; Comb-like interference : ;in, For the number of comb-like spectra, , For frequency intervals, The initial phase of the interference; Step 1.4: Synthesize the mixed signal : ;in, The noise floor is Gaussian white noise, originating from the thermal noise of RF devices and environmental noise, and meets the following requirements. ,in The noise variance is determined jointly by the interference power and the signal-to-interference-plus-noise ratio (JNR) to reproduce the noise level of the actual link.

[0010] S2: Spectrum analysis and frequency band allocation; for mixed signals Perform a Discrete Fourier Transform to convert a time-domain signal into a frequency-domain signal. Based on the spectral characteristics of the target signal and the distribution pattern of interference, three functional frequency bands are divided: reference frequency band, target frequency band, and transition frequency band.

[0011] Specifically, the reference frequency band is a clean reference region in the spectrum that contains no target signal and only the inherent noise floor and interference of the radio frequency link; the reference frequency band includes two reference sub-bands symmetrically distributed on both sides of the target frequency band, and the range is divided according to the center frequency of the target signal. and bandwidth The core function of the reference band Ref_Band is to provide an objective, target-signal-free statistical basis for noise floor gain estimation. Since this band has no target signal components, it only contains the inherent noise floor of the RF link, such as thermal noise, device noise, and external interference such as FM interference and jamming interference. Its noise floor characteristics can directly reflect the real noise floor level of the link in the vicinity of the target frequency band, providing a benchmark for the subsequent calculation of the noise floor rise in the target frequency band.

[0012] The target frequency band is the core region of the spectrum containing the core spectrum of the target signal that needs to be processed. The division of the target frequency band is directly based on the effective spectral range of the target signal, with the center frequency of the signal as the starting point. and bandwidth Based on the reference frequency band, the frequency range covers the core frequency components of the target signal. The target band is the main target of the noise reduction operation. The signal components in this band are complex, including the target signal, the noise from the same source as the reference band, and various external interferences such as frequency modulation, blocking, frequency sweeping, and comb spectrum interference. The detection and demodulation performance of the target signal is directly affected by the noise and interference in this band, thus making it the core focus area for spectrum cancellation processing.

[0013] The transition frequency band is the smooth transition region in the spectrum connecting the reference frequency band and the target frequency band; the transition frequency band includes two symmetrical transition sub-bands, corresponding to the transition portions between the two sub-bands of the reference frequency band and the target frequency band, respectively, and the division range is still based on... and Based on the reference frequency band, the core function of the Trans_Band is to avoid spectral jumps caused by abrupt changes in cancellation intensity at the junction of the reference frequency band without cancellation operation and the target frequency band with full / precise cancellation operation, thus ensuring the continuity of the spectrum and signal integrity after the entire processing. The signal components of this frequency band are between the reference frequency band and the target frequency band, including both background noise and interference, as well as a small amount of edge spectrum components of the target signal. Therefore, it is necessary to dynamically adjust the cancellation intensity through a linear weighted cancellation strategy to achieve a smooth transition from no cancellation in the reference frequency band to precise cancellation in the target frequency band.

[0014] S3: Noise Floor Gain Estimation and Noise Power Boost Quantization calculation; reference noise floor average power based on the reference frequency band Total average power of the target frequency band Calculate the noise floor gain and the lift noise power that needs to be subtracted The specific steps are as follows: 1. Reference noise floor average power in the reference frequency band calculate; First, extract the spectrum within the two reference sub-bands; Subsequently, frequency domain averaging was performed on the spectrum of each reference sub-band to obtain the average noise floor power of each sub-band; Finally, the average noise floor power of the two reference sub-bands is fused using the arithmetic mean method to obtain the reference noise floor power of the reference band. : ;in, For the reference sub-band 1, the average noise floor power, The average noise floor power of reference sub-band 2; 2. Average total power of the target frequency band Calculation; Perform frequency domain averaging on the spectrum of the target frequency band to obtain the average total power of the target frequency band, i.e.: ;in, This represents the arithmetic mean operator within the frequency domain. 3. Noise floor gain and noise power Quantization calculation; noise floor gain Defined as the average total power of the target frequency band Average power of the reference noise floor compared to the reference frequency band The ratio: ; Calculate the lift noise power that needs to be subtracted : .

