Interference suppression method based on cyclic Wiener filter iterative design

Abstract

The invention discloses an interference suppression method based on cyclic Wiener filter iterative design. The method comprises the following steps: performing down-conversion and dechirp processing on a radar echo signal to obtain a beat signal; converting the signal to a cyclic frequency domain, and calculating a cyclic power spectrum and a cyclic cross-power spectrum of the signal; extracting cyclic frequencies of linear parts of the target signal and the interference signal in the cyclic power spectrum and the cyclic cross-power spectrum by using an ordered statistical constant false alarm detection algorithm; constructing a cyclic Wiener filter through the cyclic frequency; performing multi-group cyclic Wiener filtering processing on the beat signal in parallel; and distinguishing according to the continuity of the output signal in the time domain. According to the method, the radar echo is converted to the cyclic frequency domain after signal processing, and the cyclic Wiener filter is constructed only by extracting the cyclic frequency of the target and interference signals, so that the dependence of a traditional Wiener filtering method on the statistical characteristics of the signals is overcome, and the problem that the calculation complexity is high due to the fact that the statistical characteristics need a large amount of sample data is reduced.

Description

Technical Field

[0001] The invention belongs to the technical field of radar signal interference suppression, and relates to an interference suppression method based on iterative design of a cyclic Wiener filter. Background Art

[0002] Complex electromagnetic interference can significantly reduce the detection and identification capabilities of radar systems, bringing severe challenges to the practical application of radar systems. Existing interference suppression methods mainly start from waveform design, signal processing and other aspects to achieve interference detection and suppression.

[0003] At present, for intermittent sampling forwarding interference, the Wiener filtering method is usually used to filter out the interference signal and extract the target signal to achieve interference suppression. The traditional Wiener filtering method needs to obtain the statistical characteristics of the target, interference signal and noise, such as mean, variance, power spectrum, etc. In practical applications, the statistical characteristics are complex and changeable, which makes it difficult to obtain their prior information. In addition, a large amount of sample data is required to estimate these statistical characteristics in real time, which faces the problem of high computational complexity. At the same time, typical radar transmission signals, such as linear frequency modulation signals, are non-stationary signals, which will cause the interference suppression performance of the traditional Wiener filtering method to be significantly reduced. Summary of the invention

[0004] In view of the above problems, the present invention proposes an interference suppression method based on iterative design of a cyclic Wiener filter, which well solves the problems in the prior art.

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

[0006] An interference suppression method based on iterative design of a cyclic Wiener filter, comprising:

[0007] S1, obtain a beat signal by down-converting and de-linear frequency modulation of the radar echo signal;

[0008] S2, transforming the beat signal into the cyclic frequency domain, and calculating the cyclic power spectrum and cyclic cross power spectrum of the signal;

[0009] S3, extracting the cyclic frequencies of the linear parts of the target signal and the interference signal using an ordered statistical constant false alarm detection algorithm in the cyclic power spectrum;

[0010] S4, extracting the cyclic frequency of the conjugate linear part of the target signal and the interference signal by using the ordered statistical constant false alarm detection algorithm in the cyclic cross power spectrum;

[0011] S5, constructing a cyclic Wiener filter through the cyclic frequencies in step S3 and step S4;

[0012] S6, performing multiple groups of cyclic Wiener filtering processes on the beat signal in step S1 in parallel;

[0013] S7. According to the continuity of the output signal in step S6 in the time domain, the target signal with continuous characteristics in the time domain is retained, and the forwarding interference signal with discontinuous characteristics in the time domain is discarded.

[0014] Optionally, in step S1, the beat signal obtained after down-conversion and de-linear frequency modulation of the radar echo signal is expressed as:

[0015] y(t)=S echo (t)+J(t)+n(t)

[0016] Among them, S echo (t) is the beat signal obtained by down-converting and de-linear frequency modulation of the target signal under linear frequency modulation continuous wave radar, J(t) is the beat signal obtained by down-converting and de-linear frequency modulation of the interference signal under intermittent sampling repeated forwarding interference, and n(t) is Gaussian white noise.

