RU2685972C1 - Method and device for filtering frequency-modulated signals - Google Patents

Abstract

FIELD: data processing.

SUBSTANCE: invention relates to digital processing of signals and other industries, in which digital matched filtration (compression) of signals with intrapulse modulation can be used. Result is ensured by using a corrected copy which takes into account features of the frequency-modulated (FM) signal spectrum (Fresnel pulsations), as well as distortion originating in the receiving path.

EFFECT: high efficiency of processing real FM signals, which, as compared to conventional processing, provides the required level of side interference of compressed signals in a given range of Doppler frequencies without increasing the peak duration compared to traditional weight processing.

2 cl, 4 dwg

Description

The technical field to which the invention relates.

The invention relates to the field of digital matched filtering (compression) of signals with intrapulse modulation and can be applied in all branches of technology that use frequency-modulated (FM) signals.

The level of technology

A technique of digital filtering of frequency-modulated signals is widely known in the art. It compresses an input signal in the frequency domain by multiplying its spectrum by the spectrum of a complex conjugate signal (copy) and then transforming the product into a time domain.

The known procedure of discrete convolution of the FM signal in the frequency domain is described by the expression (1):

where g (m) is a discrete compressed signal, n is the index of the spectrum of signals in the frequency domain, m-index of the signal in the time domain (m = 0,1,2 ... M-1), N is the number of samples of the spectra of the signal Ss (n) and copies Sk (n), M is the number of samples of the signal g (m) in the time domain, Ss (n) is the discrete echo spectrum described by formula (2),

where s (m) is a discrete FM input signal, Sk (n) is a discrete copy spectrum, calculated by the formula (3),

where k (m) is a discrete copy of the FM signal.

The fast Fourier transform (FFT) and the inverse fast Fourier transform (IFFT) are usually used to transform signals into the spectral and temporal domain, and in this case M is equal to N.

With matched filtering, the s (m) and k (m) signals are complex conjugate. However, in this case, the compressed signal g (m) has significant side lobes (BL). So for a linear frequency-modulated signal (chirp), the first side lobe is 13 dB below the peak value of the compressed signal, and the level of the next BL decreases from lobe to lobe by about 4 dB, which is higher than the required values. To reduce the level of side lobes (LB) of the chirp signals, copy processing is applied in the time k (m) or in the frequency Sk (n) domain. The use of the Hamming weight function for signals with a high compression ratio (Ксж> 100) makes it possible to reduce UBL to minus 42 dB, with Kszh≤50, weight processing becomes less effective and UBL≈- (26 ÷ 30) dB.

By virtue of the formula (1) used, the digital filtering system for FM signals has the following disadvantages:

- with relatively small compression ratios when processing FM signals, UBL becomes higher than the expected values;

- when sampling, quantizing and filtering, the input signal undergoes distortions, which can also reduce the efficiency of weight processing.

DISCLOSURE OF INVENTION

The technical problem addressed by the claimed technical solution is the calculation of a copy of the signal with intrapulse modulation, which makes it possible to more effectively reduce UBL during signal compression, and also take into account the influence of the receiving path (ADC, frequency converters, filters) on the input signal and its copy.

The technical result of the claimed invention is a more efficient processing of real FM signals, providing compared with traditional processing required UBL compressed signals in a given range of Doppler frequencies.

The claimed technical result is provided by a method of filtering frequency-modulated signals, which consists in the following: after sampling in the ADC, filtering and converting the discrete spectrum of the input FM signal to the frequency domain goes to a multiplier, the second input of which receives samples of the spectrum of the corrected FM signal which is carried out as follows: calculate the reference FM signal in the time domain taking into account the conversion to the ADC and filtering, by analogy with the conversion the input signal and the sample copy matched with the calculated reference FM signal (for the chirp signal, the copy envelope is weighed in accordance with the chosen weight function, for example, Hamming), then the convolution of the converted reference FM signal and its copy are calculated, obtaining a compressed FM signal which is then corrected by multiplying it in the side-lobe region by a correction function that reduces the BL to the required level, after which the calculated compressed and reference signals are converted into the frequency domain Part, calculate the quotient of the calculated spectra, which corresponds to the new corrected spectrum of the copy of the FM signal, then the spectra of the input signal and the corrected copy are multiplied to obtain the spectrum of the compressed signal, which is then transformed into the time domain using the inverse fast Fourier transform (IFFT).