[0015] S4: Spectrum cancellation and splicing processing; subtracting the boost noise power from the spectrum of the target frequency band. To eliminate the noise floor component; the frequency spectrum in the transition band is weighted by a linear weighting factor to subtract the boosted noise power. A portion of the components are used to achieve a smooth transition; the spectrum of the reference frequency band remains unchanged, thus completing spectrum cancellation; The spectra of the target frequency band, transition frequency band, and reference frequency band after cancellation are spliced ​​together to output the signal spectrum after noise reduction. .

[0016] Specifically, the specific steps for the spectrum cancellation are as follows: 1. Precise cancellation of background noise in the target frequency band; Subtract the rise noise power from the spectrum of each frequency point within the target frequency band. This is done to eliminate the excessive noise component; specifically: ;in, To cancel the spectrum of the target frequency band before the cancellation, The spectrum of the target frequency band after cancellation; 2. Linear weighted smoothing cancellation in the transition frequency band; The frequency points in transition sub-band 1 that are closer to the reference frequency band have a noise floor characteristic that is closer to the reference frequency band's baseline noise floor and do not require cancellation; the frequency points that are closer to the target frequency band have a noise floor characteristic that is closer to the target frequency band's excess noise floor and need to be fully cancelled; therefore, the weighting coefficients of transition sub-band 1... linearly increasing from 0 to The mathematical expression is: in, This refers to the frequency index within transition sub-band 1. Let be the total number of frequency points in transition sub-band 1; the cancellation formula based on this weighting coefficient is: ; For frequencies in transition sub-band 2 closer to the target frequency band, the full cancellation strength must be maintained consistent with the target frequency band; for frequencies closer to the reference frequency band, the cancellation strength must be gradually reduced to 0 to match the reference frequency band's baseline noise floor. Therefore, the weighting coefficients of this frequency band... from Decreasing linearly to 0, the mathematical expression is: in, This refers to the frequency index within transition sub-band 2. Let be the total number of frequency points in transition sub-band 2; the cancellation formula based on this weighting coefficient is: ; 3. Reference band spectrum preservation; .

[0017] S5: Performance Evaluation; The performance of the noise reduction method based on spectrum cancellation is objectively characterized using quantitative indicators. The evaluation focuses on the degree of agreement between the denoised signal spectrum and the ideal target signal spectrum. Simultaneously, it verifies the superiority of this method in improving target signal quality and suppressing complex interference, providing quantitative evidence for its engineering applicability. Specifically: Constructing the spectrum of an ideal noise-free target signal , parameters and Consistent, the root mean square error (RMSE) was used to evaluate the elimination effect. The calculation formula is as follows: ; The amplitude of the signal spectrum at the i-th frequency point after the spectrum cancellation process in step 4; Let be the amplitude of the spectrum of the ideal noiseless target signal at the i-th frequency point; The total number of frequency points participating in the evaluation covers all effective frequency points in the reference band, target band, and transition band, ensuring that the evaluation covers the entire signal processing bandwidth.

[0018] Due to the adoption of the technical solution described above, the present invention has the following advantages: 1. Adaptive noise floor suppression capability: By capturing the statistical characteristics of the noise floor in real time through the reference frequency band, the noise floor gain of the target frequency band is dynamically estimated. It can flexibly adapt to the time-varying characteristics of the noise floor under different interference scenarios and achieve accurate tracking and elimination of the noise floor without manual intervention.

[0019] 2. Spectral continuity and signal integrity: A linear weighted cancellation strategy is adopted for the transition frequency band, which effectively avoids the spectral jump and signal distortion problems caused by traditional hard threshold processing or fixed filtering, ensuring a smooth transition of the processed signal spectrum and fully preserving the detailed features of the target signal.

[0020] 3. Strong anti-interference universality: It has a good suppression effect on a variety of typical and complex interferences such as aiming interference, blocking interference, sweeping interference, and comb spectrum interference. Experimental data verify that its root mean square error (RMSE) is reduced by 15%-30% compared with traditional denoising methods, which significantly improves the signal-to-noise ratio of the target signal.