[0017] Optionally, the S echo (t) can be expressed as:

[0018] S echo (t) = A·exp(2jπrf s t+θ)

[0019] Among them, f s is the beat signal frequency; A is the beat signal amplitude; θ is the beat signal phase.

[0020] Optionally, the J(t) can be expressed as:

[0021]

[0022] Where rect(·) is the window function; T J is the interference forwarding width; T s is the interference sampling period; N max is the number of interference retransmissions in a single pulse, n represents the number of current slices; M max is the number of repetitions for a single scrambled forwarding, m is the current number of repetitions; T is the pulse width, if T / T S is an integer, N max =T / T S ,otherwise Indicates rounding down.

[0023] Optionally, the solution formula for the cyclic power spectrum and the cyclic cross power spectrum in step S2 is:

[0024]

[0025] in, is the cyclic power spectral density, is the circular autocorrelation function, R x (t, τ) is the autocorrelation function of the signal, is the cyclic cross power spectral density, is the circular cross-correlation function.

[0026] Optionally, the output signal of the cyclic Wiener filter constructed in step S5 is:

[0027]

[0028] in, α m (t) and β n (t) are the impulse response functions of the cyclic frequency obtained by calculating the cyclic power spectrum of the signal x(t), M and N are the number of cyclic frequencies of the linear part and the conjugate linear part, Represents convolution.

[0029] Optionally, the loop Wiener filter comprises constructing loop Wiener filter tap coefficients:

[0030]

[0031] Among them, A m (f) and B n (f) are the transfer functions of the linear and conjugate linear parts of the filter, A m (f) and B n (f) are respectively a m (t) and b n (t), and perform inverse Fourier transform on the above equation to obtain the tap coefficient a of the cyclic Wiener filter: m (t) and b n (t):

[0032]

[0033] Optionally, the iterative solution formulas of the tap coefficient linear part and the conjugate linear part are respectively:

[0034]

[0035]

[0036] Among them, μ is the step size parameter, which controls the update speed, and n is the number of iterations.

[0037] Optionally, the identification method in step S7 includes:

[0038] When the filter is constructed based on the target signal cycle frequency, the filtered output signal will extract the signal component corresponding to the target signal cycle frequency from the input signal. The target signal is continuous in the time domain, and the filtered output signal exhibits continuous characteristics in the time domain.

[0039] When the filter is constructed based on the cyclic frequency of the interference signal, the filtered output signal will extract the signal component corresponding to the cyclic frequency of the interference signal from the input signal. The forwarding interference signal adopts intermittent sampling forwarding in the time domain, and the filtered output signal presents discontinuous characteristics in the time domain.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. The filter is designed by using the cyclic frequency of the target and interference signals, and the continuity of the output signal after filtering in the time domain is used to distinguish the target and interference signals, and accurately distinguish and extract the target echo signal, which overcomes the dependence of the traditional Wiener filtering method on the statistical characteristics of the signal and reduces the problem of high computational complexity caused by the need for a large amount of sample data for statistical characteristics;

[0042] 2. The LCL-FRESH filtering criterion is used to calculate the tap coefficients of the cyclic Wiener filter based on the iterative solution method to reduce the computational complexity of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of an interference suppression method based on iterative design of a cyclic Wiener filter;

[0044] Figure 2 This is the schematic diagram of the forwarding interference principle;

[0045] Figure 3 Schematic diagram of the structure of the cyclic Wiener filter;

[0046] Figure 4 It is the cyclic frequency diagram of radar echo signal;

[0047] Figure 5 It is the cycle frequency diagram after the orderly statistical constant false alarm detection;

[0048] Figure 6 are the output signals after multiple groups of cyclic Wiener filtering;

[0049] Figure 7 This is the target echo signal cycle frequency diagram after interference suppression. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] See also Figure 1 , is an interference suppression method based on iterative design of a cyclic Wiener filter disclosed in an embodiment of the present invention, comprising:

