RU2490701C2 - Method for correlation processing of mixture of harmonic signal and noise - Google Patents

Abstract

FIELD: information technology.

SUBSTANCE: method comprises steps of: representing the mixture with an array of time-equidistant digital codes; deriving a primary correlation function of the array; cutting the end portion therefrom and deriving a secondary correlation function from the remaining portion; subsequent correlation functions are derived similarly, either until extracting a harmonic signal in the next correlation function, or until an unallowable reduction in duration thereof after another cutting; said array of digital codes is considered first, turned in reverse sequence and considered second; cross-correlation functions CCF11 of the first array with the second and CCF12 of the second array with the first are generated; end portions thereof are cut; the remaining portions are used to generate cross-correlation functions CCF21 between CCF11 and CCR12 and CCF22 between CCF12 and CCF11; the next CCF are similarly derived from secondary cross-correlation functions CCF21 and CCF22.

EFFECT: high signal-to-noise ratio in the derived cross-correlation functions and extracting a harmonic signal in the analysed mixture with low signal-to-noise ratio.

Description

The alleged invention relates to the technique of correlation processing of mixtures of signal and noise and can be used to isolate a harmonic signal from noise and measure its frequency.

A known method of correlation processing, implemented in a signal detector (Wolfovsky B.N., Golotvin K.G. Detector of signals // USSR Author's Certificate No. 885948. Published in B.I. No. 44. 1981) and consisting in the fact that a mixture of a harmonic signal with noise is presented both in an analog form and an array of equally spaced (in time) digital codes, which are multiplied by the analog values of the mixture; the resulting paired products of the mixture and its delayed copy are accumulated and averaged, thereby forming estimates of the ordinates of the autocorrelation function; the latter is analyzed for the presence of a harmonic signal.

Signs of an analogue that coincide with the features of the alleged invention are: presentation of the mixture by an array of equally spaced (in time) digital codes, obtaining the primary (KF1) correlation function and analyzing the latter for the presence of a harmonic signal.

The disadvantage of this method is that the possibility of isolating a harmonic signal (in the analyzed signal-to-noise mixture) by the primary autocorrelation function (ACF1) is limited not only by the signal-to-noise ratio in the said mixture, but also by its (mixture) duration and, as a result, duration AKF1.

There is also a method of correlation processing, implemented in a correlation detector (Volfovsky B.N., Vasiliev N.A. Correlation detector // Copyright certificate of the USSR No. 523419. Published in B.I. No. 28. 1976) and consisting in that a mixture of a harmonic signal with noise is represented by an array of equally spaced (in time) digital codes, a primary (ACF1) autocorrelation function of the array is obtained, the initial portion of this function is cut off (with a duration of at least the absolute interval of noise correlation) and obtained from the remainder of watts ternary (ACF2) autocorrelation function, similarly cut off its initial section and similarly obtain tertiary (AKF3), quaternary (AKF4) and subsequent autocorrelation functions, the formation of which continues either until the harmonic signal is highlighted in the next autocorrelation function, or until its duration , after cutting off the initial section, it is unacceptably reduced.

In this case, the features of the analogue that coincide with the features of the alleged invention are the presentation of the mixture by an array of equally spaced (in time) digital codes, obtaining the primary (KF1) correlation function of the array, cutting off from KF1 a portion of this function, and obtaining from the remaining part of the secondary (KF2) correlation functions, cutting off the KF2 site and obtaining the tertiary (KF3), quaternary (KF4) and subsequent correlation functions, the formation of which continues either until the harmonic As a result, in the next correlation function, or until its duration, after the next section cut-off, is unacceptably reduced.

The disadvantage of this analogue method is that in it, before the next cycle of correlation processing, only the initial portion of the correlation function is cut off (because most of the noise energy is concentrated in it) and the end portion of the same function is not cut off; the area where the variance of its values is large, due to the fact that during the formation of this area, a small number of paired products is used for averaging.