[0021] 4. High engineering applicability: The core algorithm is based on Fourier transform and linear operation, with low computational complexity. It does not require complex parameter tuning or high-order mathematical operations, and is easy to deploy and implement in RF link-related hardware systems, thus possessing strong engineering application value. Attached Figure Description

[0022] Figure 1 This is a summary diagram of the spectrum of the original signal and the enhanced signal corresponding to the four types of interference signals of this invention.

[0023] Figure 2 This is a comparison chart of the noise reduction effects of the method of the present invention and six traditional noise reduction methods under four types of interference. Detailed Implementation

[0024] The present invention will be further explained and described below with reference to the accompanying drawings and embodiments. However, this should not be construed as limiting the scope of protection of the present invention. The purpose of disclosing the present invention is to protect all technical improvements within the scope of the present invention.

[0025] Combined with appendix Figure 1-2 The method for noise reduction in an RF link based on spectrum cancellation, as shown, includes the following steps: S1: Signal modeling and acquisition; Step 1.1: Parameter initialization; System basic parameters: Define system impedance (50Ω) and reference power (1mW, for dBm unit conversion). Time parameters: Set pulse width (10μs), pulse repetition time (50μs), number of pulses (3), sampling frequency (1GHz), calculate total time (150μs) and generate time series; Target signal parameters: target power (-30dBm, corresponding to 1nW), converted to voltage amplitude; carrier frequency (200MHz), signal bandwidth (20MHz), frequency modulation slope (2e). 12 (Hz / s), and a random initial phase; Interference and noise parameters: Interference power (-20dBm, corresponding to 10nW), converted to voltage amplitude; Interference-to-noise ratio (JNR=10dB), calculated noise standard deviation (amplitude parameter of Gaussian white noise).

[0026] Step 1.2: Generate the target signal; Using a linear frequency modulated (LFM) pulse signal model: Pulse window function For Hanning Window: .

[0027] Step 1.3: Generate interference signals; Interference signal This is a typical set of interferences in complex electromagnetic environments, including frequency targeting interference. Blocking interference Frequency sweeping interference and comb-like interference ; The expression for the aiming interference is: ;in, For the amplitude of interference, The frequency modulation factor for the targeting interference; It is a modulated signal; The initial phase of the interference; Blocking interference The expression is: ;in, For blocking interference frequency modulation coefficients, The initial phase of the interference; Frequency sweeping interference The expression is: ;in For the frequency sweep rate, The frequency modulation coefficient for the sweep frequency interference. The initial phase of the interference; Comb-like interference : ;in For the number of comb-like spectra, , For frequency intervals, This is the initial phase of the interference.

[0028] Step 1.4: Synthesize the mixed signal : ;in, The noise floor is Gaussian white noise, originating from the thermal noise of RF devices and environmental noise, and meets the following requirements. ,in The noise variance is determined jointly by the interference power and the signal-to-interference-plus-noise ratio (JNR) to reproduce the noise level of the actual link.

[0029] S2: Spectrum analysis and frequency band allocation; For mixed signals Perform a Discrete Fourier Transform to convert a time-domain signal into a frequency-domain signal. Based on the spectral characteristics of the target signal and the distribution pattern of interference, three functional frequency bands are divided: reference frequency band, target frequency band, and transition frequency band.

[0030] Based on the parameters in step 1, the carrier frequency is 200MHz and the bandwidth is 20MHz, so the target frequency band range is 190MHz-210MHz; the reference frequency band is symmetrical on both sides with a width of 10MHz, so the reference sub-band 1 is 180MHz-185MHz and the reference sub-band 2 is 215MHz-220MHz; the transition frequency band 1 is 185MHz-195MHz and the transition frequency band 2 is 205MHz-215MHz.

[0031] Figure 1 The spectrograms of the original and synthesized received signals corresponding to the four types of interference are given for easy and intuitive comparison.