[0052] S1, the radar echo signal is processed by down-conversion and dechirp to obtain a beat signal;

[0053] S2, transforming the beat signal into the cyclic frequency domain, and calculating the cyclic power spectrum and cyclic cross power spectrum of the signal;

[0054] S3, extracting the cyclic frequencies of the linear parts of the target signal and the interference signal using an ordered statistical constant false alarm detection algorithm in the cyclic power spectrum;

[0055] S4, extracting the cyclic frequency of the conjugate linear part of the target signal and the interference signal by using the ordered statistical constant false alarm detection algorithm in the cyclic cross power spectrum;

[0056] S5, constructing a cyclic Wiener filter through the cyclic frequencies in step S3 and step S4;

[0057] S6, performing multiple groups of cyclic Wiener filtering processes on the beat signal in step S1 in parallel;

[0058] S7. According to the continuity of the output signal in step S6 in the time domain, the target signal with continuous characteristics in the time domain is retained, and the forwarding interference signal with discontinuous characteristics in the time domain is discarded.

[0059] In this way, the radar echo is transformed into the cyclic frequency domain after signal processing, and a cyclic Wiener filter is constructed only by extracting the cyclic frequencies of the target and interference signals. This overcomes the dependence of the traditional Wiener filtering method on the statistical characteristics of the signal and reduces the problem of high computational complexity caused by the need for a large amount of sample data due to the statistical characteristics.

[0060] In some feasible embodiments, the identification method in step S7 includes:

[0061] When the filter is constructed based on the target signal cycle frequency, the filtered output signal will extract the signal component corresponding to the target signal cycle frequency from the input signal. The target signal is continuous in the time domain, and the filtered output signal exhibits continuous characteristics in the time domain.

[0062] When the filter is constructed based on the cyclic frequency of the interference signal, the output signal after filtering will extract the signal component corresponding to the cyclic frequency of the interference signal from the input signal. The forwarding interference signal adopts the intermittent sampling forwarding form in the time domain, and the output signal after filtering presents discontinuous characteristics in the time domain.

[0063] Based on the continuity difference of the output signal in the time domain, accurate identification of the target and interference signals can be achieved, thereby retaining the target signal, eliminating the forwarding interference signal, and realizing robust suppression of forwarding interference.

[0064] As a specific implementation of an interference suppression method based on iterative design of a cyclic Wiener filter provided in the application, in step S1, the beat signal obtained after down-conversion and de-linear frequency modulation of the radar echo signal is expressed as:

[0065] y(t)=S echo (t)+J(t)+n(t)

[0066] Among them, S echo (t) is the beat signal obtained by down-converting and de-linear frequency modulation of the target signal under linear frequency modulation continuous wave radar, J(t) is the beat signal obtained by down-converting and de-linear frequency modulation of the interference signal under intermittent sampling repeated forwarding interference, and n(t) is Gaussian white noise.

[0067] In some feasible ways, in a linear frequency modulation continuous wave radar, the beat signal obtained after the target signal is down-converted and de-linear frequency modulated can be expressed as:

[0068] S echo (t) = A·exp(2jπf s t+θ)

[0069] where f s is the beat signal frequency, A is the beat signal amplitude, and θ is the beat signal phase.

[0070] See also Figure 2 Under the intermittent sampling and repeated forwarding interference, the jammer first intercepts the radar target signal, intermittently samples the target signal and repeatedly forwards it to form an interference signal. The beat signal obtained after down-conversion and de-linear frequency modulation can be expressed as:

[0071]

[0072] Where rect(·) is the window function; T J is the interference forwarding width; T s is the interference sampling period; N max is the number of interference retransmissions in a single pulse, n represents the number of current slices; M maxis the number of repetitions for a single scrambled forwarding, m is the current number of repetitions; T is the pulse width, if T / T S is an integer, N max =T / T S ,otherwise Indicates rounding down.