Closest to the claimed method in technical essence, the method described in the article (Volfovsky B.N. Multiple autocorrelation processing and its ability to detect a harmonic signal in a mixture of signal with noise, http://counter-terrorism.narod.ru/magazine1/wolf -1.htm), which consists in the fact that a mixture of a harmonic signal with noise is represented by an array of equally spaced (in time) digital codes, a primary (ACF1) autocorrelation function of the array is obtained, its initial section is cut off (with a duration of at least the absolute interval of noise correlation) and the final section and get the secondary (ACF2) autocorrelation function from the remaining part, similarly cut off its initial and final sections and similarly receive the tertiary (ACF3), quaternary (AKF4) and subsequent autocorrelation functions, the formation of which continues either until the harmonic signal is separated into the next autocorrelation function, or until its duration, after cutting off the initial section, is unacceptably reduced.

Signs that coincide with the features of the proposed invention: presentation of the mixture by an array of equally spaced (in time) digital codes, obtaining the primary (KF1) correlation function of the array, cutting off the final section from it and obtaining the correlation function from the remaining part of the secondary (KF2), cutting off the final section of this function and obtaining tertiary (CF3), quaternary (CF4) and subsequent correlation functions similar to that described, the formation of which is continued either until the harmonic signal is separated into the next correlation relational function, or until its duration, after the next cut-off, is unacceptably reduced.

The disadvantage of the prototype method is the forced cut-off of the initial section of the correlation function (with a duration of at least the absolute interval of noise correlation). Such a cutoff is advisable, since most of the noise energy is concentrated in the cut-off section and the noise correlations are strong.

The task to be solved by the alleged invention is aimed at preserving the initial portion of the correlation function, increasing, due to this, its duration, and, as a result: increasing the signal-to-noise ratio in the resulting correlation functions and isolating the harmonic signal at lower ratios than the prototype signal / noise in the analyzed mixture of harmonic signal and noise.

This technical result is achieved by the fact that the aforementioned (hereinafter referred to as the first) array is expanded backwards (i.e., the last code of the array is made first, the penultimate one is second, etc., and the first one is last) and considered the second, form a cross-correlation the function (VKF11) of the first array with the second and the cross-correlation function (VKF12) of the second array with the first, then, after cutting off the end sections from these (primary) VKF, use the remaining parts to form the cross-correlation function (VKF21) between VKF11 and VKF12 and cross-correlated onnoy function (VKF22) between VKF12 and VKF11; secondary VKF (VKF21 and VKF22), similar to that described, after cutting off their final sections, they are used to obtain tertiary (VKF31 and VKF32), quaternary (VKF41 and VKF42) and subsequent VKF.

To achieve a technical result in the method of correlation processing of a mixture of a harmonic signal with noise, which consists of the following: presentation of the mixture by an array of equally spaced (in time) digital codes, obtaining the primary (KF1) correlation function of the array, cutting off the final section from it, and obtaining from the remaining parts of the secondary (KF2) correlation function, cutting off the final section of this function and obtaining the tertiary (KF3), quaternary (KF4) and subsequent correlation functions similar to that described, form They continue either until the harmonic signal is highlighted in the next correlation function, or until its duration, after the next cut-off, is unacceptably reduced, the mentioned (hereinafter referred to as the first) array is expanded backwards (i.e., the last code of the array is made first, the penultimate - the second, etc., and the first - the last) and considered the second, form the cross-correlation function (VKF11) of the first array with the second and the cross-correlation function (VKF12) of the second array with the first, then, after cutting off from these ( primary s) CCF end portions, the remaining portion is used for forming a one-correlation function (VKF21) between VKF11 and VKF12 and cross-correlation function (VKF22) between VKF12 and VKF11; secondary VKF (VKF21 and VKF22), similar to that described, after cutting off their final sections, they are used to obtain tertiary (VKF31 and VKF32), quaternary (VKF41 and VKF42) and subsequent VKF.

Comparison of the proposed method with the prototype shows that it contains new features, i.e. meets the criterion of novelty. From a comparison with analogues, it follows that the inventive method meets the criterion of "significant differences", as in the analogues are not found new claimed features.

To prove the existence of a causal relationship between the claimed features and the achieved technical result, we consider the essence of the proposed method for the correlation processing of a mixture of a harmonic signal with noise and compare it with the prototype method and analog methods.