[0032] S3: Noise floor gain estimation and boosting noise power Quantization calculation; Reference noise floor average power based on the reference frequency band Total average power of the target frequency band Calculate the noise floor gain ; The core objective of noise floor gain estimation is to quantify the relative increase in noise floor within the target frequency band based on the statistical characteristics of the noise floor in a clean region of the reference frequency band containing only noise and interference, thereby providing core quantization parameters for accurate noise floor cancellation in the target frequency band. The specific process is as follows: 1. Reference noise floor average power in the reference frequency band calculate; First, extract the spectrum within the two reference sub-bands; Subsequently, frequency domain averaging was performed on the spectrum of each reference sub-band to obtain the average noise floor power of each sub-band; Finally, the average noise floor power of the two reference sub-bands is fused using the arithmetic mean method to obtain the reference noise floor power of the reference band. : ; in, The average noise floor power of reference frequency sub-segment 1, The average noise floor power of reference sub-band 2.

[0033] 2. Average total power of the target frequency band Calculation; The target frequency band is a composite region containing the target signal, noise floor, and interference. Its spectrum is the superposition of the target signal power, noise floor power, and interference power. To obtain the basic data on the noise floor rise within this frequency band, a frequency domain mean operation is performed on the spectrum of the target frequency band to obtain the total power mean of the target frequency band, i.e.: ;in, This represents the arithmetic mean operator within the frequency domain. Essentially, it is a composite power representation of the target signal power and the superimposed noise floor power, which is related to... The differences mainly stem from the additional increase in noise floor within the target frequency band, the distribution differences of link noise in the target frequency band, and the local superposition effect of interference.

[0034] 3. Noise floor gain Quantization calculation; noise floor gain Defined as the average total power of the target frequency band Average power of the reference noise floor compared to the reference frequency band The ratio: ; The physical meaning of this ratio is: the noise floor rise factor of the target frequency band relative to the reference frequency band, if... A value greater than 1 indicates that the noise floor level (including the contribution of interference) in the target frequency band is higher than the inherent noise floor level of the reference frequency band. The magnitude of this value directly quantifies the noise floor intensity that needs to be canceled in the target frequency band; if =1 indicates that the target frequency band noise floor is consistent with the inherent noise floor of the reference frequency band, and no additional lift is needed to cancel it out; The value serves as the core basis for subsequent target frequency band noise cancellation, ensuring the quantification and adaptive characteristics of the cancellation process and avoiding over-cancellation or under-cancellation problems caused by traditional fixed threshold cancellation.

[0035] 4. Calculate the lift noise power that needs to be subtracted. : .

[0036] S4: Spectrum cancellation and splicing processing; Subtract the noise floor gain from the spectrum of the target frequency band To eliminate the noise floor component; the noise floor gain is subtracted from the spectrum of the transition frequency band using a linear weighting coefficient. A portion of the components are used to achieve a smooth transition; the spectrum of the reference frequency band remains unchanged, thus completing spectrum cancellation; The core objective of spectrum cancellation processing is to estimate the noise floor gain based on step 3. To address the functional differences between the reference frequency band, target frequency band, and transition frequency band, a differentiated signal processing strategy is adopted. This strategy accurately eliminates the noise floor of the target frequency band while ensuring the continuity of the entire spectrum and the integrity of the target signal, thus avoiding spectral distortion caused by traditional filtering or thresholding.

[0037] Specifically: 1. Precise cancellation of background noise in the target frequency band; The target frequency band is the core region containing the target signal, noise floor, and interference. The spectrum of the target frequency band is composed of the target signal power, interference power, and the excess noise floor resulting from the noise floor increase relative to the reference frequency band. The noise floor gain is based on the noise floor increase of the target frequency band relative to the reference frequency band. The physical meaning is to perform noise floor cancellation on this frequency band: subtract the frequency spectrum of each frequency point within the target frequency band. This is done to eliminate the excessive noise component; specifically: ;in, To cancel the spectrum of the target frequency band before the cancellation, The spectrum of the target frequency band after cancellation is obtained. The advantage of this method is that it only targets and eliminates the noise floor rise, preserving the original power characteristics and spectral details of the target signal to the maximum extent, and avoiding suppression or distortion of the target signal itself.