[0073] As another specific implementation of an interference suppression method based on iterative design of a cyclic Wiener filter provided by the application, the solution formula for the cyclic power spectrum and the cyclic cross-power spectrum in step S2 is:

[0074]

[0075] in, is the cyclic power spectral density, is the circular autocorrelation function, R x (t, τ) is the autocorrelation function of the signal, is the cyclic cross power spectral density, is the circular cross-correlation function.

[0076] After solving the cyclic power spectral density and cyclic cross-power spectral density, the ordered statistical constant false alarm detection algorithm is used to extract the linear part cyclic frequencies of the target and interference signals in the cyclic power spectral density; the ordered statistical constant false alarm detection algorithm is used to extract the conjugate linear part cyclic frequencies of the target and interference signals in the cyclic cross-power spectral density.

[0077] In some feasible embodiments, the method of constructing the cyclic Wiener filter in step S5 can be found in Figure 3 The filter structure in the filter formula for periodic linear time-varying signals can be expressed as:

[0078]

[0079] Where y(t) is the filter output signal; x(t) or x(u) is the filter input signal. Since the impulse response function h(t, u) is periodic to the variable τ=tu, the impulse response function can be expressed as follows:

[0080] h(t,u)=∑ η h η (tu)exp(i2πημ) (8)

[0081] Substituting formula (8) into formula (7) and simplifying it, we can obtain:

[0082]

[0083] in represents convolution, x η(t) is the frequency shift signal obtained by multiplying x(t) or x(u) by exp(i2πημ).

[0084] From the above formula, it can be seen that filtering of linear periodic time-varying signals can be equivalent to superposition of linear time-invariant filtering results after the signal frequency shift. Radar echo is a complex signal, so the optimal time-varying multi-period filtering problem must be transformed into a dual variable filtering problem. The designed cyclic frequency filter is used to perform frequency shift processing on the beat signal and its conjugate signal respectively, and then the linear time-invariant filtering is performed on the frequency-shifted beat signal and its conjugate signal. Finally, the filtered signal and its conjugate signal are added to obtain the final filtered output signal. At this time, equation (7) can be rewritten as:

[0085]

[0086] in, α m (t) and β n (t) are impulse response functions of the cyclic frequency obtained by calculating the cyclic power spectrum of the signal x(t), M and N are the number of cyclic frequencies of the linear part and the conjugate linear part respectively.

[0087] As another specific implementation of an interference suppression method based on iterative design of a cyclic Wiener filter provided by the application, the cyclic Wiener filter includes constructing tap coefficients of the cyclic Wiener filter. The tap coefficients must meet the LCL-FRESH filtering criterion and can be expressed as:

[0088]

[0089] Among them, A m (f) and B n (f) are the transfer functions of the linear and conjugate linear parts of the filter, respectively. The upper part of the formula is the design criterion for the tap coefficients of the linear part of the cyclic Wiener filter, and the lower part of the formula is the design criterion for the tap coefficients of the conjugate linear part of the cyclic Wiener filter. A m (f) and B n (f) are respectively a m (t) and b n (t), and perform inverse Fourier transform on equation (11) to obtain the tap coefficient a of the cyclic Wiener filter: m (k) and b n (t):

[0090]

[0091] As another specific implementation of an interference suppression method based on iterative design of a cyclic Wiener filter provided by the application, based on the tap coefficients of the above-mentioned cyclic Wiener filter, an iterative solution formula for its linear part and conjugate linear part is disclosed, specifically:

[0092] To express the magnitude of the error between the output of the cyclic Wiener filter and the expected output, a filter cost function is constructed:

[0093]

[0094] in is the filter output signal, y(n) is the expected signal, and the optimal filtering effect is obtained by minimizing the cost function.