As is known, (see Gonorovsky IS Radio engineering circuits and signals. Part I, Sovetskoe Radio Publishing House M., 1967) the autocorrelation functions (ACF) of a harmonic signal and noise are significantly different. The harmonic signal ACF is the same (of the same frequency) harmonic signal throughout the delay axis. But the shape of the ACF noise is substantially uneven. So most of the energy of ACF noise is concentrated in the region of small delays; at zero delay, the ACF value of the noise is equal to the variance of the noise, and with an increase in the delay, the correlation relationships of the noise weaken.

In this connection, the ACF of a mixture of a harmonic signal and noise (hereinafter referred to as the primary ACF) can be considered as a new implementation of a mixture of a harmonic signal and noise, in which the role of a harmonic signal is played by an ACF signal and the role of noise: ACF noise, VKF signal with noise, and VKF noise with a signal. The difference between the primary ACF and the above-mentioned mixture of signal and noise is that in the ACF the signal-to-noise ratio has changed compared to the mixture. So, in the region of small delays, the signal-to-noise ratio decreased due to the concentration of most of the noise energy in this region, and in the region of medium and large delays, this ratio increased.

Since the primary ACF retains information about the signal, as well as the original mixture of the signal with noise, it is possible to apply autocorrelation processing (ACO) to obtain secondary (and subsequent) ACFs containing (by analogy; like primary ACF) information about both signal and noise. Obviously (this follows from the previous paragraph), it is advisable to cut off the initial portion from the primary ACF from the primary ACF, since the signal-to-noise ratio has decreased on it. Similarly, it is advisable to provide for clipping of the final portion of the ACF; the area where the variance of the ACF values is large, due to the fact that during its formation a small number of paired products is used for averaging.

Note that the small delays of the ACF correspond to strong noise correlations. It is precisely because of this that a large part of the noise energy is concentrated in the region of small delays. If the correlation of noise is significantly weakened, then the noise energy at low delays will decrease significantly and the signal-to-noise ratio in the initial section of the ACF will increase. As a result, this section will become expedient for AKO.

To weaken the correlation relationships of noise at low delays and achieve this effect, the proposed invention proposes to form not ACF, but VKF; first between the initial mixture of the signal with noise and the same mixture turned backwards, and then between the resulting primary, secondary, and subsequent cross-correlation functions. Such an approach, in principle, maximally weakens the correlation relationships of noise and makes it unnecessary to cut off the initial ACF section before the next AKO. Saving the initial section will allow you to: increase the duration of the correlation function, and this, in turn, will increase the signal-to-noise ratio in the correlation functions and will allow you to select a harmonic signal at lower signal-to-noise ratios than the prototype.

The implementation of the proposed method will allow you to select a harmonic signal and measure its frequency at lower signal-to-noise ratios than in the prototype method in the analyzed signal-to-noise mixture.

Claims (1)

A method for correlation processing a mixture of a harmonic signal with noise, using: presenting the mixture with an array of digital codes equally spaced in time, obtaining the primary correlation function of the array, cutting off the final section from it and obtaining the secondary correlation function from the remaining part, cutting off the final section of this function and obtaining the tertiary similar to that described , quaternary and subsequent correlation functions, the formation of which is continued either until the harmonic signal is separated into the next Relational function, or as long as its duration after the next clipping unacceptably reduced, characterized in that said array is considered the first digital codes and deploying it in the reverse sequence, i.e., the last array code is made first, the penultimate one is the second, and the first is the last; the array formed in this way is considered the second, the cross-correlation function of VKF11 of the first array with the second is formed, and the cross-correlation function of VKF12 of the second array with the first is formed, then after cutting off the end sections from the primary VKF11 and VKF12, use their remaining parts to form the cross-correlation function of VKF21 between VKF11 and VKF12 and the cross-correlation function VKF22 between VKF12 and VKF11; secondary cross-correlation functions VKF21 and VKF22, similar to that described after cutting off their end sections, are used to obtain tertiary VKF31 and VKF32, Quaternary VKF41 and VKF42 and subsequent cross-correlation functions.