[0038] 2. Linear weighted smoothing cancellation in the transition frequency band; The transition band is a critical connecting region between the uncancelled reference band and the target band with full cancellation. If a full cancellation strategy consistent with the target band is adopted, it will cause a spectral jump at the band transition, thus introducing additional distortion. Therefore, a linear weighted cancellation strategy is adopted, which dynamically adjusts the cancellation intensity of each frequency point to achieve a smooth spectral transition. The frequency points in transition sub-band 1 that are closer to the reference frequency band have a noise floor characteristic that is closer to the reference frequency band's baseline noise floor and do not require cancellation; the frequency points that are closer to the target frequency band have a noise floor characteristic that is closer to the target frequency band's excess noise floor and need to be fully cancelled; therefore, the weighting coefficients of transition sub-band 1... linearly increasing from 0 to The mathematical expression is: in, This refers to the frequency index within transition sub-band 1. Let be the total number of frequency points in transition sub-band 1; the cancellation formula based on this weighting coefficient is: ; For frequencies in transition sub-band 2 closer to the target frequency band, the full cancellation strength must be maintained consistent with the target frequency band; for frequencies closer to the reference frequency band, the cancellation strength must be gradually reduced to 0 to match the reference frequency band's baseline noise floor. Therefore, the weighting coefficients of this frequency band... linearly increasing from 0 to The mathematical expression is: in, This refers to the frequency index within transition sub-band 2. Let be the total number of frequency points in transition sub-band 2; the cancellation formula based on this weighting coefficient is: .

[0039] 3. Reference band spectrum preservation; The core function of the reference frequency band is to improve the noise floor gain. The estimated noise floor characteristics provide a baseline, which itself does not contain the target signal, and the noise floor level has been used as a reference for the entire cancellation process. Therefore, no cancellation operation is performed on the spectrum of this frequency band, preserving its original spectral characteristics. ; To ensure the stability of the noise floor baseline in subsequent signal processing, while avoiding meaningless signal interference.

[0040] 4. Spectrum splicing; The spectra of the target frequency band, transition frequency band, and reference frequency band after cancellation are superimposed to output the signal spectrum after noise reduction. .

[0041] S5: Effectiveness Evaluation: The performance of the noise reduction method based on spectrum cancellation is objectively characterized by quantitative indicators. The focus is on evaluating the degree of agreement between the denoised signal spectrum and the ideal target signal spectrum. Simultaneously, the superiority of this method in improving target signal quality and suppressing complex interference is verified, providing quantitative evidence for its engineering applicability. Specifically: Constructing the spectrum of an ideal noise-free target signal , parameters and Consistent, the root mean square error (RMSE) was used to evaluate the elimination effect. The calculation formula is as follows: ; The amplitude of the signal spectrum at the i-th frequency point after the spectrum cancellation process in step 4; Let be the amplitude of the spectrum of the ideal noiseless target signal at the i-th frequency point; The total number of frequency points participating in the evaluation covers all effective frequency points in the reference band, target band, and transition band, ensuring that the evaluation covers the entire signal processing bandwidth.

[0042] The RMSE of this method is quantitatively compared with six traditional noise reduction methods (mean denoising, CFAR denoising, filtering denoising, wavelet denoising, EMD denoising, and sparse denoising). The denoising performance of different algorithms under four types of interference is presented, such as... Figure 2 As shown in the figure. Experimental results show that, under complex scenarios such as frequency targeting interference, jamming interference, frequency sweeping interference, and comb spectrum interference, the RMSE of this method is reduced by 15%-30% compared with the traditional method, which intuitively demonstrates its significant advantages in terms of thorough noise suppression and preservation of target signal integrity.

[0043] The root mean square error (RMSE) of the seven denoising algorithms under each type of interference is calculated to quantitatively evaluate the denoising effect. The smaller the RMSE value, the better the denoising effect.

[0044] The comparison in the table below shows that the spectrum cancellation and denoising method proposed in this invention is significantly better than other denoising methods. Table: Comparison of the root mean square error of the spectrum cancellation noise reduction method of the present invention with six traditional noise floor elimination methods under four types of interference. Frequency jamming Blocking interference Frequency sweeping interference Comb-like interference Spectrum cancellation RMSE 0.5808dBm 1.8588dBm 1.3180dBm 0.6759dBm Mean Denoising RMSE 6.0932dBm 6.0242dBm 5.9580dBm 6.0856dBm CFAR noise reduction RMSE 3.8499dBm 3.8903dBm 3.8366dBm 3.8642dBm Filtering and denoising RMSE 2.6311dBm 2.6310dBm 2.6257dBm 2.6309dBm Wavelet denoising RMSE 6.0428dBm 5.9721dBm 5.9109dBm 6.0346dBm EMD noise reduction RMSE 32.7939dBm 30.9523dBm 31.1926dBm 32.8170dBm Sparse noise reduction RMSE 2.5822dBm 2.5822dBm 2.5822dBm 2.5822dBm The parts of this invention not described in detail are prior art.