This technical result is provided by a filtering device for frequency-modulated signals containing a serially connected block of discrete processing of the input signal, containing an ADC block, filtering and conversion, and an FFT unit, the signal from which is fed to the input of the two spectra multiplication unit, to its second input the unit for calculating the corrected copy, the output of the multiplier of the spectra of the input signal and the corrected copy are connected to the OBPF block; the unit for calculating the corrected copy contains a series-connected unit for generating a reference FM signal, a discrete processing unit and a convolution unit, the first input of which receives a filtered reference signal, and the second input is a signal from a copy-forming unit integrated with the reference chirp signal, and the copy weighting unit, the output from the convolution unit’s output is fed to the compressed signal correction unit, the second input of which receives a signal from the signal conditioning unit, UBL th despread signal, outputted from the block correction signal despread signal is fed to calculating unit corrected copy through FFT unit, wherein the second input unit for calculating the corrected copy arrives last FFT filtered reference signal of said discrete processing unit.

The job of the claimed FM signal compression device is to calculate a copy of the FM signal, corrected in comparison with a typical copy.

A compressed signal is initially calculated to calculate the corrected copy. It can be calculated in accordance with the algorithm shown in FIG. 1, or using convolution in the time domain by the formula (4).

where se (n) is the calculated reference FM signal.

As a reference signal, the originally calculated FM signal is used, taking into account the sampling, quantization in the ADC and filtering. The copy is complex conjugate with the reference signal, and for chirp signals using weighting processing. Traditional weight processing is advisable to carry out by modulating the envelope of the copy of the chirp signal, because in this case, the Fresnel ripple spectrum of the copy will be less than when weighting the spectrum of the copy.

Then the compressed signal in the time domain is multiplied by the correction function, reducing the LLL to the desired values. As a correction function cor (k), any function can be used to reduce side lobes in the required time interval: near the peak, in the region of possible “pair echoes” of the compressed signal or, for example, uniform outside the peak of the compressed signal (formula (5)):

where koef - coefficient taking into account how many times it is necessary to reduce the compressed UBL

signal, M1 - the initial index of the peak of the compressed signal, M2 - the final index of the peak of the compressed signal, M - the number of copies of the copy.

The compressed signal is corrected in accordance with formula (6), then its spectrum is calculated using formula (7), and the spectrum Sse (n) of the calculated reference signal is calculated using formula (2), taking into account the transformations in the FM signal path.

The spectrum of the corrected copy Skcor (n) is calculated by the formula (8) as the ratio of the spectrum of the corrected compressed signal Sgcor (n) to the spectrum of the calculated reference signal Sse (n):

The calculated corrected copy is used in formula (1) in the processing (compression) of real input signals with intrapulse modulation.

Brief Description of the Drawings

FIG. 1 shows a flow chart of a digital device for filtering FM signals with convolution in the frequency domain.

FIG. 2 is a flow chart of the claimed digital device for filtering FM signals with the calculation of the corrected copy.

FIG. 3 is a block diagram of a digital device for filtering FM signals, in which: 1 is a copy FFT unit; 2 - copy weight processing unit; 3 - a unit for generating a copy, complexly associated with a reference chirp signal; 4 - block FFT input signal; 5 is a block of digital processing of the input signal in the receiving path; 6 - ADC block; 7 is a block multiplying the spectra of the copy and the input signal; 8 - OBPF block of the compressed signal spectrum; 9 is a block calculation module of the compressed signal.

FIG. 4 is a block diagram of the proposed digital device for filtering FM signals, in which: 101 is a signal conditioning unit that corrects the compressed signal UBL; 102 — weight processing unit for the copy of the chirp signal; 103 - unit for forming a typical copy, complexly conjugated with a reference signal; 104 is a digital processing unit that simulates the conversion of the input signal in the receiving path; 105 —the unit for generating a reference FM signal; 106 - block digital input processing in the receiving path; 107 - ADC block; 108 - block convolution of a copy and a reference signal; 109 - block correction of the compressed signal; 110 - FFT block of the compressed signal; 111-unit calculating the spectrum of the corrected copy; 112 - block FFT reference signal; 113 - input FFT block; 114 —an multiplication unit of the spectra of the corrected copy and the input signal; 115 — OBPF block of the corrected compressed signal spectrum; 116 - unit calculation module of the compressed signal.

The block diagram shown in FIG. 3, is used to process FM signals and, in particular, chirp signals. In block 3, a copy of the FM signal is calculated. When using a chirp signal, the copy envelope is weighted in accordance with the traditional weighting function, for example, the Hamming function (block 2), and then the spectrum of the weighted copy is calculated (block 1). In block 6, the input echo signal is sampled and quantized into the ADC, in block 5 is converted and filtered, and in block 4, the spectrum of the input signal is calculated using an FFT. The multiplication of the spectra of the copy from the output of block 1 and the input signal from the output of block 4 is carried out in block 7. In block 8, the spectrum of the signal from block 7 is converted into the time domain using the OBPF procedure. In block 9, a compressed signal module is evaluated for further signal processing.