[0095] Based on the LCL-FRESH filtering criterion, the expression of the output signal is:

[0096]

[0097] Substituting formula (13) and formula (14) into the linear part of the filter tap coefficient in formula (12), the original cost function can be rewritten as:

[0098]

[0099] The gradient vector of the rewritten cost function can be expressed as:

[0100]

[0101] in, is an M-dimensional row vector, The specific form can be expressed as:

[0102]

[0103] Substituting formula (15) into formula (17), we can obtain:

[0104]

[0105] The above formula can be further simplified as:

[0106]

[0107] Similarly, by substituting formula (13) and formula (14) into the lower part of formula (12) and simplifying, the gradient vector expression of the conjugate linear part of the filter tap coefficient can be obtained as follows:

[0108]

[0109] Based on the derived gradient vector expressions of the linear part and the conjugate linear part of the filter tap coefficient, the steepest descent method is used for iterative update. The iterative formula of the steepest descent method can be expressed as:

[0110]

[0111] Among them, μ is the step size parameter, which controls the update speed, and n is the number of iterations.

[0112] Substituting the gradient vector of the linear part of the filter tap coefficient into formula (21), we can get its iterative update formula:

[0113]

[0114] Similarly, the iterative update formula of the conjugate linear part of the filter tap coefficient can be expressed as:

[0115]

[0116] After multiple rounds of iterations, the tap coefficients of the linear time-invariant filter in the linear part and the conjugate linear part of the cyclic Wiener filter can be obtained, and these filters are combined according to the cyclic frequency to realize the cyclic Wiener filter design.

[0117] To facilitate the verification of this application, the following verification examples are made public:

[0118] The radar transmission signal is selected as linear frequency modulation continuous wave, and the interference type is intermittent sampling repeated forwarding interference.

[0119] The beat signal obtained after down-conversion and de-linear frequency modulation of the target signal has a cycle frequency of 200 Hz, an amplitude of 1, a signal pulse width of 24 us, and a signal power of -25 dBm.

[0120] The interference parameters are set as interference forwarding period of 8us, forwarding width of 2us, number of repeated forwarding times of 3, and interference signal power of -20dBm.

[0121] See also Figure 4 , the echo signal is transformed into the cyclic frequency domain to calculate the cyclic power spectral density, and the cyclic frequency dimension of the signal is observed. It can be seen that there are 4 peak points in the cyclic frequency domain, one of which is the target peak point with a cyclic frequency of 200Hz, and the other 3 are interference peak points, which are 400Hz, 700HZ, and 1000Hz respectively.

[0122] See also Figure 5 , after using the OS-CFAR algorithm, the locations of the main peak points of the target and the interference can be obtained, and the cyclic frequencies of the four linear parts can be extracted from the cyclic power spectrum density diagram. Correspondingly, the cyclic frequencies of the four conjugate linear parts can be extracted from the cyclic cross-power spectrum density.

[0123] Furthermore, by using the cyclic frequencies of the linear part and the corresponding conjugate linear part, four groups of cyclic Wiener filters can be designed, and the tap coefficients of each group of filters can be solved by the iterative optimization method. After filtering the echo signal using the designed four cyclic Wiener filters, the results can be found in Figure 6 .

[0124] It can be seen that the target signal is continuous in the time domain after filtering, while the interference signal is discontinuous in the time domain after filtering. Figure 7 ,Based on this characteristic, we can identify the target and the interference, and extract the target signal as the final signal after filtering,,achieving robust suppression of forwarding interference.

[0125] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An interference suppression method based on iterative design of cyclic Wiener filter, characterized in that: The following steps are involved: S1, obtain a beat signal by down-converting and de-linear frequency modulation of the radar echo signal; S2, transforming the beat signal into the cyclic frequency domain, and calculating the cyclic power spectrum and cyclic cross power spectrum of the signal; S3, extracting the cyclic frequencies of the linear parts of the target signal and the interference signal using an ordered statistical constant false alarm detection algorithm in the cyclic power spectrum; S4, extracting the cyclic frequency of the conjugate linear part of the target signal and the interference signal by using the ordered statistical constant false alarm detection algorithm in the cyclic cross power spectrum; S5, constructing a cyclic Wiener filter through the cyclic frequencies in step S3 and step S4; S6, performing multiple groups of cyclic Wiener filtering processes on the beat signal in step S1 in parallel; S7. According to the continuity of the output signal in step S6 in the time domain, the target signal with continuous characteristics in the time domain is retained, and the forwarding interference signal with discontinuous characteristics in the time domain is discarded.