[0045] The embodiments selected herein for the purpose of disclosing the inventive objectives are currently considered suitable; however, it should be understood that the invention is intended to include all variations and modifications of the embodiments that fall within the scope of this concept and invention.

Claims

1. A method for noise floor cancellation in a radio frequency link based on spectrum cancellation, characterized in that: The specific steps are as follows: S1: Signal modeling and acquisition; acquisition of mixed signals in the RF link. It includes the target signal, the interference signal, and the noise floor; the mathematical model is: ;in, For target signal, This is an interference signal. This is the background noise; S2: Spectrum analysis and frequency band allocation; for mixed signals Perform a Discrete Fourier Transform to convert a time-domain signal into a frequency-domain signal. Based on the spectral characteristics of the target signal and the distribution pattern of interference, three functional frequency bands are divided: reference frequency band, target frequency band, and transition frequency band. S3: Noise Floor Gain Estimation and Noise Power Boost Quantization calculation; reference noise floor average power based on the reference frequency band Total average power of the target frequency band Calculate the noise floor gain and the lift noise power that needs to be subtracted ; S4: Spectrum cancellation and splicing processing; subtracting the boost noise power from the spectrum of the target frequency band. To eliminate the noise floor component; the frequency spectrum in the transition band is weighted by a linear weighting factor to subtract the boosted noise power. A portion of the components are used to achieve a smooth transition; the spectrum of the reference frequency band remains unchanged, thus completing spectrum cancellation; The spectra of the target frequency band, transition frequency band, and reference frequency band after cancellation are spliced ​​together to output the signal spectrum after noise reduction. ; S5: Performance Evaluation; Objectively characterize the performance of the noise reduction method based on spectrum cancellation through quantitative indicators, focusing on evaluating the degree of agreement between the denoised signal spectrum and the ideal target signal spectrum, while verifying the superiority of the method in improving the quality of the target signal and suppressing complex interference, and providing quantitative basis for its engineering applicability.

2. The method for eliminating background noise in a radio frequency link based on spectrum cancellation according to claim 1, characterized in that: Step S1 specifically includes: Step 1.1: Parameter initialization; This includes system basic parameters, time parameters, target signal parameters, and interference and noise parameters; Step 1.2: Generate the target signal; Target signal Generate a linear frequency modulated pulse signal using the radio frequency system; the expression is: In the formula, For the target signal amplitude, For pulse window functions, For the center frequency, For signal bandwidth, The frequency modulation slope is T, and the pulse width is T. Initial phase; pulse window function For Hanning Window: ; Step 1.3: Generate interference signal Interference signal This is a typical set of interferences in complex electromagnetic environments, including frequency targeting interference. Blocking interference Frequency sweeping interference and comb-like interference ; The expression for frequency interference is: ;in, For the amplitude of interference, The frequency modulation factor for the targeting interference; It is a modulated signal; The initial phase of the interference; Blocking interference The expression is: ;in, For blocking interference frequency modulation coefficients, The initial phase of the interference; Frequency sweeping interference The expression is: ;in, For the frequency sweep rate, The frequency modulation coefficient for the sweep frequency interference. The initial phase of the interference; Comb-like interference : ;in, For the number of comb-like spectra, , For frequency intervals, The initial phase of the interference; Step 1.4: Synthesize the mixed signal : ;in, The noise floor is Gaussian white noise, originating from the thermal noise of RF devices and environmental noise, and meets the following requirements. ,in The noise variance is determined jointly by the interference power and the signal-to-interference-plus-noise ratio (JNR) to reproduce the noise level of the actual link.