The block diagram shown in FIG. 4, is used to process FM signals and, in particular, chirp signals. To do this, in block 105, a reference discrete FM signal is generated, which in block 104 is converted and filtered in accordance with the conversion and filtering of echo signals in the receiving path. In block 103, a typical copy of the FM signal is calculated. When using a chirp signal, its envelope of the copy in block 102 is weighted in accordance with the traditional weighting function, for example, the Hamming function. In block 108, the convolution of the weighted sample copy and the reference chirp signal is calculated, i.e. a compressed signal is generated, which in block 109 is multiplied by a correction signal coming from block 101 and reducing the compressed signal UBL in a given time interval to the desired level. Next, in blocks 110 and 112, the signal spectra are calculated respectively from the output of blocks 104 and 109. In block 111, the spectrum of the new corrected copy is calculated equal to the quotient from dividing the spectrum of the compressed signal from the output of block 110 by the spectrum of the reference signal from the output of block 112. In block 114, the spectrum is corrected the copies from block 111 are multiplied by the spectrum of the input signal converted in blocks 107 (ADC block) and 106 (block conversion and filtering). In block 115, the spectrum of the signal from the output of block 114 using the OBPF procedure is converted into the time domain. In block 116, a compressed signal module is evaluated for further signal processing.

Thus, the use of a corrected copy, which takes into account the characteristics of the spectrum of the FM signal (Fresnel pulsations), as well as distortions in the receiving path, makes it possible, without changing the peak duration, to reduce the compressed signal UBL. For chirp signals with small compression ratios, Ксж = (26 ÷ 50) with a minimum sampling frequency (f _DISCR ) 1.2 times the frequency deviation (W), UBL can be minus (42 ÷ 49) dB. Additional processing losses will be no more than 0.9 dB. With compression ratios of KSH> 50, the average UBL will be even smaller. By doubling the sampling frequency of the signals (f _DISCR = 2W), in a small Doppler frequency range, you can reduce the EEL to minus (60-80) dB.

Claims (8)

1. The method of filtering frequency-modulated signals, namely, that sampling the input frequency-modulated (FM) signal, transform, filter, calculate its spectrum using the fast Fourier transform (FFT) and feed the multiplier, to the second input of which the spectrum is obtained, obtained as follows: - form a discrete reference FM signal, which is converted and filtered in accordance with the above conversion and filtering; - calculate a copy of the FM signal, and for linear frequency-modulated (chirp) signals are weighed using modulation of the envelope, and perform convolution (compression) of the weighted copy and the reference FM signal, then the compressed signal is multiplied by the correction signal, - calculate the spectra of the multiplied with the correction signal of the compressed FM signal and the filtered reference discrete FM signal, - form a new corrected spectrum copy of the FM signal, equal to the quotient from dividing the calculated spectra; - form the spectrum of the compressed signal, multiplying the spectra of the input FM signal and the corrected copy, and convert it into the time domain using the inverse basic Fourier transform (IFFT). 2. A device for filtering FM signals containing the ADC block connected in series, a block for converting and filtering the input signal in the receiving path and an FFT block for the input signal, the signal from which goes to the multiplier unit of the spectra of the corrected copy and the input signal, the second input of which receives a signal from the output block calculating the spectrum of the corrected copy, and the output of the block multiplying the spectra of the corrected copy and the input signal is connected to the IFFT block, characterized in that the input of the spectrum calculator the corrected copy as a divider receives the spectrum of the reference FM signal generated sequentially in the shaping unit, the filtering unit and the FFT unit, and the output from the serially connected convolution unit of the reference signal and sample copy, the compressed signal correction unit and the unit FFT, while the reference signal and the sample copy are supplied to the inputs of the convolution unit and the sample copy, weighted in the case of using the chirp signal in the weight processing unit, the input to secondly connected to a copy forming unit integrated with a reference signal, the compressed signal from the output of the convolution unit of the reference signal and the copy goes to the compressed signal correction unit, the second input of which receives a signal from the side-lobe leveling unit (UBL) of the compressed signal , from the output of the compressed signal correction block, the corrected compressed signal is converted into the frequency domain through the FFT block of the compressed signal, the spectrum of the corrected copy is calculated as the quotient the spectrum of the corrected compressed signal on the spectrum of the generated reference FM signal.

Images