2. The interference suppression method based on iterative design of cyclic Wiener filter according to claim 1, characterized in that: In step S1, the beat signal obtained after the radar echo signal is down-converted and de-linear frequency modulated is expressed as: y(t)=S echo (t)+J(t)+n(t) Among them, S echo (t) is the beat signal obtained by down-converting and de-linear frequency modulation of the target signal under linear frequency modulation continuous wave radar, J(t) is the beat signal obtained by down-converting and de-linear frequency modulation of the interference signal under intermittent sampling repeated forwarding interference, and n(t) is Gaussian white noise.

3. The interference suppression method based on iterative design of cyclic Wiener filter according to claim 2, characterized in that: The S echo (t) can be expressed as: S echo (t)=A·exp(2jπf s t+θ) Among them, f s is the beat signal frequency; A is the beat signal amplitude; θ is the beat signal phase.

4. The interference suppression method based on iterative design of cyclic Wiener filter according to claim 2, characterized in that: The J(t) can be expressed as: Where rect(·) is the window function; T J is the interference forwarding width; T s is the interference sampling period; N max is the number of interference retransmissions in a single pulse, n represents the number of current slices; M max is the number of repetitions for a single scrambled forwarding, m is the current number of repetitions; T is the pulse width, if T / T S is an integer, N max =T / T S ,otherwise Indicates rounding down.

5. The interference suppression method based on iterative design of cyclic Wiener filter according to claim 1, characterized in that: The solution formula for the cyclic power spectrum and the cyclic cross power spectrum in step S2 is: in, is the cyclic power spectral density, is the circular autocorrelation function, R x (t, τ) is the autocorrelation function of the signal, is the cyclic cross power spectral density, is the circular cross-correlation function.

6. The interference suppression method based on iterative design of cyclic Wiener filter according to claim 1, characterized in that: The output signal of the cyclic Wiener filter constructed in step S5 is: in, α m (t) and β n (t) are the impulse response functions of the cyclic frequency obtained by calculating the cyclic power spectrum of the signal x(t), M and N are the number of cyclic frequencies of the linear part and the conjugate linear part, Represents convolution.

7. The interference suppression method based on iterative design of cyclic Wiener filter according to claim 6, characterized in that: The loop Wiener filter includes constructing the loop Wiener filter tap coefficients: Among them, A m (f) and B n (f) are the transfer functions of the linear and conjugate linear parts of the filter, A m (f) and B n (f) are respectively a m (t) and b n (t), and perform inverse Fourier transform on the above equation to obtain the tap coefficient a of the cyclic Wiener filter: m (t) and b n (t):

8. The interference suppression method based on iterative design of cyclic Wiener filter according to claim 7, characterized in that: The iterative solution formulas of the tap coefficient linear part and the conjugate linear part are respectively: Among them, μ is the step size parameter, which controls the update speed, and n is the number of iterations.

9. The interference suppression method based on iterative design of cyclic Wiener filter according to claim 1, characterized in that: The identification method in step S7 includes: When the filter is constructed based on the target signal cycle frequency, the filtered output signal will extract the signal component corresponding to the target signal cycle frequency from the input signal. The target signal is continuous in the time domain, and the filtered output signal exhibits continuous characteristics in the time domain. When the filter is constructed based on the cyclic frequency of the interference signal, the filtered output signal will extract the signal component corresponding to the cyclic frequency of the interference signal from the input signal. The forwarding interference signal adopts intermittent sampling forwarding in the time domain, and the filtered output signal presents discontinuous characteristics in the time domain.

Images