3. The method for eliminating background noise in a radio frequency link based on spectrum cancellation according to claim 1, characterized in that: In step S2, the reference frequency band is a clean reference region in the spectrum that contains no target signal and only the inherent noise floor and interference of the radio frequency link; the reference frequency band includes two reference sub-bands symmetrically distributed on both sides of the target frequency band, and the range is divided according to the center frequency of the target signal. and bandwidth As a benchmark; The target frequency band is the core region of the spectrum containing the core spectrum of the target signal that needs to be processed. The division of the target frequency band is directly based on the effective spectral range of the target signal, with the center frequency of the signal as the starting point. and bandwidth Based on this, the frequency range covers the core frequency components of the target signal; The transition frequency band is the smooth transition region in the spectrum connecting the reference frequency band and the target frequency band; the transition frequency band includes two symmetrical transition sub-bands, corresponding to the transition portions between the two sub-bands of the reference frequency band and the target frequency band, respectively, and the division range is still based on... and Based on.

4. The method for eliminating background noise in a radio frequency link based on spectrum cancellation according to claim 3, characterized in that: The specific steps of step S3 are as follows:

1. Reference noise floor average power in the reference frequency band calculate; First, extract the spectrum within the two reference sub-bands; Subsequently, frequency domain averaging was performed on the spectrum of each reference sub-band to obtain the average noise floor power of each sub-band; Finally, the average noise floor power of the two reference sub-bands is fused using the arithmetic mean method to obtain the reference noise floor power of the reference band. : ;in, For the reference sub-band 1, the average noise floor power, The average noise floor power of reference sub-band 2; 2. Average total power of the target frequency band Calculation; Perform frequency domain averaging on the spectrum of the target frequency band to obtain the average total power of the target frequency band, i.e.: ;in, This represents the arithmetic mean operator within the frequency domain.

3. Noise floor gain and noise power Quantization calculation; noise floor gain Defined as the average total power of the target frequency band Average power of the reference noise floor compared to the reference frequency band The ratio: ; Calculate the lift noise power that needs to be subtracted : .

5. The method for eliminating background noise in a radio frequency link based on spectrum cancellation according to claim 3, characterized in that: The specific steps of spectrum cancellation in step S4 are as follows:

1. Precise cancellation of background noise in the target frequency band; Subtract the rise noise power from the spectrum of each frequency point within the target frequency band. This is done to eliminate the excessive noise component; specifically: ;in, To cancel the spectrum of the target frequency band before the cancellation, The spectrum of the target frequency band after cancellation; 2. Linear weighted smoothing cancellation in the transition frequency band; The frequency points in transition sub-band 1 that are closer to the reference frequency band have a noise floor characteristic that is closer to the reference frequency band's baseline noise floor and do not require cancellation; the frequency points that are closer to the target frequency band have a noise floor characteristic that is closer to the target frequency band's excess noise floor and need to be fully cancelled; therefore, the weighting coefficients of transition sub-band 1... linearly increasing from 0 to The mathematical expression is: in, This refers to the frequency index within transition sub-band 1. Let be the total number of frequency points in transition sub-band 1; the cancellation formula based on this weighting coefficient is: ; For frequencies in transition sub-band 2 closer to the target frequency band, the full cancellation strength must be maintained consistent with the target frequency band; for frequencies closer to the reference frequency band, the cancellation strength must be gradually reduced to 0 to match the reference frequency band's baseline noise floor. Therefore, the weighting coefficients of this frequency band... from Decreasing linearly to 0, the mathematical expression is: in, This refers to the frequency index within transition sub-band 2. Let be the total number of frequency points in transition sub-band 2; the cancellation formula based on this weighting coefficient is: ; 3. Reference band spectrum preservation; 。 6. The method for eliminating background noise in a radio frequency link based on spectrum cancellation according to claim 1, characterized in that: The specific steps of step S5 are as follows: Constructing the spectrum of an ideal noise-free target signal , parameters and Consistent, the root mean square error (RMSE) was used to evaluate the elimination effect. The calculation formula is as follows: ; The amplitude of the signal spectrum at the i-th frequency point after the spectrum cancellation process in step 4; Let be the amplitude of the spectrum of the ideal noiseless target signal at the i-th frequency point; The total number of frequency points participating in the evaluation covers all effective frequency points in the reference band, target band, and transition band, ensuring that the evaluation covers the entire signal processing bandwidth.

Images